diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,56788 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 7.9922192151556155,
+  "global_step": 94500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 11.352,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 10.3758,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8e-05,
+      "loss": 9.1034,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 8.175,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 6.9829,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.6e-05,
+      "loss": 5.3579,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00011200000000000001,
+      "loss": 3.4022,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00012800000000000002,
+      "loss": 1.8856,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000144,
+      "loss": 1.8634,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00016,
+      "loss": 1.271,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017600000000000002,
+      "loss": 1.2577,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000192,
+      "loss": 1.1367,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020800000000000001,
+      "loss": 1.1801,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022400000000000002,
+      "loss": 0.9615,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00024,
+      "loss": 0.9087,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00025600000000000004,
+      "loss": 0.9564,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00027200000000000005,
+      "loss": 1.011,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000288,
+      "loss": 1.0056,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000304,
+      "loss": 0.9391,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00032,
+      "loss": 0.937,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000336,
+      "loss": 0.8199,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00035200000000000005,
+      "loss": 0.8833,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00036800000000000005,
+      "loss": 0.8461,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000384,
+      "loss": 0.9581,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004,
+      "loss": 0.8207,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00041600000000000003,
+      "loss": 0.8886,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00043200000000000004,
+      "loss": 0.8363,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00044800000000000005,
+      "loss": 0.846,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000464,
+      "loss": 0.8868,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00048,
+      "loss": 0.8315,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000496,
+      "loss": 0.7763,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005120000000000001,
+      "loss": 0.8158,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000528,
+      "loss": 0.9013,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005440000000000001,
+      "loss": 0.7732,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00056,
+      "loss": 0.8564,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000576,
+      "loss": 0.7771,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000592,
+      "loss": 0.8326,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000608,
+      "loss": 0.7996,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0006240000000000001,
+      "loss": 0.9489,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00064,
+      "loss": 0.7488,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000656,
+      "loss": 0.8316,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000672,
+      "loss": 0.8597,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000688,
+      "loss": 0.6988,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007040000000000001,
+      "loss": 0.8528,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00072,
+      "loss": 0.9214,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007360000000000001,
+      "loss": 0.8793,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000752,
+      "loss": 0.8323,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000768,
+      "loss": 0.9091,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000784,
+      "loss": 0.7685,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0008,
+      "loss": 0.7964,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007999149768311866,
+      "loss": 0.7333,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000799829953662373,
+      "loss": 0.7987,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007997449304935595,
+      "loss": 0.8004,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000799659907324746,
+      "loss": 0.7653,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007995748841559325,
+      "loss": 0.8146,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007994898609871191,
+      "loss": 0.8128,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007994048378183055,
+      "loss": 0.8755,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000799319814649492,
+      "loss": 0.7958,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007992347914806785,
+      "loss": 0.8564,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000799149768311865,
+      "loss": 0.8716,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007990647451430516,
+      "loss": 0.8574,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000798979721974238,
+      "loss": 0.7663,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007988946988054245,
+      "loss": 0.8213,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000798809675636611,
+      "loss": 0.7871,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007987246524677975,
+      "loss": 0.7427,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000798639629298984,
+      "loss": 0.7615,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007985546061301705,
+      "loss": 0.7643,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007984695829613569,
+      "loss": 0.8137,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007983845597925436,
+      "loss": 0.7785,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007982995366237301,
+      "loss": 0.7567,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007982145134549165,
+      "loss": 0.8446,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000798129490286103,
+      "loss": 0.8621,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007980444671172894,
+      "loss": 0.912,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000797959443948476,
+      "loss": 0.7489,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007978744207796625,
+      "loss": 0.7163,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000797789397610849,
+      "loss": 0.7954,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007977043744420355,
+      "loss": 0.8123,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000797619351273222,
+      "loss": 0.7708,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007975343281044085,
+      "loss": 0.661,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000797449304935595,
+      "loss": 0.7536,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007973642817667815,
+      "loss": 0.7689,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007972792585979679,
+      "loss": 0.8043,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007971942354291546,
+      "loss": 0.8051,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000797109212260341,
+      "loss": 0.7241,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007970241890915275,
+      "loss": 0.8862,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000796939165922714,
+      "loss": 0.7362,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0007968541427539004,
+      "loss": 0.8252,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000796769119585087,
+      "loss": 0.7583,
+      "step": 880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007966840964162735,
+      "loss": 0.8086,
+      "step": 890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00079659907324746,
+      "loss": 0.7874,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007965140500786465,
+      "loss": 0.7866,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000796429026909833,
+      "loss": 0.8041,
+      "step": 920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007963440037410195,
+      "loss": 0.7342,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000796258980572206,
+      "loss": 0.8099,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007961739574033924,
+      "loss": 0.7572,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007960889342345789,
+      "loss": 0.8761,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007960039110657655,
+      "loss": 0.7575,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000795918887896952,
+      "loss": 0.7929,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007958338647281385,
+      "loss": 0.7473,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0007957488415593249,
+      "loss": 0.7014,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007956638183905114,
+      "loss": 0.7967,
+      "step": 1010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000795578795221698,
+      "loss": 0.7737,
+      "step": 1020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007954937720528844,
+      "loss": 0.7436,
+      "step": 1030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007954087488840709,
+      "loss": 0.7839,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007953237257152574,
+      "loss": 0.7652,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000795238702546444,
+      "loss": 0.7857,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007951536793776305,
+      "loss": 0.6987,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007950686562088169,
+      "loss": 0.8973,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007949836330400034,
+      "loss": 0.7156,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007948986098711899,
+      "loss": 0.7097,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0007948135867023764,
+      "loss": 0.7933,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000794728563533563,
+      "loss": 0.7223,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007946435403647494,
+      "loss": 0.717,
+      "step": 1130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007945585171959359,
+      "loss": 0.7175,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007944734940271225,
+      "loss": 0.833,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007943884708583089,
+      "loss": 0.6525,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007943034476894954,
+      "loss": 0.6711,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007942184245206819,
+      "loss": 0.7691,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007941334013518683,
+      "loss": 0.7558,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000794048378183055,
+      "loss": 0.7444,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007939633550142415,
+      "loss": 0.7778,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007938783318454279,
+      "loss": 0.7128,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007937933086766144,
+      "loss": 0.7672,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0007937082855078008,
+      "loss": 0.815,
+      "step": 1240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007936232623389874,
+      "loss": 0.7632,
+      "step": 1250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007935382391701739,
+      "loss": 0.7548,
+      "step": 1260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007934532160013604,
+      "loss": 0.7188,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007933681928325469,
+      "loss": 0.6822,
+      "step": 1280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007932831696637335,
+      "loss": 0.7717,
+      "step": 1290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007931981464949199,
+      "loss": 0.7345,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007931131233261064,
+      "loss": 0.8165,
+      "step": 1310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007930281001572929,
+      "loss": 0.7428,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007929430769884793,
+      "loss": 0.7375,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000792858053819666,
+      "loss": 0.8057,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0007927730306508524,
+      "loss": 0.8524,
+      "step": 1350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007926880074820389,
+      "loss": 0.7634,
+      "step": 1360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007926029843132254,
+      "loss": 0.7759,
+      "step": 1370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007925179611444119,
+      "loss": 0.7515,
+      "step": 1380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007924329379755984,
+      "loss": 0.7386,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007923479148067849,
+      "loss": 0.7769,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007922628916379714,
+      "loss": 0.7603,
+      "step": 1410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007921778684691579,
+      "loss": 0.7233,
+      "step": 1420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007920928453003444,
+      "loss": 0.7061,
+      "step": 1430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007920078221315309,
+      "loss": 0.7832,
+      "step": 1440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007919227989627174,
+      "loss": 0.694,
+      "step": 1450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007918377757939038,
+      "loss": 0.667,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0007917527526250903,
+      "loss": 0.8031,
+      "step": 1470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007916677294562769,
+      "loss": 0.7311,
+      "step": 1480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007915827062874634,
+      "loss": 0.8335,
+      "step": 1490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007914976831186499,
+      "loss": 0.709,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007914126599498363,
+      "loss": 0.7925,
+      "step": 1510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007913276367810229,
+      "loss": 0.799,
+      "step": 1520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007912426136122094,
+      "loss": 0.6857,
+      "step": 1530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007911575904433958,
+      "loss": 0.7542,
+      "step": 1540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007910725672745823,
+      "loss": 0.6397,
+      "step": 1550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007909875441057688,
+      "loss": 0.721,
+      "step": 1560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007909025209369554,
+      "loss": 0.702,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007908174977681419,
+      "loss": 0.8134,
+      "step": 1580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0007907324745993283,
+      "loss": 0.8437,
+      "step": 1590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007906474514305148,
+      "loss": 0.8504,
+      "step": 1600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007905624282617014,
+      "loss": 0.7399,
+      "step": 1610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007904774050928878,
+      "loss": 0.7501,
+      "step": 1620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007903923819240744,
+      "loss": 0.8159,
+      "step": 1630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007903073587552609,
+      "loss": 0.8245,
+      "step": 1640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007902223355864473,
+      "loss": 0.8313,
+      "step": 1650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007901373124176339,
+      "loss": 0.8256,
+      "step": 1660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007900522892488203,
+      "loss": 0.7847,
+      "step": 1670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007899672660800068,
+      "loss": 0.854,
+      "step": 1680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007898822429111933,
+      "loss": 0.7702,
+      "step": 1690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007897972197423799,
+      "loss": 0.803,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0007897121965735664,
+      "loss": 0.7813,
+      "step": 1710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007896271734047529,
+      "loss": 0.6711,
+      "step": 1720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007895421502359393,
+      "loss": 0.7832,
+      "step": 1730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007894571270671258,
+      "loss": 0.7324,
+      "step": 1740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007893721038983124,
+      "loss": 0.7272,
+      "step": 1750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007892870807294988,
+      "loss": 0.7219,
+      "step": 1760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007892020575606853,
+      "loss": 0.7141,
+      "step": 1770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007891170343918718,
+      "loss": 0.6251,
+      "step": 1780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007890320112230583,
+      "loss": 0.813,
+      "step": 1790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007889469880542449,
+      "loss": 0.8317,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007888619648854313,
+      "loss": 0.8059,
+      "step": 1810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007887769417166178,
+      "loss": 0.7287,
+      "step": 1820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0007886919185478043,
+      "loss": 0.7898,
+      "step": 1830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007886068953789907,
+      "loss": 0.6575,
+      "step": 1840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007885218722101774,
+      "loss": 0.8035,
+      "step": 1850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007884368490413638,
+      "loss": 0.6962,
+      "step": 1860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007883518258725503,
+      "loss": 0.699,
+      "step": 1870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007882668027037368,
+      "loss": 0.779,
+      "step": 1880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007881817795349233,
+      "loss": 0.7021,
+      "step": 1890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007880967563661098,
+      "loss": 0.7576,
+      "step": 1900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007880117331972963,
+      "loss": 0.7192,
+      "step": 1910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007879267100284828,
+      "loss": 0.8109,
+      "step": 1920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007878416868596694,
+      "loss": 0.7804,
+      "step": 1930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007877566636908558,
+      "loss": 0.7852,
+      "step": 1940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007876716405220423,
+      "loss": 0.7984,
+      "step": 1950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007875866173532288,
+      "loss": 0.7354,
+      "step": 1960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007875015941844152,
+      "loss": 0.6181,
+      "step": 1970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007874165710156018,
+      "loss": 0.7995,
+      "step": 1980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007873315478467883,
+      "loss": 0.6643,
+      "step": 1990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007872465246779748,
+      "loss": 0.7594,
+      "step": 2000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007871615015091613,
+      "loss": 0.6107,
+      "step": 2010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007870764783403477,
+      "loss": 0.7455,
+      "step": 2020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007869914551715343,
+      "loss": 0.7935,
+      "step": 2030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007869064320027208,
+      "loss": 0.7216,
+      "step": 2040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007868214088339072,
+      "loss": 0.7392,
+      "step": 2050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007867363856650937,
+      "loss": 0.7437,
+      "step": 2060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007866513624962804,
+      "loss": 0.7498,
+      "step": 2070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007865663393274668,
+      "loss": 0.7152,
+      "step": 2080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007864813161586533,
+      "loss": 0.6961,
+      "step": 2090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007863962929898397,
+      "loss": 0.6733,
+      "step": 2100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007863112698210262,
+      "loss": 0.705,
+      "step": 2110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007862262466522128,
+      "loss": 0.7757,
+      "step": 2120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007861412234833992,
+      "loss": 0.7846,
+      "step": 2130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007860562003145858,
+      "loss": 0.7939,
+      "step": 2140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007859711771457723,
+      "loss": 0.7659,
+      "step": 2150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007858861539769588,
+      "loss": 0.8492,
+      "step": 2160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007858011308081453,
+      "loss": 0.7348,
+      "step": 2170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007857161076393318,
+      "loss": 0.7493,
+      "step": 2180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007856310844705182,
+      "loss": 0.7487,
+      "step": 2190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007855460613017047,
+      "loss": 0.7144,
+      "step": 2200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007854610381328913,
+      "loss": 0.7617,
+      "step": 2210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007853760149640778,
+      "loss": 0.6897,
+      "step": 2220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007852909917952643,
+      "loss": 0.7544,
+      "step": 2230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007852059686264507,
+      "loss": 0.7277,
+      "step": 2240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007851209454576372,
+      "loss": 0.7164,
+      "step": 2250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007850359222888238,
+      "loss": 0.6949,
+      "step": 2260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007849508991200102,
+      "loss": 0.757,
+      "step": 2270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007848658759511968,
+      "loss": 0.7453,
+      "step": 2280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007847808527823832,
+      "loss": 0.8222,
+      "step": 2290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007846958296135698,
+      "loss": 0.8956,
+      "step": 2300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007846108064447563,
+      "loss": 0.7678,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007845257832759427,
+      "loss": 0.6938,
+      "step": 2320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007844407601071292,
+      "loss": 0.7392,
+      "step": 2330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007843557369383157,
+      "loss": 0.7481,
+      "step": 2340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007842707137695022,
+      "loss": 0.7659,
+      "step": 2350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007841856906006888,
+      "loss": 0.713,
+      "step": 2360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007841006674318752,
+      "loss": 0.7069,
+      "step": 2370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007840156442630617,
+      "loss": 0.7748,
+      "step": 2380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007839306210942483,
+      "loss": 0.6603,
+      "step": 2390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007838455979254347,
+      "loss": 0.7605,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007837605747566212,
+      "loss": 0.7021,
+      "step": 2410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007836755515878077,
+      "loss": 0.757,
+      "step": 2420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007835905284189942,
+      "loss": 0.7613,
+      "step": 2430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007835055052501808,
+      "loss": 0.7767,
+      "step": 2440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007834204820813672,
+      "loss": 0.8062,
+      "step": 2450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007833354589125537,
+      "loss": 0.7052,
+      "step": 2460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007832504357437402,
+      "loss": 0.7192,
+      "step": 2470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007831654125749266,
+      "loss": 0.7301,
+      "step": 2480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007830803894061132,
+      "loss": 0.7131,
+      "step": 2490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007829953662372997,
+      "loss": 0.6935,
+      "step": 2500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007829103430684862,
+      "loss": 0.7095,
+      "step": 2510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007828253198996727,
+      "loss": 0.6776,
+      "step": 2520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007827402967308592,
+      "loss": 0.7232,
+      "step": 2530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007826552735620457,
+      "loss": 0.7073,
+      "step": 2540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007825702503932322,
+      "loss": 0.7722,
+      "step": 2550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007824852272244186,
+      "loss": 0.6119,
+      "step": 2560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007824002040556051,
+      "loss": 0.7608,
+      "step": 2570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007823151808867918,
+      "loss": 0.6227,
+      "step": 2580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007822301577179782,
+      "loss": 0.7126,
+      "step": 2590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007821451345491647,
+      "loss": 0.6767,
+      "step": 2600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007820601113803511,
+      "loss": 0.8066,
+      "step": 2610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007819750882115376,
+      "loss": 0.7678,
+      "step": 2620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007818900650427242,
+      "loss": 0.6726,
+      "step": 2630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007818050418739106,
+      "loss": 0.7921,
+      "step": 2640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007817200187050972,
+      "loss": 0.7714,
+      "step": 2650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007816349955362837,
+      "loss": 0.693,
+      "step": 2660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007815499723674702,
+      "loss": 0.7722,
+      "step": 2670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007814649491986567,
+      "loss": 0.7474,
+      "step": 2680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007813799260298432,
+      "loss": 0.6708,
+      "step": 2690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007812949028610296,
+      "loss": 0.7534,
+      "step": 2700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007812098796922161,
+      "loss": 0.6444,
+      "step": 2710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007811248565234027,
+      "loss": 0.727,
+      "step": 2720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007810398333545892,
+      "loss": 0.7627,
+      "step": 2730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007809548101857757,
+      "loss": 0.6833,
+      "step": 2740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007808697870169621,
+      "loss": 0.654,
+      "step": 2750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007807847638481487,
+      "loss": 0.8051,
+      "step": 2760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007806997406793352,
+      "loss": 0.7636,
+      "step": 2770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007806147175105216,
+      "loss": 0.7097,
+      "step": 2780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007805296943417082,
+      "loss": 0.7033,
+      "step": 2790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007804446711728946,
+      "loss": 0.708,
+      "step": 2800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007803596480040812,
+      "loss": 0.772,
+      "step": 2810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007802746248352677,
+      "loss": 0.7172,
+      "step": 2820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007801896016664541,
+      "loss": 0.6634,
+      "step": 2830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007801045784976406,
+      "loss": 0.6616,
+      "step": 2840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007800195553288272,
+      "loss": 0.687,
+      "step": 2850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007799345321600136,
+      "loss": 0.7416,
+      "step": 2860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007798495089912002,
+      "loss": 0.9109,
+      "step": 2870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007797644858223866,
+      "loss": 0.6803,
+      "step": 2880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0007796794626535731,
+      "loss": 0.6477,
+      "step": 2890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007795944394847597,
+      "loss": 0.7915,
+      "step": 2900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007795094163159461,
+      "loss": 0.6594,
+      "step": 2910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007794243931471326,
+      "loss": 0.6788,
+      "step": 2920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007793393699783191,
+      "loss": 0.7576,
+      "step": 2930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007792543468095057,
+      "loss": 0.7752,
+      "step": 2940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007791693236406922,
+      "loss": 0.7395,
+      "step": 2950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007790843004718786,
+      "loss": 0.7189,
+      "step": 2960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007789992773030651,
+      "loss": 0.7133,
+      "step": 2970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007789142541342516,
+      "loss": 0.7829,
+      "step": 2980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000778829230965438,
+      "loss": 0.654,
+      "step": 2990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007787442077966246,
+      "loss": 0.7112,
+      "step": 3000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0007786591846278112,
+      "loss": 0.6978,
+      "step": 3010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007785741614589976,
+      "loss": 0.7266,
+      "step": 3020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007784891382901841,
+      "loss": 0.7459,
+      "step": 3030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007784041151213706,
+      "loss": 0.6782,
+      "step": 3040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007783190919525571,
+      "loss": 0.717,
+      "step": 3050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007782340687837436,
+      "loss": 0.7709,
+      "step": 3060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00077814904561493,
+      "loss": 0.7078,
+      "step": 3070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007780640224461167,
+      "loss": 0.6715,
+      "step": 3080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007779789992773032,
+      "loss": 0.6962,
+      "step": 3090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007778939761084896,
+      "loss": 0.7161,
+      "step": 3100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007778089529396761,
+      "loss": 0.6423,
+      "step": 3110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007777239297708626,
+      "loss": 0.7303,
+      "step": 3120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0007776389066020491,
+      "loss": 0.7575,
+      "step": 3130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007775538834332356,
+      "loss": 0.7039,
+      "step": 3140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000777468860264422,
+      "loss": 0.7102,
+      "step": 3150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007773838370956086,
+      "loss": 0.6789,
+      "step": 3160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007772988139267952,
+      "loss": 0.6803,
+      "step": 3170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007772137907579816,
+      "loss": 0.7946,
+      "step": 3180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007771287675891681,
+      "loss": 0.7099,
+      "step": 3190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007770437444203546,
+      "loss": 0.7355,
+      "step": 3200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000776958721251541,
+      "loss": 0.7575,
+      "step": 3210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007768736980827276,
+      "loss": 0.7422,
+      "step": 3220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007767886749139141,
+      "loss": 0.6728,
+      "step": 3230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007767036517451006,
+      "loss": 0.7061,
+      "step": 3240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0007766186285762871,
+      "loss": 0.6997,
+      "step": 3250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007765336054074735,
+      "loss": 0.6632,
+      "step": 3260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007764485822386601,
+      "loss": 0.5969,
+      "step": 3270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007763635590698466,
+      "loss": 0.6515,
+      "step": 3280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000776278535901033,
+      "loss": 0.6988,
+      "step": 3290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007761935127322196,
+      "loss": 0.6692,
+      "step": 3300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000776108489563406,
+      "loss": 0.7407,
+      "step": 3310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007760234663945926,
+      "loss": 0.6621,
+      "step": 3320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007759384432257791,
+      "loss": 0.7047,
+      "step": 3330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007758534200569655,
+      "loss": 0.6649,
+      "step": 3340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000775768396888152,
+      "loss": 0.7466,
+      "step": 3350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0007756833737193386,
+      "loss": 0.8027,
+      "step": 3360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007755983505505251,
+      "loss": 0.6974,
+      "step": 3370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007755133273817116,
+      "loss": 0.6662,
+      "step": 3380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000775428304212898,
+      "loss": 0.7549,
+      "step": 3390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007753432810440845,
+      "loss": 0.7117,
+      "step": 3400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007752582578752711,
+      "loss": 0.6722,
+      "step": 3410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007751732347064575,
+      "loss": 0.7276,
+      "step": 3420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000775088211537644,
+      "loss": 0.7092,
+      "step": 3430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007750031883688305,
+      "loss": 0.6493,
+      "step": 3440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007749181652000171,
+      "loss": 0.7223,
+      "step": 3450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007748331420312036,
+      "loss": 0.6892,
+      "step": 3460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00077474811886239,
+      "loss": 0.7683,
+      "step": 3470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0007746630956935765,
+      "loss": 0.7251,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000774578072524763,
+      "loss": 0.6871,
+      "step": 3490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007744930493559495,
+      "loss": 0.6653,
+      "step": 3500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000774408026187136,
+      "loss": 0.7666,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007743230030183226,
+      "loss": 0.6413,
+      "step": 3520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000774237979849509,
+      "loss": 0.7404,
+      "step": 3530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007741529566806956,
+      "loss": 0.7443,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000774067933511882,
+      "loss": 0.6234,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007739829103430685,
+      "loss": 0.7206,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000773897887174255,
+      "loss": 0.7005,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007738128640054414,
+      "loss": 0.7152,
+      "step": 3580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007737278408366281,
+      "loss": 0.6789,
+      "step": 3590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0007736428176678146,
+      "loss": 0.7389,
+      "step": 3600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000773557794499001,
+      "loss": 0.7544,
+      "step": 3610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007734727713301875,
+      "loss": 0.8116,
+      "step": 3620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000773387748161374,
+      "loss": 0.7118,
+      "step": 3630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007733027249925605,
+      "loss": 0.6055,
+      "step": 3640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000773217701823747,
+      "loss": 0.7238,
+      "step": 3650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007731326786549334,
+      "loss": 0.5918,
+      "step": 3660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00077304765548612,
+      "loss": 0.7289,
+      "step": 3670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007729626323173066,
+      "loss": 0.8281,
+      "step": 3680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000772877609148493,
+      "loss": 0.672,
+      "step": 3690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007727925859796795,
+      "loss": 0.7042,
+      "step": 3700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000772707562810866,
+      "loss": 0.7994,
+      "step": 3710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0007726225396420524,
+      "loss": 0.7359,
+      "step": 3720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000772537516473239,
+      "loss": 0.7361,
+      "step": 3730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007724524933044255,
+      "loss": 0.7305,
+      "step": 3740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000772367470135612,
+      "loss": 0.7288,
+      "step": 3750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007722824469667985,
+      "loss": 0.6774,
+      "step": 3760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000772197423797985,
+      "loss": 0.7819,
+      "step": 3770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007721124006291715,
+      "loss": 0.6595,
+      "step": 3780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000772027377460358,
+      "loss": 0.705,
+      "step": 3790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007719423542915444,
+      "loss": 0.7132,
+      "step": 3800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000771857331122731,
+      "loss": 0.7375,
+      "step": 3810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007717723079539175,
+      "loss": 0.6283,
+      "step": 3820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000771687284785104,
+      "loss": 0.6815,
+      "step": 3830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007716022616162905,
+      "loss": 0.7184,
+      "step": 3840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007715172384474769,
+      "loss": 0.7061,
+      "step": 3850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007714322152786634,
+      "loss": 0.7373,
+      "step": 3860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00077134719210985,
+      "loss": 0.7147,
+      "step": 3870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007712621689410365,
+      "loss": 0.6838,
+      "step": 3880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000771177145772223,
+      "loss": 0.6691,
+      "step": 3890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007710921226034094,
+      "loss": 0.7164,
+      "step": 3900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000771007099434596,
+      "loss": 0.6456,
+      "step": 3910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007709220762657825,
+      "loss": 0.6684,
+      "step": 3920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007708370530969689,
+      "loss": 0.6953,
+      "step": 3930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007707520299281554,
+      "loss": 0.7288,
+      "step": 3940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000770667006759342,
+      "loss": 0.7789,
+      "step": 3950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007705819835905285,
+      "loss": 0.7132,
+      "step": 3960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000770496960421715,
+      "loss": 0.7505,
+      "step": 3970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007704119372529014,
+      "loss": 0.7742,
+      "step": 3980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007703269140840879,
+      "loss": 0.6201,
+      "step": 3990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007702418909152745,
+      "loss": 0.7321,
+      "step": 4000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007701568677464609,
+      "loss": 0.7361,
+      "step": 4010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007700718445776474,
+      "loss": 0.7302,
+      "step": 4020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000769986821408834,
+      "loss": 0.7512,
+      "step": 4030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007699017982400204,
+      "loss": 0.6857,
+      "step": 4040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000769816775071207,
+      "loss": 0.7013,
+      "step": 4050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007697317519023935,
+      "loss": 0.6133,
+      "step": 4060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007696467287335799,
+      "loss": 0.6962,
+      "step": 4070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007695617055647664,
+      "loss": 0.6808,
+      "step": 4080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007694766823959528,
+      "loss": 0.6303,
+      "step": 4090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007693916592271395,
+      "loss": 0.7292,
+      "step": 4100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000769306636058326,
+      "loss": 0.768,
+      "step": 4110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007692216128895124,
+      "loss": 0.7236,
+      "step": 4120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007691365897206989,
+      "loss": 0.8015,
+      "step": 4130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007690515665518855,
+      "loss": 0.7448,
+      "step": 4140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007689665433830719,
+      "loss": 0.6139,
+      "step": 4150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007688815202142584,
+      "loss": 0.678,
+      "step": 4160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007687964970454449,
+      "loss": 0.6328,
+      "step": 4170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007687114738766314,
+      "loss": 0.7164,
+      "step": 4180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000768626450707818,
+      "loss": 0.7158,
+      "step": 4190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007685414275390044,
+      "loss": 0.7882,
+      "step": 4200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007684564043701909,
+      "loss": 0.7636,
+      "step": 4210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007683713812013774,
+      "loss": 0.7856,
+      "step": 4220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007682863580325639,
+      "loss": 0.6878,
+      "step": 4230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007682013348637504,
+      "loss": 0.6704,
+      "step": 4240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007681163116949369,
+      "loss": 0.7391,
+      "step": 4250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007680312885261234,
+      "loss": 0.6507,
+      "step": 4260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00076794626535731,
+      "loss": 0.7249,
+      "step": 4270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007678612421884964,
+      "loss": 0.6175,
+      "step": 4280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007677762190196829,
+      "loss": 0.5912,
+      "step": 4290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007676911958508694,
+      "loss": 0.7189,
+      "step": 4300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007676061726820558,
+      "loss": 0.7479,
+      "step": 4310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007675211495132425,
+      "loss": 0.7201,
+      "step": 4320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007674361263444289,
+      "loss": 0.7613,
+      "step": 4330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007673511031756154,
+      "loss": 0.6413,
+      "step": 4340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007672660800068019,
+      "loss": 0.6809,
+      "step": 4350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007671810568379883,
+      "loss": 0.7114,
+      "step": 4360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007670960336691749,
+      "loss": 0.7544,
+      "step": 4370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007670110105003614,
+      "loss": 0.6518,
+      "step": 4380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007669259873315479,
+      "loss": 0.7294,
+      "step": 4390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007668409641627344,
+      "loss": 0.6749,
+      "step": 4400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007667559409939208,
+      "loss": 0.7264,
+      "step": 4410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007666709178251074,
+      "loss": 0.7251,
+      "step": 4420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007665858946562939,
+      "loss": 0.7337,
+      "step": 4430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007665008714874803,
+      "loss": 0.7678,
+      "step": 4440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007664158483186668,
+      "loss": 0.7846,
+      "step": 4450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007663308251498535,
+      "loss": 0.6983,
+      "step": 4460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007662458019810399,
+      "loss": 0.7187,
+      "step": 4470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007661607788122264,
+      "loss": 0.6009,
+      "step": 4480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007660757556434129,
+      "loss": 0.6935,
+      "step": 4490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007659907324745993,
+      "loss": 0.7455,
+      "step": 4500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007659057093057859,
+      "loss": 0.7753,
+      "step": 4510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007658206861369723,
+      "loss": 0.6608,
+      "step": 4520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007657356629681588,
+      "loss": 0.6807,
+      "step": 4530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007656506397993454,
+      "loss": 0.6874,
+      "step": 4540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007655656166305319,
+      "loss": 0.7762,
+      "step": 4550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007654805934617184,
+      "loss": 0.7613,
+      "step": 4560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007653955702929049,
+      "loss": 0.6904,
+      "step": 4570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007653105471240913,
+      "loss": 0.6786,
+      "step": 4580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007652255239552778,
+      "loss": 0.6694,
+      "step": 4590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007651405007864644,
+      "loss": 0.7172,
+      "step": 4600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007650554776176509,
+      "loss": 0.7225,
+      "step": 4610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007649704544488374,
+      "loss": 0.6926,
+      "step": 4620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007648854312800238,
+      "loss": 0.7572,
+      "step": 4630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007648004081112103,
+      "loss": 0.7897,
+      "step": 4640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007647153849423969,
+      "loss": 0.6763,
+      "step": 4650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007646303617735833,
+      "loss": 0.809,
+      "step": 4660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007645453386047698,
+      "loss": 0.7678,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007644603154359563,
+      "loss": 0.624,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007643752922671429,
+      "loss": 0.731,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007642902690983294,
+      "loss": 0.7494,
+      "step": 4700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007642052459295158,
+      "loss": 0.7323,
+      "step": 4710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007641202227607023,
+      "loss": 0.7292,
+      "step": 4720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007640351995918888,
+      "loss": 0.7321,
+      "step": 4730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007639501764230753,
+      "loss": 0.7439,
+      "step": 4740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007638651532542619,
+      "loss": 0.7044,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007637801300854483,
+      "loss": 0.7102,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007636951069166348,
+      "loss": 0.6605,
+      "step": 4770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0007636100837478214,
+      "loss": 0.6988,
+      "step": 4780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007635250605790078,
+      "loss": 0.6523,
+      "step": 4790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007634400374101943,
+      "loss": 0.6828,
+      "step": 4800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007633550142413808,
+      "loss": 0.714,
+      "step": 4810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007632699910725672,
+      "loss": 0.7565,
+      "step": 4820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007631849679037539,
+      "loss": 0.6712,
+      "step": 4830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007630999447349403,
+      "loss": 0.662,
+      "step": 4840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007630149215661268,
+      "loss": 0.6726,
+      "step": 4850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007629298983973133,
+      "loss": 0.7241,
+      "step": 4860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007628448752284997,
+      "loss": 0.7914,
+      "step": 4870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007627598520596863,
+      "loss": 0.7068,
+      "step": 4880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007626748288908728,
+      "loss": 0.7077,
+      "step": 4890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0007625898057220593,
+      "loss": 0.7564,
+      "step": 4900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007625047825532458,
+      "loss": 0.7748,
+      "step": 4910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007624197593844323,
+      "loss": 0.772,
+      "step": 4920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007623347362156188,
+      "loss": 0.7422,
+      "step": 4930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007622497130468053,
+      "loss": 0.5846,
+      "step": 4940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007621646898779917,
+      "loss": 0.7109,
+      "step": 4950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007620796667091782,
+      "loss": 0.7692,
+      "step": 4960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007619946435403649,
+      "loss": 0.7238,
+      "step": 4970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007619096203715513,
+      "loss": 0.6938,
+      "step": 4980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007618245972027378,
+      "loss": 0.7215,
+      "step": 4990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007617395740339243,
+      "loss": 0.6589,
+      "step": 5000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007616545508651108,
+      "loss": 0.7576,
+      "step": 5010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0007615695276962973,
+      "loss": 0.755,
+      "step": 5020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007614845045274837,
+      "loss": 0.7218,
+      "step": 5030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007613994813586702,
+      "loss": 0.7433,
+      "step": 5040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007613144581898568,
+      "loss": 0.6895,
+      "step": 5050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007612294350210433,
+      "loss": 0.6702,
+      "step": 5060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007611444118522298,
+      "loss": 0.7038,
+      "step": 5070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007610593886834163,
+      "loss": 0.7272,
+      "step": 5080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007609743655146027,
+      "loss": 0.7057,
+      "step": 5090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007608893423457892,
+      "loss": 0.7239,
+      "step": 5100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007608043191769758,
+      "loss": 0.7045,
+      "step": 5110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007607192960081623,
+      "loss": 0.6336,
+      "step": 5120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007606342728393488,
+      "loss": 0.7212,
+      "step": 5130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0007605492496705352,
+      "loss": 0.6987,
+      "step": 5140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007604642265017218,
+      "loss": 0.6216,
+      "step": 5150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007603792033329083,
+      "loss": 0.6824,
+      "step": 5160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007602941801640947,
+      "loss": 0.705,
+      "step": 5170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007602091569952812,
+      "loss": 0.6263,
+      "step": 5180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007601241338264677,
+      "loss": 0.7186,
+      "step": 5190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007600391106576543,
+      "loss": 0.8147,
+      "step": 5200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007599540874888408,
+      "loss": 0.6626,
+      "step": 5210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007598690643200272,
+      "loss": 0.6822,
+      "step": 5220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007597840411512137,
+      "loss": 0.689,
+      "step": 5230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007596990179824003,
+      "loss": 0.6718,
+      "step": 5240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007596139948135867,
+      "loss": 0.6669,
+      "step": 5250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0007595289716447733,
+      "loss": 0.6519,
+      "step": 5260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007594439484759597,
+      "loss": 0.7903,
+      "step": 5270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007593589253071462,
+      "loss": 0.6302,
+      "step": 5280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007592739021383328,
+      "loss": 0.7561,
+      "step": 5290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007591888789695192,
+      "loss": 0.6895,
+      "step": 5300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007591038558007057,
+      "loss": 0.7676,
+      "step": 5310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007590188326318922,
+      "loss": 0.6854,
+      "step": 5320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007589338094630786,
+      "loss": 0.6189,
+      "step": 5330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007588487862942653,
+      "loss": 0.7854,
+      "step": 5340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007587637631254517,
+      "loss": 0.6958,
+      "step": 5350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007586787399566382,
+      "loss": 0.6886,
+      "step": 5360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0007585937167878247,
+      "loss": 0.6719,
+      "step": 5370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007585086936190112,
+      "loss": 0.7076,
+      "step": 5380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007584236704501977,
+      "loss": 0.6356,
+      "step": 5390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007583386472813842,
+      "loss": 0.6954,
+      "step": 5400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007582536241125707,
+      "loss": 0.671,
+      "step": 5410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007581686009437572,
+      "loss": 0.7579,
+      "step": 5420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007580835777749438,
+      "loss": 0.7106,
+      "step": 5430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007579985546061302,
+      "loss": 0.6864,
+      "step": 5440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007579135314373167,
+      "loss": 0.7031,
+      "step": 5450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007578285082685031,
+      "loss": 0.7211,
+      "step": 5460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007577434850996897,
+      "loss": 0.7038,
+      "step": 5470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007576584619308763,
+      "loss": 0.6919,
+      "step": 5480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0007575734387620627,
+      "loss": 0.6557,
+      "step": 5490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007574884155932492,
+      "loss": 0.6823,
+      "step": 5500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007574033924244357,
+      "loss": 0.7419,
+      "step": 5510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007573183692556222,
+      "loss": 0.6807,
+      "step": 5520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007572333460868087,
+      "loss": 0.7434,
+      "step": 5530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007571483229179952,
+      "loss": 0.7314,
+      "step": 5540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007570632997491817,
+      "loss": 0.6414,
+      "step": 5550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007569782765803683,
+      "loss": 0.7562,
+      "step": 5560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007568932534115547,
+      "loss": 0.7281,
+      "step": 5570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007568082302427412,
+      "loss": 0.6913,
+      "step": 5580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007567232070739277,
+      "loss": 0.6921,
+      "step": 5590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007566381839051141,
+      "loss": 0.6635,
+      "step": 5600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0007565531607363007,
+      "loss": 0.6595,
+      "step": 5610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007564681375674872,
+      "loss": 0.7338,
+      "step": 5620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007563831143986737,
+      "loss": 0.7199,
+      "step": 5630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007562980912298602,
+      "loss": 0.6408,
+      "step": 5640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007562130680610466,
+      "loss": 0.7148,
+      "step": 5650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007561280448922332,
+      "loss": 0.655,
+      "step": 5660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007560430217234197,
+      "loss": 0.6604,
+      "step": 5670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007559579985546061,
+      "loss": 0.7158,
+      "step": 5680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007558729753857926,
+      "loss": 0.6591,
+      "step": 5690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007557879522169792,
+      "loss": 0.643,
+      "step": 5700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007557029290481657,
+      "loss": 0.581,
+      "step": 5710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007556179058793522,
+      "loss": 0.6965,
+      "step": 5720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0007555328827105386,
+      "loss": 0.7058,
+      "step": 5730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007554478595417251,
+      "loss": 0.6903,
+      "step": 5740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007553628363729117,
+      "loss": 0.7155,
+      "step": 5750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007552778132040981,
+      "loss": 0.6786,
+      "step": 5760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007551927900352847,
+      "loss": 0.7005,
+      "step": 5770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007551077668664711,
+      "loss": 0.6494,
+      "step": 5780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007550227436976577,
+      "loss": 0.6784,
+      "step": 5790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007549377205288442,
+      "loss": 0.7343,
+      "step": 5800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007548526973600306,
+      "loss": 0.6875,
+      "step": 5810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007547676741912171,
+      "loss": 0.668,
+      "step": 5820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007546826510224036,
+      "loss": 0.6673,
+      "step": 5830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007545976278535902,
+      "loss": 0.6596,
+      "step": 5840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0007545126046847767,
+      "loss": 0.7206,
+      "step": 5850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007544275815159632,
+      "loss": 0.7849,
+      "step": 5860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007543425583471496,
+      "loss": 0.6805,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007542575351783361,
+      "loss": 0.7302,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007541725120095226,
+      "loss": 0.6838,
+      "step": 5890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007540874888407091,
+      "loss": 0.6555,
+      "step": 5900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007540024656718956,
+      "loss": 0.7326,
+      "step": 5910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007539174425030821,
+      "loss": 0.7246,
+      "step": 5920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007538324193342687,
+      "loss": 0.6881,
+      "step": 5930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007537473961654552,
+      "loss": 0.6667,
+      "step": 5940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007536623729966416,
+      "loss": 0.634,
+      "step": 5950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007535773498278281,
+      "loss": 0.6474,
+      "step": 5960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0007534923266590146,
+      "loss": 0.6187,
+      "step": 5970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007534073034902011,
+      "loss": 0.7666,
+      "step": 5980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007533222803213877,
+      "loss": 0.7198,
+      "step": 5990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007532372571525741,
+      "loss": 0.696,
+      "step": 6000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007531522339837606,
+      "loss": 0.7566,
+      "step": 6010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007530672108149472,
+      "loss": 0.63,
+      "step": 6020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007529821876461336,
+      "loss": 0.6581,
+      "step": 6030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007528971644773201,
+      "loss": 0.6976,
+      "step": 6040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007528121413085066,
+      "loss": 0.6176,
+      "step": 6050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007527271181396931,
+      "loss": 0.6785,
+      "step": 6060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007526420949708797,
+      "loss": 0.7164,
+      "step": 6070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0007525570718020661,
+      "loss": 0.6452,
+      "step": 6080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007524720486332526,
+      "loss": 0.6935,
+      "step": 6090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007523870254644391,
+      "loss": 0.7008,
+      "step": 6100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007523020022956255,
+      "loss": 0.6934,
+      "step": 6110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007522169791268121,
+      "loss": 0.6781,
+      "step": 6120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007521319559579986,
+      "loss": 0.6786,
+      "step": 6130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007520469327891851,
+      "loss": 0.7028,
+      "step": 6140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007519619096203716,
+      "loss": 0.67,
+      "step": 6150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007518768864515581,
+      "loss": 0.6842,
+      "step": 6160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007517918632827446,
+      "loss": 0.7099,
+      "step": 6170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007517068401139311,
+      "loss": 0.7041,
+      "step": 6180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0007516218169451175,
+      "loss": 0.743,
+      "step": 6190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000751536793776304,
+      "loss": 0.7169,
+      "step": 6200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007514517706074906,
+      "loss": 0.7623,
+      "step": 6210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007513667474386771,
+      "loss": 0.6625,
+      "step": 6220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007512817242698636,
+      "loss": 0.6688,
+      "step": 6230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00075119670110105,
+      "loss": 0.6879,
+      "step": 6240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007511116779322366,
+      "loss": 0.7571,
+      "step": 6250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007510266547634231,
+      "loss": 0.709,
+      "step": 6260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007509416315946095,
+      "loss": 0.6918,
+      "step": 6270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007508566084257961,
+      "loss": 0.7722,
+      "step": 6280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007507715852569825,
+      "loss": 0.6913,
+      "step": 6290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007506865620881691,
+      "loss": 0.6587,
+      "step": 6300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0007506015389193556,
+      "loss": 0.6962,
+      "step": 6310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000750516515750542,
+      "loss": 0.6856,
+      "step": 6320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007504314925817285,
+      "loss": 0.6659,
+      "step": 6330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000750346469412915,
+      "loss": 0.657,
+      "step": 6340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007502614462441016,
+      "loss": 0.6189,
+      "step": 6350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007501764230752881,
+      "loss": 0.6475,
+      "step": 6360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007500913999064746,
+      "loss": 0.8106,
+      "step": 6370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000750006376737661,
+      "loss": 0.7119,
+      "step": 6380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007499213535688476,
+      "loss": 0.7514,
+      "step": 6390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000749836330400034,
+      "loss": 0.6992,
+      "step": 6400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007497513072312205,
+      "loss": 0.6595,
+      "step": 6410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000749666284062407,
+      "loss": 0.6774,
+      "step": 6420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007495812608935935,
+      "loss": 0.7182,
+      "step": 6430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0007494962377247801,
+      "loss": 0.6552,
+      "step": 6440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007494112145559666,
+      "loss": 0.6276,
+      "step": 6450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000749326191387153,
+      "loss": 0.6982,
+      "step": 6460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007492411682183395,
+      "loss": 0.6596,
+      "step": 6470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007491561450495261,
+      "loss": 0.6908,
+      "step": 6480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007490711218807125,
+      "loss": 0.6987,
+      "step": 6490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007489860987118991,
+      "loss": 0.677,
+      "step": 6500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007489010755430855,
+      "loss": 0.7492,
+      "step": 6510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000748816052374272,
+      "loss": 0.6414,
+      "step": 6520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007487310292054586,
+      "loss": 0.7329,
+      "step": 6530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000748646006036645,
+      "loss": 0.6092,
+      "step": 6540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0007485609828678315,
+      "loss": 0.7037,
+      "step": 6550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000748475959699018,
+      "loss": 0.6448,
+      "step": 6560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007483909365302046,
+      "loss": 0.7304,
+      "step": 6570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007483059133613911,
+      "loss": 0.7186,
+      "step": 6580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007482208901925775,
+      "loss": 0.6727,
+      "step": 6590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000748135867023764,
+      "loss": 0.7504,
+      "step": 6600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007480508438549505,
+      "loss": 0.6347,
+      "step": 6610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000747965820686137,
+      "loss": 0.6889,
+      "step": 6620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007478807975173235,
+      "loss": 0.6919,
+      "step": 6630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00074779577434851,
+      "loss": 0.7278,
+      "step": 6640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007477107511796965,
+      "loss": 0.706,
+      "step": 6650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000747625728010883,
+      "loss": 0.6918,
+      "step": 6660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0007475407048420695,
+      "loss": 0.6988,
+      "step": 6670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000747455681673256,
+      "loss": 0.648,
+      "step": 6680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007473706585044425,
+      "loss": 0.6988,
+      "step": 6690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007472856353356289,
+      "loss": 0.6687,
+      "step": 6700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007472006121668155,
+      "loss": 0.6471,
+      "step": 6710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000747115588998002,
+      "loss": 0.6575,
+      "step": 6720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007470305658291885,
+      "loss": 0.7523,
+      "step": 6730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000746945542660375,
+      "loss": 0.7099,
+      "step": 6740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007468605194915614,
+      "loss": 0.7465,
+      "step": 6750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000746775496322748,
+      "loss": 0.6884,
+      "step": 6760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007466904731539345,
+      "loss": 0.6615,
+      "step": 6770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007466054499851209,
+      "loss": 0.7045,
+      "step": 6780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0007465204268163075,
+      "loss": 0.71,
+      "step": 6790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007464354036474941,
+      "loss": 0.7377,
+      "step": 6800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007463503804786805,
+      "loss": 0.788,
+      "step": 6810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000746265357309867,
+      "loss": 0.6459,
+      "step": 6820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007461803341410534,
+      "loss": 0.7031,
+      "step": 6830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007460953109722399,
+      "loss": 0.7351,
+      "step": 6840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007460102878034265,
+      "loss": 0.6491,
+      "step": 6850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000745925264634613,
+      "loss": 0.6393,
+      "step": 6860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007458402414657995,
+      "loss": 0.7237,
+      "step": 6870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000745755218296986,
+      "loss": 0.7898,
+      "step": 6880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007456701951281724,
+      "loss": 0.7257,
+      "step": 6890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000745585171959359,
+      "loss": 0.6982,
+      "step": 6900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007455001487905455,
+      "loss": 0.6504,
+      "step": 6910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007454151256217319,
+      "loss": 0.7206,
+      "step": 6920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007453301024529185,
+      "loss": 0.7374,
+      "step": 6930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000745245079284105,
+      "loss": 0.6388,
+      "step": 6940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007451600561152915,
+      "loss": 0.694,
+      "step": 6950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000745075032946478,
+      "loss": 0.7532,
+      "step": 6960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007449900097776644,
+      "loss": 0.7296,
+      "step": 6970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007449049866088509,
+      "loss": 0.6811,
+      "step": 6980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007448199634400375,
+      "loss": 0.7213,
+      "step": 6990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007447349402712239,
+      "loss": 0.707,
+      "step": 7000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007446499171024105,
+      "loss": 0.6966,
+      "step": 7010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007445648939335969,
+      "loss": 0.7702,
+      "step": 7020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007444798707647835,
+      "loss": 0.6878,
+      "step": 7030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00074439484759597,
+      "loss": 0.6578,
+      "step": 7040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007443098244271564,
+      "loss": 0.6328,
+      "step": 7050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007442248012583429,
+      "loss": 0.7383,
+      "step": 7060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007441397780895294,
+      "loss": 0.7126,
+      "step": 7070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000744054754920716,
+      "loss": 0.675,
+      "step": 7080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007439697317519025,
+      "loss": 0.7084,
+      "step": 7090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007438847085830889,
+      "loss": 0.6615,
+      "step": 7100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007437996854142754,
+      "loss": 0.7478,
+      "step": 7110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000743714662245462,
+      "loss": 0.6806,
+      "step": 7120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007436296390766484,
+      "loss": 0.6572,
+      "step": 7130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007435446159078349,
+      "loss": 0.6538,
+      "step": 7140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007434595927390214,
+      "loss": 0.7205,
+      "step": 7150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007433745695702079,
+      "loss": 0.7214,
+      "step": 7160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007432895464013945,
+      "loss": 0.7325,
+      "step": 7170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007432045232325809,
+      "loss": 0.7118,
+      "step": 7180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007431195000637674,
+      "loss": 0.6976,
+      "step": 7190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007430344768949539,
+      "loss": 0.7029,
+      "step": 7200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007429494537261403,
+      "loss": 0.6336,
+      "step": 7210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000742864430557327,
+      "loss": 0.7095,
+      "step": 7220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007427794073885134,
+      "loss": 0.7861,
+      "step": 7230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007426943842196999,
+      "loss": 0.7064,
+      "step": 7240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007426093610508864,
+      "loss": 0.6818,
+      "step": 7250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007425243378820728,
+      "loss": 0.7326,
+      "step": 7260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007424393147132594,
+      "loss": 0.7345,
+      "step": 7270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007423542915444459,
+      "loss": 0.6686,
+      "step": 7280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007422692683756323,
+      "loss": 0.6743,
+      "step": 7290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007421842452068189,
+      "loss": 0.6413,
+      "step": 7300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007420992220380055,
+      "loss": 0.7402,
+      "step": 7310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007420141988691919,
+      "loss": 0.6669,
+      "step": 7320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007419291757003784,
+      "loss": 0.6883,
+      "step": 7330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007418441525315648,
+      "loss": 0.7222,
+      "step": 7340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007417591293627513,
+      "loss": 0.7447,
+      "step": 7350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007416741061939379,
+      "loss": 0.6347,
+      "step": 7360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007415890830251244,
+      "loss": 0.6974,
+      "step": 7370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007415040598563109,
+      "loss": 0.6847,
+      "step": 7380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0007414190366874974,
+      "loss": 0.7519,
+      "step": 7390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007413340135186839,
+      "loss": 0.6959,
+      "step": 7400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007412489903498704,
+      "loss": 0.6601,
+      "step": 7410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007411639671810569,
+      "loss": 0.6581,
+      "step": 7420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007410789440122433,
+      "loss": 0.6767,
+      "step": 7430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00074099392084343,
+      "loss": 0.708,
+      "step": 7440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007409088976746164,
+      "loss": 0.6596,
+      "step": 7450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007408238745058029,
+      "loss": 0.7314,
+      "step": 7460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007407388513369894,
+      "loss": 0.6278,
+      "step": 7470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007406538281681758,
+      "loss": 0.6386,
+      "step": 7480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007405688049993624,
+      "loss": 0.7413,
+      "step": 7490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007404837818305489,
+      "loss": 0.7307,
+      "step": 7500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007403987586617353,
+      "loss": 0.6925,
+      "step": 7510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007403137354929219,
+      "loss": 0.6456,
+      "step": 7520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007402287123241083,
+      "loss": 0.7142,
+      "step": 7530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007401436891552949,
+      "loss": 0.6787,
+      "step": 7540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007400586659864814,
+      "loss": 0.6938,
+      "step": 7550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007399736428176678,
+      "loss": 0.7385,
+      "step": 7560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007398886196488543,
+      "loss": 0.6141,
+      "step": 7570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007398035964800408,
+      "loss": 0.6976,
+      "step": 7580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007397185733112274,
+      "loss": 0.6613,
+      "step": 7590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007396335501424139,
+      "loss": 0.6509,
+      "step": 7600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007395485269736003,
+      "loss": 0.6217,
+      "step": 7610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007394635038047868,
+      "loss": 0.6486,
+      "step": 7620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007393784806359734,
+      "loss": 0.8031,
+      "step": 7630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007392934574671598,
+      "loss": 0.6746,
+      "step": 7640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007392084342983463,
+      "loss": 0.7116,
+      "step": 7650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007391234111295328,
+      "loss": 0.6962,
+      "step": 7660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007390383879607193,
+      "loss": 0.7209,
+      "step": 7670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007389533647919059,
+      "loss": 0.6569,
+      "step": 7680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007388683416230923,
+      "loss": 0.7466,
+      "step": 7690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007387833184542788,
+      "loss": 0.6846,
+      "step": 7700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007386982952854653,
+      "loss": 0.7095,
+      "step": 7710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007386132721166517,
+      "loss": 0.713,
+      "step": 7720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007385282489478384,
+      "loss": 0.6436,
+      "step": 7730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007384432257790249,
+      "loss": 0.7201,
+      "step": 7740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007383582026102113,
+      "loss": 0.69,
+      "step": 7750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007382731794413978,
+      "loss": 0.7046,
+      "step": 7760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007381881562725843,
+      "loss": 0.6142,
+      "step": 7770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007381031331037708,
+      "loss": 0.6201,
+      "step": 7780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007380181099349573,
+      "loss": 0.6895,
+      "step": 7790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007379330867661437,
+      "loss": 0.6599,
+      "step": 7800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007378480635973304,
+      "loss": 0.6011,
+      "step": 7810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007377630404285169,
+      "loss": 0.666,
+      "step": 7820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007376780172597033,
+      "loss": 0.7257,
+      "step": 7830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007375929940908898,
+      "loss": 0.6057,
+      "step": 7840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007375079709220763,
+      "loss": 0.7392,
+      "step": 7850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007374229477532628,
+      "loss": 0.5946,
+      "step": 7860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007373379245844493,
+      "loss": 0.6486,
+      "step": 7870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007372529014156358,
+      "loss": 0.6908,
+      "step": 7880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007371678782468223,
+      "loss": 0.7561,
+      "step": 7890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007370828550780088,
+      "loss": 0.6299,
+      "step": 7900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007369978319091953,
+      "loss": 0.7471,
+      "step": 7910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007369128087403818,
+      "loss": 0.7032,
+      "step": 7920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007368277855715683,
+      "loss": 0.668,
+      "step": 7930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007367427624027547,
+      "loss": 0.6945,
+      "step": 7940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007366577392339414,
+      "loss": 0.732,
+      "step": 7950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007365727160651278,
+      "loss": 0.6786,
+      "step": 7960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007364876928963143,
+      "loss": 0.753,
+      "step": 7970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007364026697275008,
+      "loss": 0.7275,
+      "step": 7980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007363176465586872,
+      "loss": 0.5795,
+      "step": 7990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007362326233898738,
+      "loss": 0.6814,
+      "step": 8000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007361476002210603,
+      "loss": 0.6497,
+      "step": 8010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007360625770522468,
+      "loss": 0.6204,
+      "step": 8020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007359775538834333,
+      "loss": 0.6675,
+      "step": 8030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007358925307146197,
+      "loss": 0.6844,
+      "step": 8040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007358075075458063,
+      "loss": 0.6687,
+      "step": 8050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007357224843769928,
+      "loss": 0.6867,
+      "step": 8060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007356374612081792,
+      "loss": 0.7153,
+      "step": 8070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007355524380393657,
+      "loss": 0.7521,
+      "step": 8080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007354674148705523,
+      "loss": 0.6723,
+      "step": 8090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007353823917017388,
+      "loss": 0.7408,
+      "step": 8100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007352973685329253,
+      "loss": 0.7065,
+      "step": 8110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007352123453641117,
+      "loss": 0.6953,
+      "step": 8120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007351273221952982,
+      "loss": 0.7136,
+      "step": 8130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007350422990264848,
+      "loss": 0.73,
+      "step": 8140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007349572758576712,
+      "loss": 0.6346,
+      "step": 8150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007348722526888577,
+      "loss": 0.675,
+      "step": 8160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007347872295200443,
+      "loss": 0.6373,
+      "step": 8170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007347022063512308,
+      "loss": 0.7409,
+      "step": 8180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007346171831824173,
+      "loss": 0.7434,
+      "step": 8190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007345321600136037,
+      "loss": 0.5941,
+      "step": 8200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007344471368447902,
+      "loss": 0.7175,
+      "step": 8210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007343621136759767,
+      "loss": 0.639,
+      "step": 8220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007342770905071632,
+      "loss": 0.6216,
+      "step": 8230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007341920673383498,
+      "loss": 0.7303,
+      "step": 8240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007341070441695363,
+      "loss": 0.6968,
+      "step": 8250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007340220210007227,
+      "loss": 0.6345,
+      "step": 8260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007339369978319093,
+      "loss": 0.6667,
+      "step": 8270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007338519746630958,
+      "loss": 0.7285,
+      "step": 8280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007337669514942822,
+      "loss": 0.7073,
+      "step": 8290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007336819283254687,
+      "loss": 0.6227,
+      "step": 8300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007335969051566552,
+      "loss": 0.6674,
+      "step": 8310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007335118819878418,
+      "loss": 0.6652,
+      "step": 8320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007334268588190283,
+      "loss": 0.742,
+      "step": 8330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007333418356502147,
+      "loss": 0.6106,
+      "step": 8340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007332568124814012,
+      "loss": 0.6261,
+      "step": 8350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007331717893125877,
+      "loss": 0.6795,
+      "step": 8360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007330867661437742,
+      "loss": 0.6575,
+      "step": 8370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007330017429749607,
+      "loss": 0.7074,
+      "step": 8380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007329167198061472,
+      "loss": 0.7225,
+      "step": 8390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007328316966373337,
+      "loss": 0.6656,
+      "step": 8400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007327466734685203,
+      "loss": 0.6831,
+      "step": 8410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007326616502997067,
+      "loss": 0.678,
+      "step": 8420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007325766271308932,
+      "loss": 0.7097,
+      "step": 8430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007324916039620797,
+      "loss": 0.65,
+      "step": 8440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007324065807932661,
+      "loss": 0.6058,
+      "step": 8450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007323215576244528,
+      "loss": 0.6802,
+      "step": 8460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007322365344556392,
+      "loss": 0.7313,
+      "step": 8470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007321515112868257,
+      "loss": 0.7629,
+      "step": 8480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007320664881180122,
+      "loss": 0.6994,
+      "step": 8490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007319814649491986,
+      "loss": 0.5773,
+      "step": 8500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007318964417803852,
+      "loss": 0.6731,
+      "step": 8510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007318114186115717,
+      "loss": 0.7186,
+      "step": 8520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007317263954427582,
+      "loss": 0.7226,
+      "step": 8530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007316413722739447,
+      "loss": 0.7031,
+      "step": 8540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007315563491051312,
+      "loss": 0.6822,
+      "step": 8550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007314713259363177,
+      "loss": 0.6392,
+      "step": 8560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007313863027675042,
+      "loss": 0.7175,
+      "step": 8570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007313012795986906,
+      "loss": 0.6115,
+      "step": 8580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007312162564298771,
+      "loss": 0.7225,
+      "step": 8590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007311312332610637,
+      "loss": 0.6353,
+      "step": 8600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007310462100922502,
+      "loss": 0.6418,
+      "step": 8610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007309611869234367,
+      "loss": 0.6024,
+      "step": 8620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007308761637546231,
+      "loss": 0.6519,
+      "step": 8630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007307911405858097,
+      "loss": 0.6752,
+      "step": 8640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007307061174169962,
+      "loss": 0.7079,
+      "step": 8650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007306210942481826,
+      "loss": 0.6836,
+      "step": 8660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007305360710793691,
+      "loss": 0.662,
+      "step": 8670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007304510479105557,
+      "loss": 0.7419,
+      "step": 8680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007303660247417422,
+      "loss": 0.6297,
+      "step": 8690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007302810015729287,
+      "loss": 0.6629,
+      "step": 8700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007301959784041151,
+      "loss": 0.6309,
+      "step": 8710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007301109552353016,
+      "loss": 0.6596,
+      "step": 8720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007300259320664881,
+      "loss": 0.7007,
+      "step": 8730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007299409088976746,
+      "loss": 0.6553,
+      "step": 8740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007298558857288612,
+      "loss": 0.6591,
+      "step": 8750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007297708625600477,
+      "loss": 0.7096,
+      "step": 8760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007296858393912341,
+      "loss": 0.6946,
+      "step": 8770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007296008162224207,
+      "loss": 0.6812,
+      "step": 8780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007295157930536072,
+      "loss": 0.6996,
+      "step": 8790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007294307698847936,
+      "loss": 0.6447,
+      "step": 8800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007293457467159801,
+      "loss": 0.6659,
+      "step": 8810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007292607235471666,
+      "loss": 0.7292,
+      "step": 8820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007291757003783532,
+      "loss": 0.7674,
+      "step": 8830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007290906772095397,
+      "loss": 0.6138,
+      "step": 8840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007290056540407261,
+      "loss": 0.7218,
+      "step": 8850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007289206308719126,
+      "loss": 0.6716,
+      "step": 8860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007288356077030992,
+      "loss": 0.7389,
+      "step": 8870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007287505845342856,
+      "loss": 0.672,
+      "step": 8880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007286655613654721,
+      "loss": 0.6498,
+      "step": 8890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007285805381966586,
+      "loss": 0.6911,
+      "step": 8900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007284955150278451,
+      "loss": 0.6535,
+      "step": 8910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007284104918590317,
+      "loss": 0.7016,
+      "step": 8920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007283254686902181,
+      "loss": 0.7125,
+      "step": 8930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007282404455214046,
+      "loss": 0.6637,
+      "step": 8940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007281554223525911,
+      "loss": 0.7209,
+      "step": 8950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007280703991837775,
+      "loss": 0.663,
+      "step": 8960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007279853760149642,
+      "loss": 0.7859,
+      "step": 8970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007279003528461506,
+      "loss": 0.7401,
+      "step": 8980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007278153296773371,
+      "loss": 0.6419,
+      "step": 8990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007277303065085236,
+      "loss": 0.6524,
+      "step": 9000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007276452833397101,
+      "loss": 0.5726,
+      "step": 9010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007275602601708966,
+      "loss": 0.6179,
+      "step": 9020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007274752370020831,
+      "loss": 0.7117,
+      "step": 9030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007273902138332696,
+      "loss": 0.6381,
+      "step": 9040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007273051906644562,
+      "loss": 0.6686,
+      "step": 9050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007272201674956426,
+      "loss": 0.7258,
+      "step": 9060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007271351443268291,
+      "loss": 0.6228,
+      "step": 9070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007270501211580156,
+      "loss": 0.5929,
+      "step": 9080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000726965097989202,
+      "loss": 0.6458,
+      "step": 9090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007268800748203886,
+      "loss": 0.7027,
+      "step": 9100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007267950516515752,
+      "loss": 0.6846,
+      "step": 9110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007267100284827616,
+      "loss": 0.6466,
+      "step": 9120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007266250053139481,
+      "loss": 0.6544,
+      "step": 9130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007265399821451345,
+      "loss": 0.6659,
+      "step": 9140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007264549589763211,
+      "loss": 0.6541,
+      "step": 9150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007263699358075076,
+      "loss": 0.632,
+      "step": 9160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000726284912638694,
+      "loss": 0.6499,
+      "step": 9170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007261998894698805,
+      "loss": 0.7963,
+      "step": 9180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007261148663010672,
+      "loss": 0.5941,
+      "step": 9190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007260298431322536,
+      "loss": 0.6131,
+      "step": 9200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007259448199634401,
+      "loss": 0.7644,
+      "step": 9210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007258597967946266,
+      "loss": 0.6881,
+      "step": 9220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000725774773625813,
+      "loss": 0.7044,
+      "step": 9230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007256897504569996,
+      "loss": 0.7193,
+      "step": 9240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000725604727288186,
+      "loss": 0.6542,
+      "step": 9250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007255197041193726,
+      "loss": 0.6979,
+      "step": 9260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007254346809505591,
+      "loss": 0.7027,
+      "step": 9270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007253496577817455,
+      "loss": 0.6942,
+      "step": 9280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007252646346129321,
+      "loss": 0.6435,
+      "step": 9290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007251796114441186,
+      "loss": 0.6307,
+      "step": 9300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000725094588275305,
+      "loss": 0.6377,
+      "step": 9310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007250095651064915,
+      "loss": 0.6828,
+      "step": 9320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007249245419376781,
+      "loss": 0.6519,
+      "step": 9330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007248395187688646,
+      "loss": 0.6451,
+      "step": 9340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007247544956000511,
+      "loss": 0.68,
+      "step": 9350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007246694724312375,
+      "loss": 0.6655,
+      "step": 9360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000724584449262424,
+      "loss": 0.7024,
+      "step": 9370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007244994260936106,
+      "loss": 0.7844,
+      "step": 9380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000724414402924797,
+      "loss": 0.7587,
+      "step": 9390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0007243293797559836,
+      "loss": 0.6746,
+      "step": 9400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00072424435658717,
+      "loss": 0.6819,
+      "step": 9410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007241593334183566,
+      "loss": 0.6479,
+      "step": 9420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007240743102495431,
+      "loss": 0.6772,
+      "step": 9430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007239892870807295,
+      "loss": 0.6591,
+      "step": 9440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000723904263911916,
+      "loss": 0.6359,
+      "step": 9450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007238192407431025,
+      "loss": 0.7065,
+      "step": 9460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000723734217574289,
+      "loss": 0.7812,
+      "step": 9470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007236491944054756,
+      "loss": 0.7136,
+      "step": 9480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000723564171236662,
+      "loss": 0.653,
+      "step": 9490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0007234791480678485,
+      "loss": 0.6108,
+      "step": 9500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000723394124899035,
+      "loss": 0.6106,
+      "step": 9510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007233091017302215,
+      "loss": 0.739,
+      "step": 9520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000723224078561408,
+      "loss": 0.696,
+      "step": 9530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007231390553925945,
+      "loss": 0.6606,
+      "step": 9540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000723054032223781,
+      "loss": 0.6999,
+      "step": 9550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007229690090549676,
+      "loss": 0.6744,
+      "step": 9560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000722883985886154,
+      "loss": 0.7159,
+      "step": 9570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007227989627173405,
+      "loss": 0.6905,
+      "step": 9580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000722713939548527,
+      "loss": 0.7583,
+      "step": 9590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007226289163797134,
+      "loss": 0.6986,
+      "step": 9600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007225438932109,
+      "loss": 0.6375,
+      "step": 9610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0007224588700420866,
+      "loss": 0.6248,
+      "step": 9620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000722373846873273,
+      "loss": 0.7079,
+      "step": 9630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007222888237044595,
+      "loss": 0.7074,
+      "step": 9640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000722203800535646,
+      "loss": 0.6627,
+      "step": 9650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007221187773668325,
+      "loss": 0.6063,
+      "step": 9660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000722033754198019,
+      "loss": 0.7221,
+      "step": 9670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007219487310292054,
+      "loss": 0.6818,
+      "step": 9680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000721863707860392,
+      "loss": 0.6969,
+      "step": 9690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007217786846915786,
+      "loss": 0.6663,
+      "step": 9700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000721693661522765,
+      "loss": 0.6826,
+      "step": 9710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007216086383539515,
+      "loss": 0.7055,
+      "step": 9720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000721523615185138,
+      "loss": 0.6992,
+      "step": 9730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0007214385920163244,
+      "loss": 0.7099,
+      "step": 9740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000721353568847511,
+      "loss": 0.6007,
+      "step": 9750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007212685456786974,
+      "loss": 0.7426,
+      "step": 9760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000721183522509884,
+      "loss": 0.6913,
+      "step": 9770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007210984993410705,
+      "loss": 0.7048,
+      "step": 9780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000721013476172257,
+      "loss": 0.6473,
+      "step": 9790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007209284530034435,
+      "loss": 0.668,
+      "step": 9800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00072084342983463,
+      "loss": 0.6212,
+      "step": 9810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007207584066658164,
+      "loss": 0.6578,
+      "step": 9820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007206733834970029,
+      "loss": 0.671,
+      "step": 9830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007205883603281895,
+      "loss": 0.693,
+      "step": 9840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000720503337159376,
+      "loss": 0.6909,
+      "step": 9850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007204183139905625,
+      "loss": 0.6747,
+      "step": 9860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007203332908217489,
+      "loss": 0.6498,
+      "step": 9870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007202482676529355,
+      "loss": 0.7061,
+      "step": 9880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000720163244484122,
+      "loss": 0.6402,
+      "step": 9890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007200782213153084,
+      "loss": 0.6089,
+      "step": 9900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000719993198146495,
+      "loss": 0.6106,
+      "step": 9910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007199081749776814,
+      "loss": 0.6137,
+      "step": 9920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000719823151808868,
+      "loss": 0.6721,
+      "step": 9930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007197381286400545,
+      "loss": 0.7512,
+      "step": 9940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007196531054712409,
+      "loss": 0.6888,
+      "step": 9950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007195680823024274,
+      "loss": 0.6624,
+      "step": 9960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000719483059133614,
+      "loss": 0.6682,
+      "step": 9970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007193980359648004,
+      "loss": 0.6549,
+      "step": 9980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000719313012795987,
+      "loss": 0.6673,
+      "step": 9990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007192279896271734,
+      "loss": 0.5905,
+      "step": 10000
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.6319361925125122,
+      "eval_runtime": 312.5988,
+      "eval_samples_per_second": 16.811,
+      "eval_steps_per_second": 2.102,
+      "step": 10000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007191429664583599,
+      "loss": 0.6789,
+      "step": 10010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007190579432895465,
+      "loss": 0.6871,
+      "step": 10020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007189729201207329,
+      "loss": 0.7094,
+      "step": 10030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007188878969519194,
+      "loss": 0.6923,
+      "step": 10040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007188028737831059,
+      "loss": 0.6289,
+      "step": 10050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007187178506142924,
+      "loss": 0.6998,
+      "step": 10060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000718632827445479,
+      "loss": 0.6901,
+      "step": 10070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007185478042766654,
+      "loss": 0.6723,
+      "step": 10080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007184627811078519,
+      "loss": 0.6502,
+      "step": 10090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007183777579390384,
+      "loss": 0.5832,
+      "step": 10100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007182927347702248,
+      "loss": 0.6678,
+      "step": 10110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007182077116014114,
+      "loss": 0.6305,
+      "step": 10120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000718122688432598,
+      "loss": 0.6305,
+      "step": 10130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007180376652637844,
+      "loss": 0.6637,
+      "step": 10140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007179526420949709,
+      "loss": 0.647,
+      "step": 10150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007178676189261575,
+      "loss": 0.6736,
+      "step": 10160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007177825957573439,
+      "loss": 0.6386,
+      "step": 10170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007176975725885304,
+      "loss": 0.7459,
+      "step": 10180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007176125494197168,
+      "loss": 0.6879,
+      "step": 10190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007175275262509035,
+      "loss": 0.6971,
+      "step": 10200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00071744250308209,
+      "loss": 0.68,
+      "step": 10210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007173574799132764,
+      "loss": 0.6286,
+      "step": 10220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007172724567444629,
+      "loss": 0.7676,
+      "step": 10230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007171874335756494,
+      "loss": 0.6728,
+      "step": 10240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007171024104068359,
+      "loss": 0.66,
+      "step": 10250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007170173872380224,
+      "loss": 0.6417,
+      "step": 10260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007169323640692089,
+      "loss": 0.6295,
+      "step": 10270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007168473409003954,
+      "loss": 0.69,
+      "step": 10280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000716762317731582,
+      "loss": 0.6983,
+      "step": 10290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007166772945627684,
+      "loss": 0.6262,
+      "step": 10300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007165922713939549,
+      "loss": 0.595,
+      "step": 10310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007165072482251414,
+      "loss": 0.6508,
+      "step": 10320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007164222250563278,
+      "loss": 0.708,
+      "step": 10330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007163372018875144,
+      "loss": 0.6477,
+      "step": 10340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007162521787187009,
+      "loss": 0.7383,
+      "step": 10350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007161671555498874,
+      "loss": 0.7153,
+      "step": 10360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007160821323810739,
+      "loss": 0.6278,
+      "step": 10370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007159971092122603,
+      "loss": 0.6203,
+      "step": 10380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007159120860434469,
+      "loss": 0.6674,
+      "step": 10390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007158270628746334,
+      "loss": 0.6823,
+      "step": 10400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007157420397058198,
+      "loss": 0.693,
+      "step": 10410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007156570165370064,
+      "loss": 0.6372,
+      "step": 10420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007155719933681928,
+      "loss": 0.6994,
+      "step": 10430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007154869701993794,
+      "loss": 0.6519,
+      "step": 10440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007154019470305659,
+      "loss": 0.7357,
+      "step": 10450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007153169238617523,
+      "loss": 0.6649,
+      "step": 10460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007152319006929388,
+      "loss": 0.7048,
+      "step": 10470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007151468775241254,
+      "loss": 0.6739,
+      "step": 10480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007150618543553119,
+      "loss": 0.6626,
+      "step": 10490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007149768311864984,
+      "loss": 0.6689,
+      "step": 10500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007148918080176848,
+      "loss": 0.7281,
+      "step": 10510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007148067848488713,
+      "loss": 0.6348,
+      "step": 10520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007147217616800579,
+      "loss": 0.6716,
+      "step": 10530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007146367385112443,
+      "loss": 0.6078,
+      "step": 10540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007145517153424308,
+      "loss": 0.6578,
+      "step": 10550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007144666921736173,
+      "loss": 0.6568,
+      "step": 10560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007143816690048039,
+      "loss": 0.6023,
+      "step": 10570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007142966458359904,
+      "loss": 0.7061,
+      "step": 10580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007142116226671769,
+      "loss": 0.6395,
+      "step": 10590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007141265994983633,
+      "loss": 0.6675,
+      "step": 10600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007140415763295498,
+      "loss": 0.6242,
+      "step": 10610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007139565531607363,
+      "loss": 0.6612,
+      "step": 10620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007138715299919228,
+      "loss": 0.6844,
+      "step": 10630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007137865068231094,
+      "loss": 0.6949,
+      "step": 10640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007137014836542958,
+      "loss": 0.6081,
+      "step": 10650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007136164604854824,
+      "loss": 0.6177,
+      "step": 10660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007135314373166689,
+      "loss": 0.7081,
+      "step": 10670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007134464141478553,
+      "loss": 0.7484,
+      "step": 10680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007133613909790418,
+      "loss": 0.6647,
+      "step": 10690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007132763678102283,
+      "loss": 0.652,
+      "step": 10700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007131913446414149,
+      "loss": 0.6842,
+      "step": 10710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007131063214726014,
+      "loss": 0.7122,
+      "step": 10720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007130212983037878,
+      "loss": 0.7427,
+      "step": 10730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007129362751349743,
+      "loss": 0.6746,
+      "step": 10740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007128512519661608,
+      "loss": 0.7206,
+      "step": 10750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007127662287973473,
+      "loss": 0.6362,
+      "step": 10760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007126812056285338,
+      "loss": 0.6724,
+      "step": 10770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007125961824597203,
+      "loss": 0.6416,
+      "step": 10780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007125111592909068,
+      "loss": 0.6676,
+      "step": 10790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007124261361220934,
+      "loss": 0.7071,
+      "step": 10800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007123411129532798,
+      "loss": 0.6981,
+      "step": 10810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007122560897844663,
+      "loss": 0.7514,
+      "step": 10820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007121710666156528,
+      "loss": 0.5922,
+      "step": 10830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007120860434468392,
+      "loss": 0.5998,
+      "step": 10840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007120010202780258,
+      "loss": 0.6907,
+      "step": 10850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007119159971092123,
+      "loss": 0.7298,
+      "step": 10860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007118309739403988,
+      "loss": 0.6897,
+      "step": 10870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007117459507715853,
+      "loss": 0.6679,
+      "step": 10880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007116609276027717,
+      "loss": 0.719,
+      "step": 10890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007115759044339583,
+      "loss": 0.6699,
+      "step": 10900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007114908812651448,
+      "loss": 0.6414,
+      "step": 10910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007114058580963312,
+      "loss": 0.7017,
+      "step": 10920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007113208349275178,
+      "loss": 0.6731,
+      "step": 10930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007112358117587043,
+      "loss": 0.6499,
+      "step": 10940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007111507885898908,
+      "loss": 0.696,
+      "step": 10950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007110657654210773,
+      "loss": 0.6108,
+      "step": 10960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007109807422522637,
+      "loss": 0.6767,
+      "step": 10970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007108957190834502,
+      "loss": 0.7509,
+      "step": 10980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007108106959146368,
+      "loss": 0.6861,
+      "step": 10990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007107256727458233,
+      "loss": 0.6816,
+      "step": 11000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007106406495770098,
+      "loss": 0.6532,
+      "step": 11010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007105556264081962,
+      "loss": 0.6083,
+      "step": 11020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007104706032393828,
+      "loss": 0.7984,
+      "step": 11030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007103855800705693,
+      "loss": 0.6795,
+      "step": 11040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007103005569017557,
+      "loss": 0.7322,
+      "step": 11050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007102155337329422,
+      "loss": 0.6793,
+      "step": 11060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007101305105641287,
+      "loss": 0.6374,
+      "step": 11070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007100454873953153,
+      "loss": 0.7026,
+      "step": 11080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007099604642265018,
+      "loss": 0.6364,
+      "step": 11090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007098754410576883,
+      "loss": 0.7107,
+      "step": 11100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007097904178888747,
+      "loss": 0.6699,
+      "step": 11110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007097053947200613,
+      "loss": 0.6776,
+      "step": 11120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007096203715512477,
+      "loss": 0.6319,
+      "step": 11130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007095353483824342,
+      "loss": 0.6751,
+      "step": 11140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007094503252136208,
+      "loss": 0.7059,
+      "step": 11150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007093653020448072,
+      "loss": 0.6138,
+      "step": 11160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007092802788759938,
+      "loss": 0.659,
+      "step": 11170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007091952557071803,
+      "loss": 0.6837,
+      "step": 11180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007091102325383667,
+      "loss": 0.6384,
+      "step": 11190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007090252093695532,
+      "loss": 0.6051,
+      "step": 11200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007089401862007397,
+      "loss": 0.6213,
+      "step": 11210
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007088551630319263,
+      "loss": 0.6991,
+      "step": 11220
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007087701398631128,
+      "loss": 0.6314,
+      "step": 11230
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007086851166942992,
+      "loss": 0.6698,
+      "step": 11240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007086000935254857,
+      "loss": 0.6725,
+      "step": 11250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007085150703566723,
+      "loss": 0.6568,
+      "step": 11260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007084300471878587,
+      "loss": 0.7167,
+      "step": 11270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007083450240190452,
+      "loss": 0.643,
+      "step": 11280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007082600008502317,
+      "loss": 0.6747,
+      "step": 11290
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007081749776814182,
+      "loss": 0.7258,
+      "step": 11300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007080899545126048,
+      "loss": 0.6478,
+      "step": 11310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007080049313437912,
+      "loss": 0.7128,
+      "step": 11320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007079199081749777,
+      "loss": 0.5898,
+      "step": 11330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007078348850061642,
+      "loss": 0.581,
+      "step": 11340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007077498618373506,
+      "loss": 0.6505,
+      "step": 11350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007076648386685372,
+      "loss": 0.7108,
+      "step": 11360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007075798154997237,
+      "loss": 0.711,
+      "step": 11370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007074947923309102,
+      "loss": 0.6474,
+      "step": 11380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007074097691620967,
+      "loss": 0.7057,
+      "step": 11390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007073247459932832,
+      "loss": 0.6475,
+      "step": 11400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007072397228244697,
+      "loss": 0.6925,
+      "step": 11410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007071546996556562,
+      "loss": 0.6755,
+      "step": 11420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007070696764868426,
+      "loss": 0.6923,
+      "step": 11430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007069846533180293,
+      "loss": 0.6547,
+      "step": 11440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007068996301492157,
+      "loss": 0.6409,
+      "step": 11450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007068146069804022,
+      "loss": 0.7035,
+      "step": 11460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007067295838115887,
+      "loss": 0.7256,
+      "step": 11470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007066445606427751,
+      "loss": 0.7193,
+      "step": 11480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007065595374739617,
+      "loss": 0.6473,
+      "step": 11490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007064745143051482,
+      "loss": 0.6871,
+      "step": 11500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007063894911363347,
+      "loss": 0.6019,
+      "step": 11510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007063044679675212,
+      "loss": 0.6605,
+      "step": 11520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007062194447987077,
+      "loss": 0.6854,
+      "step": 11530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007061344216298942,
+      "loss": 0.6985,
+      "step": 11540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007060493984610807,
+      "loss": 0.6711,
+      "step": 11550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007059643752922671,
+      "loss": 0.6368,
+      "step": 11560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007058793521234536,
+      "loss": 0.7391,
+      "step": 11570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007057943289546403,
+      "loss": 0.7121,
+      "step": 11580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007057093057858267,
+      "loss": 0.7251,
+      "step": 11590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007056242826170132,
+      "loss": 0.6889,
+      "step": 11600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007055392594481997,
+      "loss": 0.6876,
+      "step": 11610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007054542362793861,
+      "loss": 0.6714,
+      "step": 11620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007053692131105727,
+      "loss": 0.6564,
+      "step": 11630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007052841899417592,
+      "loss": 0.6807,
+      "step": 11640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007051991667729456,
+      "loss": 0.6623,
+      "step": 11650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007051141436041322,
+      "loss": 0.6738,
+      "step": 11660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007050291204353186,
+      "loss": 0.677,
+      "step": 11670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007049440972665052,
+      "loss": 0.7038,
+      "step": 11680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007048590740976917,
+      "loss": 0.6479,
+      "step": 11690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007047740509288781,
+      "loss": 0.7181,
+      "step": 11700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007046890277600646,
+      "loss": 0.7659,
+      "step": 11710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007046040045912512,
+      "loss": 0.6785,
+      "step": 11720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007045189814224377,
+      "loss": 0.6657,
+      "step": 11730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007044339582536242,
+      "loss": 0.6598,
+      "step": 11740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007043489350848106,
+      "loss": 0.6212,
+      "step": 11750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007042639119159971,
+      "loss": 0.6652,
+      "step": 11760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007041788887471837,
+      "loss": 0.6845,
+      "step": 11770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007040938655783701,
+      "loss": 0.5694,
+      "step": 11780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007040088424095566,
+      "loss": 0.6317,
+      "step": 11790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007039238192407431,
+      "loss": 0.7716,
+      "step": 11800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007038387960719297,
+      "loss": 0.6049,
+      "step": 11810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007037537729031162,
+      "loss": 0.6253,
+      "step": 11820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007036687497343026,
+      "loss": 0.6638,
+      "step": 11830
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007035837265654891,
+      "loss": 0.5967,
+      "step": 11840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007034987033966756,
+      "loss": 0.6731,
+      "step": 11850
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007034136802278621,
+      "loss": 0.5814,
+      "step": 11860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007033286570590487,
+      "loss": 0.6264,
+      "step": 11870
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007032436338902351,
+      "loss": 0.6251,
+      "step": 11880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007031586107214216,
+      "loss": 0.5878,
+      "step": 11890
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007030735875526082,
+      "loss": 0.6962,
+      "step": 11900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007029885643837946,
+      "loss": 0.6692,
+      "step": 11910
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007029035412149811,
+      "loss": 0.585,
+      "step": 11920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007028185180461676,
+      "loss": 0.6201,
+      "step": 11930
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000702733494877354,
+      "loss": 0.6635,
+      "step": 11940
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007026484717085407,
+      "loss": 0.6184,
+      "step": 11950
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007025634485397271,
+      "loss": 0.6285,
+      "step": 11960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007024784253709136,
+      "loss": 0.6508,
+      "step": 11970
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007023934022021001,
+      "loss": 0.5851,
+      "step": 11980
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007023083790332865,
+      "loss": 0.6092,
+      "step": 11990
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007022233558644731,
+      "loss": 0.6972,
+      "step": 12000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007021383326956596,
+      "loss": 0.6293,
+      "step": 12010
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007020533095268461,
+      "loss": 0.6988,
+      "step": 12020
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007019682863580326,
+      "loss": 0.7152,
+      "step": 12030
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007018832631892192,
+      "loss": 0.604,
+      "step": 12040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007017982400204056,
+      "loss": 0.5897,
+      "step": 12050
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007017132168515921,
+      "loss": 0.5893,
+      "step": 12060
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007016281936827785,
+      "loss": 0.6436,
+      "step": 12070
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000701543170513965,
+      "loss": 0.609,
+      "step": 12080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007014581473451517,
+      "loss": 0.6767,
+      "step": 12090
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007013731241763381,
+      "loss": 0.6093,
+      "step": 12100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007012881010075246,
+      "loss": 0.6908,
+      "step": 12110
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007012030778387111,
+      "loss": 0.6846,
+      "step": 12120
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007011180546698975,
+      "loss": 0.6878,
+      "step": 12130
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007010330315010841,
+      "loss": 0.6158,
+      "step": 12140
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007009480083322706,
+      "loss": 0.633,
+      "step": 12150
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007008629851634571,
+      "loss": 0.5731,
+      "step": 12160
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007007779619946436,
+      "loss": 0.6381,
+      "step": 12170
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007006929388258301,
+      "loss": 0.6203,
+      "step": 12180
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007006079156570166,
+      "loss": 0.6034,
+      "step": 12190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007005228924882031,
+      "loss": 0.6177,
+      "step": 12200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007004378693193895,
+      "loss": 0.6394,
+      "step": 12210
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000700352846150576,
+      "loss": 0.6598,
+      "step": 12220
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007002678229817626,
+      "loss": 0.627,
+      "step": 12230
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0007001827998129491,
+      "loss": 0.6489,
+      "step": 12240
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0007000977766441356,
+      "loss": 0.6155,
+      "step": 12250
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000700012753475322,
+      "loss": 0.6468,
+      "step": 12260
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006999277303065086,
+      "loss": 0.6219,
+      "step": 12270
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006998427071376951,
+      "loss": 0.5752,
+      "step": 12280
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006997576839688815,
+      "loss": 0.6046,
+      "step": 12290
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000699672660800068,
+      "loss": 0.6161,
+      "step": 12300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006995876376312545,
+      "loss": 0.5981,
+      "step": 12310
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006995026144624411,
+      "loss": 0.6516,
+      "step": 12320
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006994175912936276,
+      "loss": 0.657,
+      "step": 12330
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000699332568124814,
+      "loss": 0.6995,
+      "step": 12340
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0006992475449560005,
+      "loss": 0.6473,
+      "step": 12350
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000699162521787187,
+      "loss": 0.6714,
+      "step": 12360
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006990774986183735,
+      "loss": 0.681,
+      "step": 12370
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006989924754495601,
+      "loss": 0.5957,
+      "step": 12380
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006989074522807465,
+      "loss": 0.6032,
+      "step": 12390
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000698822429111933,
+      "loss": 0.5996,
+      "step": 12400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006987374059431196,
+      "loss": 0.6827,
+      "step": 12410
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000698652382774306,
+      "loss": 0.5956,
+      "step": 12420
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006985673596054925,
+      "loss": 0.6859,
+      "step": 12430
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000698482336436679,
+      "loss": 0.6387,
+      "step": 12440
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006983973132678654,
+      "loss": 0.5967,
+      "step": 12450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006983122900990521,
+      "loss": 0.7185,
+      "step": 12460
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0006982272669302386,
+      "loss": 0.712,
+      "step": 12470
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000698142243761425,
+      "loss": 0.7476,
+      "step": 12480
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006980572205926115,
+      "loss": 0.6387,
+      "step": 12490
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000697972197423798,
+      "loss": 0.6555,
+      "step": 12500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006978871742549845,
+      "loss": 0.6564,
+      "step": 12510
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000697802151086171,
+      "loss": 0.6382,
+      "step": 12520
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006977171279173575,
+      "loss": 0.6427,
+      "step": 12530
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000697632104748544,
+      "loss": 0.6831,
+      "step": 12540
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006975470815797306,
+      "loss": 0.673,
+      "step": 12550
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000697462058410917,
+      "loss": 0.6311,
+      "step": 12560
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006973770352421035,
+      "loss": 0.6341,
+      "step": 12570
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00069729201207329,
+      "loss": 0.707,
+      "step": 12580
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0006972069889044764,
+      "loss": 0.6321,
+      "step": 12590
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006971219657356631,
+      "loss": 0.6604,
+      "step": 12600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006970369425668495,
+      "loss": 0.7235,
+      "step": 12610
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000696951919398036,
+      "loss": 0.6091,
+      "step": 12620
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006968668962292225,
+      "loss": 0.5634,
+      "step": 12630
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000696781873060409,
+      "loss": 0.6656,
+      "step": 12640
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006966968498915955,
+      "loss": 0.6184,
+      "step": 12650
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000696611826722782,
+      "loss": 0.6646,
+      "step": 12660
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006965268035539685,
+      "loss": 0.6005,
+      "step": 12670
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000696441780385155,
+      "loss": 0.6169,
+      "step": 12680
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006963567572163415,
+      "loss": 0.6173,
+      "step": 12690
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000696271734047528,
+      "loss": 0.7332,
+      "step": 12700
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0006961867108787145,
+      "loss": 0.5985,
+      "step": 12710
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006961016877099009,
+      "loss": 0.655,
+      "step": 12720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006960166645410875,
+      "loss": 0.7142,
+      "step": 12730
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000695931641372274,
+      "loss": 0.6524,
+      "step": 12740
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006958466182034605,
+      "loss": 0.6473,
+      "step": 12750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000695761595034647,
+      "loss": 0.6056,
+      "step": 12760
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006956765718658334,
+      "loss": 0.5348,
+      "step": 12770
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00069559154869702,
+      "loss": 0.6487,
+      "step": 12780
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006955065255282065,
+      "loss": 0.6777,
+      "step": 12790
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006954215023593929,
+      "loss": 0.6275,
+      "step": 12800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0006953364791905794,
+      "loss": 0.6359,
+      "step": 12810
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000695251456021766,
+      "loss": 0.6603,
+      "step": 12820
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006951664328529525,
+      "loss": 0.6623,
+      "step": 12830
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000695081409684139,
+      "loss": 0.6298,
+      "step": 12840
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006949963865153254,
+      "loss": 0.7219,
+      "step": 12850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006949113633465119,
+      "loss": 0.6224,
+      "step": 12860
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006948263401776985,
+      "loss": 0.628,
+      "step": 12870
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006947413170088849,
+      "loss": 0.5654,
+      "step": 12880
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006946562938400715,
+      "loss": 0.5656,
+      "step": 12890
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000694571270671258,
+      "loss": 0.6243,
+      "step": 12900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006944862475024444,
+      "loss": 0.6455,
+      "step": 12910
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000694401224333631,
+      "loss": 0.5999,
+      "step": 12920
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006943162011648174,
+      "loss": 0.6184,
+      "step": 12930
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0006942311779960039,
+      "loss": 0.6589,
+      "step": 12940
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006941461548271904,
+      "loss": 0.6364,
+      "step": 12950
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000694061131658377,
+      "loss": 0.6838,
+      "step": 12960
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006939761084895635,
+      "loss": 0.6521,
+      "step": 12970
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00069389108532075,
+      "loss": 0.5987,
+      "step": 12980
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006938060621519364,
+      "loss": 0.6874,
+      "step": 12990
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006937210389831229,
+      "loss": 0.6493,
+      "step": 13000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006936360158143095,
+      "loss": 0.6935,
+      "step": 13010
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006935509926454959,
+      "loss": 0.5875,
+      "step": 13020
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006934659694766824,
+      "loss": 0.6322,
+      "step": 13030
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006933809463078689,
+      "loss": 0.6566,
+      "step": 13040
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006932959231390555,
+      "loss": 0.6257,
+      "step": 13050
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000693210899970242,
+      "loss": 0.6477,
+      "step": 13060
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006931258768014284,
+      "loss": 0.6476,
+      "step": 13070
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006930408536326149,
+      "loss": 0.675,
+      "step": 13080
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006929558304638014,
+      "loss": 0.6283,
+      "step": 13090
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006928708072949879,
+      "loss": 0.6688,
+      "step": 13100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006927857841261745,
+      "loss": 0.5732,
+      "step": 13110
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006927007609573609,
+      "loss": 0.6362,
+      "step": 13120
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006926157377885474,
+      "loss": 0.5925,
+      "step": 13130
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000692530714619734,
+      "loss": 0.6407,
+      "step": 13140
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006924456914509204,
+      "loss": 0.6646,
+      "step": 13150
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006923606682821069,
+      "loss": 0.5876,
+      "step": 13160
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006922756451132934,
+      "loss": 0.6128,
+      "step": 13170
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006921906219444799,
+      "loss": 0.5998,
+      "step": 13180
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006921055987756665,
+      "loss": 0.7011,
+      "step": 13190
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006920205756068529,
+      "loss": 0.6018,
+      "step": 13200
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006919355524380394,
+      "loss": 0.6322,
+      "step": 13210
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006918505292692259,
+      "loss": 0.6697,
+      "step": 13220
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006917655061004123,
+      "loss": 0.6242,
+      "step": 13230
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006916804829315989,
+      "loss": 0.5387,
+      "step": 13240
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006915954597627854,
+      "loss": 0.5849,
+      "step": 13250
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006915104365939719,
+      "loss": 0.7081,
+      "step": 13260
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006914254134251584,
+      "loss": 0.5329,
+      "step": 13270
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006913403902563448,
+      "loss": 0.5903,
+      "step": 13280
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006912553670875314,
+      "loss": 0.6398,
+      "step": 13290
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006911703439187179,
+      "loss": 0.62,
+      "step": 13300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006910853207499043,
+      "loss": 0.6262,
+      "step": 13310
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006910002975810908,
+      "loss": 0.6596,
+      "step": 13320
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006909152744122774,
+      "loss": 0.6929,
+      "step": 13330
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006908302512434639,
+      "loss": 0.5919,
+      "step": 13340
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006907452280746504,
+      "loss": 0.6271,
+      "step": 13350
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006906602049058368,
+      "loss": 0.5902,
+      "step": 13360
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006905751817370233,
+      "loss": 0.6081,
+      "step": 13370
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006904901585682099,
+      "loss": 0.7113,
+      "step": 13380
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006904051353993963,
+      "loss": 0.6352,
+      "step": 13390
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006903201122305829,
+      "loss": 0.633,
+      "step": 13400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006902350890617694,
+      "loss": 0.6739,
+      "step": 13410
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006901500658929559,
+      "loss": 0.6977,
+      "step": 13420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006900650427241424,
+      "loss": 0.6277,
+      "step": 13430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006899800195553288,
+      "loss": 0.6523,
+      "step": 13440
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006898949963865153,
+      "loss": 0.5445,
+      "step": 13450
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006898099732177018,
+      "loss": 0.5686,
+      "step": 13460
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006897249500488884,
+      "loss": 0.6612,
+      "step": 13470
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006896399268800749,
+      "loss": 0.6107,
+      "step": 13480
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006895549037112614,
+      "loss": 0.5696,
+      "step": 13490
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006894698805424478,
+      "loss": 0.6018,
+      "step": 13500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006893848573736344,
+      "loss": 0.6161,
+      "step": 13510
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006892998342048209,
+      "loss": 0.6329,
+      "step": 13520
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006892148110360073,
+      "loss": 0.6701,
+      "step": 13530
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006891297878671938,
+      "loss": 0.6764,
+      "step": 13540
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006890447646983803,
+      "loss": 0.5256,
+      "step": 13550
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006889597415295669,
+      "loss": 0.6482,
+      "step": 13560
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006888747183607534,
+      "loss": 0.641,
+      "step": 13570
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006887896951919398,
+      "loss": 0.6139,
+      "step": 13580
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006887046720231263,
+      "loss": 0.5854,
+      "step": 13590
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006886196488543129,
+      "loss": 0.5807,
+      "step": 13600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006885346256854993,
+      "loss": 0.7449,
+      "step": 13610
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006884496025166859,
+      "loss": 0.6385,
+      "step": 13620
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006883645793478723,
+      "loss": 0.6494,
+      "step": 13630
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006882795561790588,
+      "loss": 0.6713,
+      "step": 13640
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006881945330102454,
+      "loss": 0.5874,
+      "step": 13650
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006881095098414318,
+      "loss": 0.6903,
+      "step": 13660
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006880244866726183,
+      "loss": 0.6359,
+      "step": 13670
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006879394635038048,
+      "loss": 0.6827,
+      "step": 13680
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006878544403349913,
+      "loss": 0.6318,
+      "step": 13690
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006877694171661779,
+      "loss": 0.6518,
+      "step": 13700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006876843939973643,
+      "loss": 0.6295,
+      "step": 13710
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006875993708285508,
+      "loss": 0.6713,
+      "step": 13720
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006875143476597373,
+      "loss": 0.6288,
+      "step": 13730
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006874293244909237,
+      "loss": 0.6914,
+      "step": 13740
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006873443013221103,
+      "loss": 0.5858,
+      "step": 13750
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006872592781532968,
+      "loss": 0.6604,
+      "step": 13760
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006871742549844833,
+      "loss": 0.6284,
+      "step": 13770
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006870892318156698,
+      "loss": 0.7019,
+      "step": 13780
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006870042086468563,
+      "loss": 0.6344,
+      "step": 13790
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006869191854780428,
+      "loss": 0.5867,
+      "step": 13800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006868341623092293,
+      "loss": 0.6323,
+      "step": 13810
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006867491391404157,
+      "loss": 0.6093,
+      "step": 13820
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006866641159716022,
+      "loss": 0.6589,
+      "step": 13830
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006865790928027889,
+      "loss": 0.6672,
+      "step": 13840
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006864940696339753,
+      "loss": 0.5932,
+      "step": 13850
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006864090464651618,
+      "loss": 0.6528,
+      "step": 13860
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006863240232963482,
+      "loss": 0.6138,
+      "step": 13870
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006862390001275348,
+      "loss": 0.6784,
+      "step": 13880
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006861539769587213,
+      "loss": 0.6407,
+      "step": 13890
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006860689537899077,
+      "loss": 0.7027,
+      "step": 13900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006859839306210943,
+      "loss": 0.632,
+      "step": 13910
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006858989074522809,
+      "loss": 0.6775,
+      "step": 13920
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006858138842834673,
+      "loss": 0.6245,
+      "step": 13930
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006857288611146538,
+      "loss": 0.7834,
+      "step": 13940
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006856438379458403,
+      "loss": 0.5945,
+      "step": 13950
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006855588147770267,
+      "loss": 0.6108,
+      "step": 13960
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006854737916082133,
+      "loss": 0.6246,
+      "step": 13970
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006853887684393998,
+      "loss": 0.5782,
+      "step": 13980
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006853037452705863,
+      "loss": 0.6127,
+      "step": 13990
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006852187221017728,
+      "loss": 0.6479,
+      "step": 14000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006851336989329592,
+      "loss": 0.6581,
+      "step": 14010
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006850486757641458,
+      "loss": 0.6579,
+      "step": 14020
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006849636525953323,
+      "loss": 0.5804,
+      "step": 14030
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006848786294265187,
+      "loss": 0.6062,
+      "step": 14040
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006847936062577053,
+      "loss": 0.6306,
+      "step": 14050
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006847085830888917,
+      "loss": 0.5839,
+      "step": 14060
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006846235599200783,
+      "loss": 0.6345,
+      "step": 14070
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006845385367512648,
+      "loss": 0.6519,
+      "step": 14080
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006844535135824512,
+      "loss": 0.5878,
+      "step": 14090
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006843684904136377,
+      "loss": 0.5898,
+      "step": 14100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006842834672448243,
+      "loss": 0.6915,
+      "step": 14110
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006841984440760107,
+      "loss": 0.5847,
+      "step": 14120
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006841134209071973,
+      "loss": 0.5876,
+      "step": 14130
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006840283977383837,
+      "loss": 0.6717,
+      "step": 14140
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006839433745695702,
+      "loss": 0.6529,
+      "step": 14150
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006838583514007568,
+      "loss": 0.5887,
+      "step": 14160
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006837733282319432,
+      "loss": 0.5749,
+      "step": 14170
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006836883050631297,
+      "loss": 0.651,
+      "step": 14180
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006836032818943162,
+      "loss": 0.599,
+      "step": 14190
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006835182587255028,
+      "loss": 0.6103,
+      "step": 14200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006834332355566893,
+      "loss": 0.6652,
+      "step": 14210
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006833482123878757,
+      "loss": 0.5999,
+      "step": 14220
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006832631892190622,
+      "loss": 0.6093,
+      "step": 14230
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006831781660502487,
+      "loss": 0.5878,
+      "step": 14240
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006830931428814352,
+      "loss": 0.6195,
+      "step": 14250
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006830081197126217,
+      "loss": 0.6273,
+      "step": 14260
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006829230965438083,
+      "loss": 0.6674,
+      "step": 14270
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006828380733749947,
+      "loss": 0.6507,
+      "step": 14280
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006827530502061813,
+      "loss": 0.6357,
+      "step": 14290
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006826680270373677,
+      "loss": 0.5782,
+      "step": 14300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006825830038685542,
+      "loss": 0.6286,
+      "step": 14310
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006824979806997407,
+      "loss": 0.7234,
+      "step": 14320
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006824129575309271,
+      "loss": 0.6044,
+      "step": 14330
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006823279343621138,
+      "loss": 0.5853,
+      "step": 14340
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006822429111933003,
+      "loss": 0.6159,
+      "step": 14350
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006821578880244867,
+      "loss": 0.579,
+      "step": 14360
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006820728648556732,
+      "loss": 0.6938,
+      "step": 14370
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006819878416868598,
+      "loss": 0.6696,
+      "step": 14380
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006819028185180462,
+      "loss": 0.6987,
+      "step": 14390
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006818177953492327,
+      "loss": 0.5635,
+      "step": 14400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006817327721804191,
+      "loss": 0.6307,
+      "step": 14410
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006816477490116057,
+      "loss": 0.5884,
+      "step": 14420
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006815627258427923,
+      "loss": 0.6521,
+      "step": 14430
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006814777026739787,
+      "loss": 0.6551,
+      "step": 14440
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006813926795051652,
+      "loss": 0.6499,
+      "step": 14450
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006813076563363517,
+      "loss": 0.6057,
+      "step": 14460
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006812226331675381,
+      "loss": 0.6299,
+      "step": 14470
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006811376099987247,
+      "loss": 0.6775,
+      "step": 14480
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006810525868299112,
+      "loss": 0.7163,
+      "step": 14490
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006809675636610977,
+      "loss": 0.5479,
+      "step": 14500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006808825404922842,
+      "loss": 0.6751,
+      "step": 14510
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006807975173234706,
+      "loss": 0.6114,
+      "step": 14520
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006807124941546572,
+      "loss": 0.7029,
+      "step": 14530
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006806274709858437,
+      "loss": 0.7293,
+      "step": 14540
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006805424478170301,
+      "loss": 0.6532,
+      "step": 14550
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006804574246482167,
+      "loss": 0.683,
+      "step": 14560
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006803724014794032,
+      "loss": 0.5697,
+      "step": 14570
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006802873783105897,
+      "loss": 0.6907,
+      "step": 14580
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006802023551417762,
+      "loss": 0.6121,
+      "step": 14590
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006801173319729626,
+      "loss": 0.6569,
+      "step": 14600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006800323088041491,
+      "loss": 0.5861,
+      "step": 14610
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006799472856353357,
+      "loss": 0.6926,
+      "step": 14620
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006798622624665222,
+      "loss": 0.6982,
+      "step": 14630
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006797772392977087,
+      "loss": 0.7099,
+      "step": 14640
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006796922161288951,
+      "loss": 0.6486,
+      "step": 14650
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006796071929600817,
+      "loss": 0.5724,
+      "step": 14660
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006795221697912682,
+      "loss": 0.6721,
+      "step": 14670
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006794371466224546,
+      "loss": 0.6571,
+      "step": 14680
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006793521234536411,
+      "loss": 0.6377,
+      "step": 14690
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006792671002848276,
+      "loss": 0.5749,
+      "step": 14700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006791820771160142,
+      "loss": 0.6621,
+      "step": 14710
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006790970539472007,
+      "loss": 0.677,
+      "step": 14720
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006790120307783871,
+      "loss": 0.7123,
+      "step": 14730
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006789270076095736,
+      "loss": 0.6547,
+      "step": 14740
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006788419844407602,
+      "loss": 0.6293,
+      "step": 14750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006787569612719466,
+      "loss": 0.6967,
+      "step": 14760
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006786719381031331,
+      "loss": 0.6419,
+      "step": 14770
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006785869149343197,
+      "loss": 0.645,
+      "step": 14780
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006785018917655061,
+      "loss": 0.5573,
+      "step": 14790
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006784168685966927,
+      "loss": 0.6093,
+      "step": 14800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006783318454278791,
+      "loss": 0.6241,
+      "step": 14810
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006782468222590656,
+      "loss": 0.5859,
+      "step": 14820
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006781617990902521,
+      "loss": 0.7315,
+      "step": 14830
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006780767759214385,
+      "loss": 0.6019,
+      "step": 14840
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006779917527526252,
+      "loss": 0.5927,
+      "step": 14850
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006779067295838117,
+      "loss": 0.6407,
+      "step": 14860
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006778217064149981,
+      "loss": 0.6998,
+      "step": 14870
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006777366832461846,
+      "loss": 0.6094,
+      "step": 14880
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006776516600773712,
+      "loss": 0.5675,
+      "step": 14890
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006775666369085576,
+      "loss": 0.6037,
+      "step": 14900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006774816137397441,
+      "loss": 0.6494,
+      "step": 14910
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006773965905709305,
+      "loss": 0.6265,
+      "step": 14920
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006773115674021171,
+      "loss": 0.6164,
+      "step": 14930
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006772265442333037,
+      "loss": 0.5876,
+      "step": 14940
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006771415210644901,
+      "loss": 0.6721,
+      "step": 14950
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006770564978956766,
+      "loss": 0.6544,
+      "step": 14960
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006769714747268631,
+      "loss": 0.6176,
+      "step": 14970
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006768864515580495,
+      "loss": 0.6246,
+      "step": 14980
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006768014283892361,
+      "loss": 0.5439,
+      "step": 14990
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006767164052204226,
+      "loss": 0.656,
+      "step": 15000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006766313820516091,
+      "loss": 0.6065,
+      "step": 15010
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006765463588827956,
+      "loss": 0.612,
+      "step": 15020
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006764613357139821,
+      "loss": 0.5881,
+      "step": 15030
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006763763125451686,
+      "loss": 0.6167,
+      "step": 15040
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006762912893763551,
+      "loss": 0.58,
+      "step": 15050
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006762062662075415,
+      "loss": 0.6652,
+      "step": 15060
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006761212430387282,
+      "loss": 0.6623,
+      "step": 15070
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006760362198699146,
+      "loss": 0.6561,
+      "step": 15080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006759511967011011,
+      "loss": 0.6165,
+      "step": 15090
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006758661735322876,
+      "loss": 0.6087,
+      "step": 15100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000675781150363474,
+      "loss": 0.622,
+      "step": 15110
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006756961271946606,
+      "loss": 0.6743,
+      "step": 15120
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006756111040258471,
+      "loss": 0.7102,
+      "step": 15130
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006755260808570336,
+      "loss": 0.6243,
+      "step": 15140
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006754410576882201,
+      "loss": 0.6421,
+      "step": 15150
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006753560345194065,
+      "loss": 0.5543,
+      "step": 15160
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006752710113505931,
+      "loss": 0.5803,
+      "step": 15170
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006751859881817796,
+      "loss": 0.63,
+      "step": 15180
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000675100965012966,
+      "loss": 0.6631,
+      "step": 15190
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006750159418441525,
+      "loss": 0.6722,
+      "step": 15200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000674930918675339,
+      "loss": 0.6335,
+      "step": 15210
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006748458955065256,
+      "loss": 0.6304,
+      "step": 15220
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006747608723377121,
+      "loss": 0.6244,
+      "step": 15230
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006746758491688985,
+      "loss": 0.6031,
+      "step": 15240
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000674590826000085,
+      "loss": 0.6185,
+      "step": 15250
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006745058028312716,
+      "loss": 0.5495,
+      "step": 15260
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000674420779662458,
+      "loss": 0.6747,
+      "step": 15270
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006743357564936445,
+      "loss": 0.5853,
+      "step": 15280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006742507333248311,
+      "loss": 0.6584,
+      "step": 15290
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006741657101560175,
+      "loss": 0.6062,
+      "step": 15300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006740806869872041,
+      "loss": 0.6699,
+      "step": 15310
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006739956638183906,
+      "loss": 0.6161,
+      "step": 15320
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000673910640649577,
+      "loss": 0.5842,
+      "step": 15330
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006738256174807635,
+      "loss": 0.6493,
+      "step": 15340
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00067374059431195,
+      "loss": 0.5659,
+      "step": 15350
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006736555711431366,
+      "loss": 0.7078,
+      "step": 15360
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006735705479743231,
+      "loss": 0.6239,
+      "step": 15370
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006734855248055095,
+      "loss": 0.6917,
+      "step": 15380
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000673400501636696,
+      "loss": 0.673,
+      "step": 15390
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006733154784678826,
+      "loss": 0.6343,
+      "step": 15400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000673230455299069,
+      "loss": 0.579,
+      "step": 15410
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006731454321302555,
+      "loss": 0.5902,
+      "step": 15420
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000673060408961442,
+      "loss": 0.6107,
+      "step": 15430
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006729753857926286,
+      "loss": 0.6977,
+      "step": 15440
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006728903626238151,
+      "loss": 0.6146,
+      "step": 15450
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006728053394550015,
+      "loss": 0.6554,
+      "step": 15460
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000672720316286188,
+      "loss": 0.5924,
+      "step": 15470
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006726352931173745,
+      "loss": 0.6189,
+      "step": 15480
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000672550269948561,
+      "loss": 0.6584,
+      "step": 15490
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006724652467797475,
+      "loss": 0.6226,
+      "step": 15500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000672380223610934,
+      "loss": 0.6341,
+      "step": 15510
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006722952004421205,
+      "loss": 0.6491,
+      "step": 15520
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000672210177273307,
+      "loss": 0.6624,
+      "step": 15530
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006721251541044935,
+      "loss": 0.6754,
+      "step": 15540
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00067204013093568,
+      "loss": 0.6486,
+      "step": 15550
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006719551077668665,
+      "loss": 0.6246,
+      "step": 15560
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006718700845980529,
+      "loss": 0.6298,
+      "step": 15570
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006717850614292396,
+      "loss": 0.6331,
+      "step": 15580
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000671700038260426,
+      "loss": 0.6251,
+      "step": 15590
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006716150150916125,
+      "loss": 0.6326,
+      "step": 15600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000671529991922799,
+      "loss": 0.6267,
+      "step": 15610
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006714449687539854,
+      "loss": 0.6299,
+      "step": 15620
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000671359945585172,
+      "loss": 0.6421,
+      "step": 15630
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006712749224163585,
+      "loss": 0.7019,
+      "step": 15640
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000671189899247545,
+      "loss": 0.5954,
+      "step": 15650
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006711048760787315,
+      "loss": 0.5758,
+      "step": 15660
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000671019852909918,
+      "loss": 0.6725,
+      "step": 15670
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006709348297411045,
+      "loss": 0.6538,
+      "step": 15680
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000670849806572291,
+      "loss": 0.6582,
+      "step": 15690
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006707647834034774,
+      "loss": 0.6477,
+      "step": 15700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006706797602346639,
+      "loss": 0.6205,
+      "step": 15710
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006705947370658506,
+      "loss": 0.5861,
+      "step": 15720
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000670509713897037,
+      "loss": 0.5974,
+      "step": 15730
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006704246907282235,
+      "loss": 0.6358,
+      "step": 15740
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006703396675594099,
+      "loss": 0.6575,
+      "step": 15750
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006702546443905964,
+      "loss": 0.629,
+      "step": 15760
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000670169621221783,
+      "loss": 0.6586,
+      "step": 15770
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006700845980529694,
+      "loss": 0.6364,
+      "step": 15780
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006699995748841559,
+      "loss": 0.6033,
+      "step": 15790
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006699145517153425,
+      "loss": 0.5734,
+      "step": 15800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000669829528546529,
+      "loss": 0.666,
+      "step": 15810
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006697445053777155,
+      "loss": 0.692,
+      "step": 15820
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000669659482208902,
+      "loss": 0.6384,
+      "step": 15830
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006695744590400884,
+      "loss": 0.6153,
+      "step": 15840
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006694894358712749,
+      "loss": 0.6735,
+      "step": 15850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006694044127024614,
+      "loss": 0.6756,
+      "step": 15860
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000669319389533648,
+      "loss": 0.5997,
+      "step": 15870
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006692343663648345,
+      "loss": 0.6285,
+      "step": 15880
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006691493431960209,
+      "loss": 0.6316,
+      "step": 15890
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006690643200272075,
+      "loss": 0.6357,
+      "step": 15900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000668979296858394,
+      "loss": 0.5711,
+      "step": 15910
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006688942736895804,
+      "loss": 0.668,
+      "step": 15920
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006688092505207669,
+      "loss": 0.7186,
+      "step": 15930
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006687242273519534,
+      "loss": 0.6537,
+      "step": 15940
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00066863920418314,
+      "loss": 0.6476,
+      "step": 15950
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006685541810143265,
+      "loss": 0.5556,
+      "step": 15960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006684691578455129,
+      "loss": 0.6413,
+      "step": 15970
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006683841346766994,
+      "loss": 0.6193,
+      "step": 15980
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000668299111507886,
+      "loss": 0.6969,
+      "step": 15990
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006682140883390724,
+      "loss": 0.5854,
+      "step": 16000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006681290651702589,
+      "loss": 0.675,
+      "step": 16010
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006680440420014454,
+      "loss": 0.6281,
+      "step": 16020
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006679590188326319,
+      "loss": 0.6256,
+      "step": 16030
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006678739956638185,
+      "loss": 0.6486,
+      "step": 16040
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006677889724950049,
+      "loss": 0.6507,
+      "step": 16050
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006677039493261914,
+      "loss": 0.7338,
+      "step": 16060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006676189261573779,
+      "loss": 0.633,
+      "step": 16070
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006675339029885643,
+      "loss": 0.6741,
+      "step": 16080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.000667448879819751,
+      "loss": 0.6197,
+      "step": 16090
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006673638566509374,
+      "loss": 0.6905,
+      "step": 16100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006672788334821239,
+      "loss": 0.6674,
+      "step": 16110
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006671938103133104,
+      "loss": 0.6866,
+      "step": 16120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006671087871444969,
+      "loss": 0.5837,
+      "step": 16130
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006670237639756834,
+      "loss": 0.6274,
+      "step": 16140
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006669387408068699,
+      "loss": 0.6031,
+      "step": 16150
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006668537176380564,
+      "loss": 0.6939,
+      "step": 16160
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000666768694469243,
+      "loss": 0.5627,
+      "step": 16170
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006666836713004294,
+      "loss": 0.6351,
+      "step": 16180
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006665986481316159,
+      "loss": 0.6406,
+      "step": 16190
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006665136249628024,
+      "loss": 0.6297,
+      "step": 16200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006664286017939888,
+      "loss": 0.6116,
+      "step": 16210
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006663435786251753,
+      "loss": 0.7192,
+      "step": 16220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000666258555456362,
+      "loss": 0.6464,
+      "step": 16230
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006661735322875484,
+      "loss": 0.7399,
+      "step": 16240
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006660885091187349,
+      "loss": 0.6628,
+      "step": 16250
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006660034859499214,
+      "loss": 0.6122,
+      "step": 16260
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006659184627811079,
+      "loss": 0.6485,
+      "step": 16270
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006658334396122944,
+      "loss": 0.6322,
+      "step": 16280
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006657484164434808,
+      "loss": 0.6453,
+      "step": 16290
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006656633932746673,
+      "loss": 0.6106,
+      "step": 16300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000665578370105854,
+      "loss": 0.6615,
+      "step": 16310
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006654933469370404,
+      "loss": 0.5757,
+      "step": 16320
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006654083237682269,
+      "loss": 0.6702,
+      "step": 16330
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006653233005994134,
+      "loss": 0.7033,
+      "step": 16340
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006652382774305998,
+      "loss": 0.6715,
+      "step": 16350
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006651532542617864,
+      "loss": 0.6603,
+      "step": 16360
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0006650682310929729,
+      "loss": 0.5759,
+      "step": 16370
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006649832079241594,
+      "loss": 0.6158,
+      "step": 16380
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006648981847553459,
+      "loss": 0.5996,
+      "step": 16390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006648131615865323,
+      "loss": 0.6527,
+      "step": 16400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006647281384177189,
+      "loss": 0.5682,
+      "step": 16410
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006646431152489054,
+      "loss": 0.6511,
+      "step": 16420
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006645580920800918,
+      "loss": 0.6751,
+      "step": 16430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006644730689112783,
+      "loss": 0.647,
+      "step": 16440
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006643880457424649,
+      "loss": 0.6398,
+      "step": 16450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006643030225736514,
+      "loss": 0.6304,
+      "step": 16460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006642179994048379,
+      "loss": 0.658,
+      "step": 16470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006641329762360243,
+      "loss": 0.6597,
+      "step": 16480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0006640479530672108,
+      "loss": 0.5897,
+      "step": 16490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006639629298983974,
+      "loss": 0.5612,
+      "step": 16500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006638779067295838,
+      "loss": 0.6192,
+      "step": 16510
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006637928835607704,
+      "loss": 0.693,
+      "step": 16520
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006637078603919568,
+      "loss": 0.6747,
+      "step": 16530
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006636228372231433,
+      "loss": 0.6147,
+      "step": 16540
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006635378140543299,
+      "loss": 0.6639,
+      "step": 16550
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006634527908855163,
+      "loss": 0.6348,
+      "step": 16560
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006633677677167028,
+      "loss": 0.5477,
+      "step": 16570
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006632827445478893,
+      "loss": 0.5359,
+      "step": 16580
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006631977213790758,
+      "loss": 0.5852,
+      "step": 16590
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006631126982102624,
+      "loss": 0.692,
+      "step": 16600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0006630276750414488,
+      "loss": 0.6528,
+      "step": 16610
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006629426518726353,
+      "loss": 0.5982,
+      "step": 16620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006628576287038218,
+      "loss": 0.6624,
+      "step": 16630
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006627726055350083,
+      "loss": 0.6382,
+      "step": 16640
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006626875823661948,
+      "loss": 0.6606,
+      "step": 16650
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006626025591973813,
+      "loss": 0.6058,
+      "step": 16660
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006625175360285678,
+      "loss": 0.6803,
+      "step": 16670
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006624325128597544,
+      "loss": 0.6186,
+      "step": 16680
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006623474896909409,
+      "loss": 0.5997,
+      "step": 16690
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006622624665221273,
+      "loss": 0.5622,
+      "step": 16700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006621774433533138,
+      "loss": 0.6391,
+      "step": 16710
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006620924201845002,
+      "loss": 0.6425,
+      "step": 16720
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0006620073970156868,
+      "loss": 0.6406,
+      "step": 16730
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006619223738468734,
+      "loss": 0.6142,
+      "step": 16740
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006618373506780598,
+      "loss": 0.6417,
+      "step": 16750
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006617523275092463,
+      "loss": 0.5519,
+      "step": 16760
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006616673043404329,
+      "loss": 0.6402,
+      "step": 16770
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006615822811716193,
+      "loss": 0.6714,
+      "step": 16780
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006614972580028058,
+      "loss": 0.7512,
+      "step": 16790
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006614122348339923,
+      "loss": 0.5777,
+      "step": 16800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006613272116651788,
+      "loss": 0.5727,
+      "step": 16810
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006612421884963654,
+      "loss": 0.684,
+      "step": 16820
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006611571653275518,
+      "loss": 0.5889,
+      "step": 16830
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0006610721421587383,
+      "loss": 0.6746,
+      "step": 16840
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006609871189899248,
+      "loss": 0.6085,
+      "step": 16850
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006609020958211112,
+      "loss": 0.5712,
+      "step": 16860
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006608170726522978,
+      "loss": 0.6081,
+      "step": 16870
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006607320494834843,
+      "loss": 0.6436,
+      "step": 16880
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006606470263146708,
+      "loss": 0.59,
+      "step": 16890
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006605620031458573,
+      "loss": 0.6643,
+      "step": 16900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006604769799770438,
+      "loss": 0.6393,
+      "step": 16910
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006603919568082303,
+      "loss": 0.5838,
+      "step": 16920
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006603069336394168,
+      "loss": 0.5946,
+      "step": 16930
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006602219104706032,
+      "loss": 0.6474,
+      "step": 16940
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006601368873017897,
+      "loss": 0.6423,
+      "step": 16950
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0006600518641329763,
+      "loss": 0.6414,
+      "step": 16960
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006599668409641628,
+      "loss": 0.6181,
+      "step": 16970
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006598818177953493,
+      "loss": 0.6748,
+      "step": 16980
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006597967946265357,
+      "loss": 0.5736,
+      "step": 16990
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006597117714577222,
+      "loss": 0.6547,
+      "step": 17000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006596267482889088,
+      "loss": 0.6175,
+      "step": 17010
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006595417251200952,
+      "loss": 0.6103,
+      "step": 17020
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006594567019512818,
+      "loss": 0.6873,
+      "step": 17030
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006593716787824682,
+      "loss": 0.6239,
+      "step": 17040
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006592866556136548,
+      "loss": 0.6108,
+      "step": 17050
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006592016324448413,
+      "loss": 0.6037,
+      "step": 17060
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006591166092760277,
+      "loss": 0.6241,
+      "step": 17070
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0006590315861072142,
+      "loss": 0.6263,
+      "step": 17080
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006589465629384007,
+      "loss": 0.5496,
+      "step": 17090
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006588615397695873,
+      "loss": 0.6508,
+      "step": 17100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006587765166007738,
+      "loss": 0.6102,
+      "step": 17110
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006586914934319602,
+      "loss": 0.5461,
+      "step": 17120
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006586064702631467,
+      "loss": 0.598,
+      "step": 17130
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006585214470943333,
+      "loss": 0.6016,
+      "step": 17140
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006584364239255197,
+      "loss": 0.667,
+      "step": 17150
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006583514007567062,
+      "loss": 0.5713,
+      "step": 17160
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006582663775878927,
+      "loss": 0.6233,
+      "step": 17170
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006581813544190792,
+      "loss": 0.6442,
+      "step": 17180
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006580963312502658,
+      "loss": 0.6071,
+      "step": 17190
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0006580113080814523,
+      "loss": 0.6158,
+      "step": 17200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006579262849126387,
+      "loss": 0.6963,
+      "step": 17210
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006578412617438252,
+      "loss": 0.6824,
+      "step": 17220
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006577562385750116,
+      "loss": 0.6287,
+      "step": 17230
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006576712154061982,
+      "loss": 0.6541,
+      "step": 17240
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006575861922373848,
+      "loss": 0.5445,
+      "step": 17250
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006575011690685712,
+      "loss": 0.645,
+      "step": 17260
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006574161458997577,
+      "loss": 0.6251,
+      "step": 17270
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006573311227309443,
+      "loss": 0.6093,
+      "step": 17280
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006572460995621307,
+      "loss": 0.6909,
+      "step": 17290
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006571610763933172,
+      "loss": 0.6464,
+      "step": 17300
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006570760532245037,
+      "loss": 0.6008,
+      "step": 17310
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0006569910300556902,
+      "loss": 0.6332,
+      "step": 17320
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006569060068868768,
+      "loss": 0.6969,
+      "step": 17330
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006568209837180632,
+      "loss": 0.5589,
+      "step": 17340
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006567359605492497,
+      "loss": 0.572,
+      "step": 17350
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006566509373804362,
+      "loss": 0.5922,
+      "step": 17360
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006565659142116227,
+      "loss": 0.557,
+      "step": 17370
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006564808910428092,
+      "loss": 0.6815,
+      "step": 17380
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006563958678739957,
+      "loss": 0.6904,
+      "step": 17390
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006563108447051822,
+      "loss": 0.6555,
+      "step": 17400
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006562258215363687,
+      "loss": 0.6086,
+      "step": 17410
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006561407983675552,
+      "loss": 0.5629,
+      "step": 17420
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006560557751987417,
+      "loss": 0.6203,
+      "step": 17430
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0006559707520299282,
+      "loss": 0.6193,
+      "step": 17440
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006558857288611146,
+      "loss": 0.6908,
+      "step": 17450
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006558007056923011,
+      "loss": 0.675,
+      "step": 17460
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006557156825234877,
+      "loss": 0.5937,
+      "step": 17470
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006556306593546742,
+      "loss": 0.6046,
+      "step": 17480
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006555456361858607,
+      "loss": 0.6827,
+      "step": 17490
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006554606130170471,
+      "loss": 0.6025,
+      "step": 17500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006553755898482337,
+      "loss": 0.7122,
+      "step": 17510
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006552905666794202,
+      "loss": 0.6124,
+      "step": 17520
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006552055435106066,
+      "loss": 0.5959,
+      "step": 17530
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006551205203417932,
+      "loss": 0.6287,
+      "step": 17540
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0006550354971729796,
+      "loss": 0.6117,
+      "step": 17550
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006549504740041662,
+      "loss": 0.6519,
+      "step": 17560
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006548654508353527,
+      "loss": 0.611,
+      "step": 17570
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006547804276665391,
+      "loss": 0.6371,
+      "step": 17580
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006546954044977256,
+      "loss": 0.6325,
+      "step": 17590
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006546103813289122,
+      "loss": 0.6307,
+      "step": 17600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006545253581600987,
+      "loss": 0.5223,
+      "step": 17610
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006544403349912852,
+      "loss": 0.6413,
+      "step": 17620
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006543553118224717,
+      "loss": 0.6353,
+      "step": 17630
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006542702886536581,
+      "loss": 0.6813,
+      "step": 17640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006541852654848447,
+      "loss": 0.6978,
+      "step": 17650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006541002423160311,
+      "loss": 0.5529,
+      "step": 17660
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0006540152191472176,
+      "loss": 0.6946,
+      "step": 17670
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006539301959784041,
+      "loss": 0.5963,
+      "step": 17680
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006538451728095907,
+      "loss": 0.6161,
+      "step": 17690
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006537601496407772,
+      "loss": 0.6446,
+      "step": 17700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006536751264719637,
+      "loss": 0.7243,
+      "step": 17710
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006535901033031501,
+      "loss": 0.5344,
+      "step": 17720
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006535050801343366,
+      "loss": 0.6534,
+      "step": 17730
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006534200569655232,
+      "loss": 0.6022,
+      "step": 17740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006533350337967096,
+      "loss": 0.5981,
+      "step": 17750
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006532500106278962,
+      "loss": 0.626,
+      "step": 17760
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006531649874590826,
+      "loss": 0.6177,
+      "step": 17770
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006530799642902691,
+      "loss": 0.6243,
+      "step": 17780
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0006529949411214557,
+      "loss": 0.6385,
+      "step": 17790
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006529099179526421,
+      "loss": 0.6484,
+      "step": 17800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006528248947838286,
+      "loss": 0.6293,
+      "step": 17810
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006527398716150151,
+      "loss": 0.6375,
+      "step": 17820
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006526548484462017,
+      "loss": 0.652,
+      "step": 17830
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006525698252773882,
+      "loss": 0.6572,
+      "step": 17840
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006524848021085746,
+      "loss": 0.6541,
+      "step": 17850
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006523997789397611,
+      "loss": 0.7286,
+      "step": 17860
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006523147557709476,
+      "loss": 0.5719,
+      "step": 17870
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006522297326021341,
+      "loss": 0.6259,
+      "step": 17880
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006521447094333206,
+      "loss": 0.5662,
+      "step": 17890
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006520596862645071,
+      "loss": 0.5958,
+      "step": 17900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0006519746630956936,
+      "loss": 0.6353,
+      "step": 17910
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006518896399268802,
+      "loss": 0.6545,
+      "step": 17920
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006518046167580666,
+      "loss": 0.6769,
+      "step": 17930
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006517195935892531,
+      "loss": 0.6654,
+      "step": 17940
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006516345704204396,
+      "loss": 0.5177,
+      "step": 17950
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.000651549547251626,
+      "loss": 0.6445,
+      "step": 17960
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006514645240828126,
+      "loss": 0.6796,
+      "step": 17970
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006513795009139991,
+      "loss": 0.5916,
+      "step": 17980
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006512944777451856,
+      "loss": 0.6102,
+      "step": 17990
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006512094545763721,
+      "loss": 0.6325,
+      "step": 18000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006511244314075585,
+      "loss": 0.586,
+      "step": 18010
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006510394082387451,
+      "loss": 0.6034,
+      "step": 18020
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0006509543850699316,
+      "loss": 0.7447,
+      "step": 18030
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000650869361901118,
+      "loss": 0.6502,
+      "step": 18040
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006507843387323046,
+      "loss": 0.6522,
+      "step": 18050
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000650699315563491,
+      "loss": 0.6175,
+      "step": 18060
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006506142923946776,
+      "loss": 0.6449,
+      "step": 18070
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006505292692258641,
+      "loss": 0.6083,
+      "step": 18080
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006504442460570505,
+      "loss": 0.6523,
+      "step": 18090
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000650359222888237,
+      "loss": 0.6528,
+      "step": 18100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006502741997194236,
+      "loss": 0.6906,
+      "step": 18110
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006501891765506101,
+      "loss": 0.5754,
+      "step": 18120
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006501041533817966,
+      "loss": 0.5772,
+      "step": 18130
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0006500191302129831,
+      "loss": 0.6141,
+      "step": 18140
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006499341070441696,
+      "loss": 0.6624,
+      "step": 18150
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006498490838753561,
+      "loss": 0.6062,
+      "step": 18160
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006497640607065425,
+      "loss": 0.6193,
+      "step": 18170
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.000649679037537729,
+      "loss": 0.6948,
+      "step": 18180
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006495940143689156,
+      "loss": 0.6087,
+      "step": 18190
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006495089912001021,
+      "loss": 0.6423,
+      "step": 18200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006494239680312886,
+      "loss": 0.5489,
+      "step": 18210
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006493389448624751,
+      "loss": 0.5991,
+      "step": 18220
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006492539216936615,
+      "loss": 0.6705,
+      "step": 18230
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.000649168898524848,
+      "loss": 0.6152,
+      "step": 18240
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0006490838753560346,
+      "loss": 0.6382,
+      "step": 18250
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.000648998852187221,
+      "loss": 0.617,
+      "step": 18260
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006489138290184076,
+      "loss": 0.6064,
+      "step": 18270
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000648828805849594,
+      "loss": 0.6289,
+      "step": 18280
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006487437826807806,
+      "loss": 0.639,
+      "step": 18290
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006486587595119671,
+      "loss": 0.7014,
+      "step": 18300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006485737363431535,
+      "loss": 0.5693,
+      "step": 18310
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00064848871317434,
+      "loss": 0.6621,
+      "step": 18320
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006484036900055265,
+      "loss": 0.7052,
+      "step": 18330
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006483186668367131,
+      "loss": 0.646,
+      "step": 18340
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006482336436678996,
+      "loss": 0.6876,
+      "step": 18350
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000648148620499086,
+      "loss": 0.5869,
+      "step": 18360
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006480635973302725,
+      "loss": 0.6565,
+      "step": 18370
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0006479785741614591,
+      "loss": 0.5837,
+      "step": 18380
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006478935509926455,
+      "loss": 0.5929,
+      "step": 18390
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000647808527823832,
+      "loss": 0.6511,
+      "step": 18400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006477235046550185,
+      "loss": 0.6886,
+      "step": 18410
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000647638481486205,
+      "loss": 0.6368,
+      "step": 18420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006475534583173916,
+      "loss": 0.5958,
+      "step": 18430
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000647468435148578,
+      "loss": 0.5972,
+      "step": 18440
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006473834119797645,
+      "loss": 0.6906,
+      "step": 18450
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000647298388810951,
+      "loss": 0.6543,
+      "step": 18460
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006472133656421374,
+      "loss": 0.6608,
+      "step": 18470
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000647128342473324,
+      "loss": 0.5981,
+      "step": 18480
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0006470433193045105,
+      "loss": 0.5872,
+      "step": 18490
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000646958296135697,
+      "loss": 0.5986,
+      "step": 18500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006468732729668835,
+      "loss": 0.6411,
+      "step": 18510
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00064678824979807,
+      "loss": 0.5512,
+      "step": 18520
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006467032266292565,
+      "loss": 0.6077,
+      "step": 18530
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000646618203460443,
+      "loss": 0.5775,
+      "step": 18540
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006465331802916294,
+      "loss": 0.7422,
+      "step": 18550
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000646448157122816,
+      "loss": 0.657,
+      "step": 18560
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006463631339540026,
+      "loss": 0.5832,
+      "step": 18570
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000646278110785189,
+      "loss": 0.5656,
+      "step": 18580
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006461930876163755,
+      "loss": 0.6158,
+      "step": 18590
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006461080644475619,
+      "loss": 0.6417,
+      "step": 18600
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0006460230412787484,
+      "loss": 0.663,
+      "step": 18610
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000645938018109935,
+      "loss": 0.6638,
+      "step": 18620
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006458529949411215,
+      "loss": 0.6728,
+      "step": 18630
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.000645767971772308,
+      "loss": 0.6932,
+      "step": 18640
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006456829486034945,
+      "loss": 0.5898,
+      "step": 18650
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.000645597925434681,
+      "loss": 0.6499,
+      "step": 18660
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006455129022658675,
+      "loss": 0.6707,
+      "step": 18670
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.000645427879097054,
+      "loss": 0.685,
+      "step": 18680
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006453428559282404,
+      "loss": 0.6747,
+      "step": 18690
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006452578327594271,
+      "loss": 0.7202,
+      "step": 18700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006451728095906135,
+      "loss": 0.6065,
+      "step": 18710
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006450877864218,
+      "loss": 0.5883,
+      "step": 18720
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006450027632529865,
+      "loss": 0.6277,
+      "step": 18730
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0006449177400841729,
+      "loss": 0.6589,
+      "step": 18740
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006448327169153595,
+      "loss": 0.6952,
+      "step": 18750
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000644747693746546,
+      "loss": 0.6713,
+      "step": 18760
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006446626705777324,
+      "loss": 0.5802,
+      "step": 18770
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000644577647408919,
+      "loss": 0.6017,
+      "step": 18780
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006444926242401054,
+      "loss": 0.6841,
+      "step": 18790
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000644407601071292,
+      "loss": 0.5652,
+      "step": 18800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006443225779024785,
+      "loss": 0.604,
+      "step": 18810
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006442375547336649,
+      "loss": 0.637,
+      "step": 18820
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006441525315648514,
+      "loss": 0.6376,
+      "step": 18830
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000644067508396038,
+      "loss": 0.6783,
+      "step": 18840
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0006439824852272245,
+      "loss": 0.6168,
+      "step": 18850
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.000643897462058411,
+      "loss": 0.5904,
+      "step": 18860
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006438124388895974,
+      "loss": 0.6345,
+      "step": 18870
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006437274157207839,
+      "loss": 0.616,
+      "step": 18880
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006436423925519705,
+      "loss": 0.6263,
+      "step": 18890
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006435573693831569,
+      "loss": 0.5982,
+      "step": 18900
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006434723462143434,
+      "loss": 0.6632,
+      "step": 18910
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006433873230455299,
+      "loss": 0.6199,
+      "step": 18920
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006433022998767165,
+      "loss": 0.6251,
+      "step": 18930
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.000643217276707903,
+      "loss": 0.5733,
+      "step": 18940
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006431322535390894,
+      "loss": 0.5859,
+      "step": 18950
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006430472303702759,
+      "loss": 0.6607,
+      "step": 18960
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0006429622072014624,
+      "loss": 0.6278,
+      "step": 18970
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006428771840326489,
+      "loss": 0.6033,
+      "step": 18980
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006427921608638355,
+      "loss": 0.5949,
+      "step": 18990
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000642707137695022,
+      "loss": 0.6522,
+      "step": 19000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006426221145262084,
+      "loss": 0.7001,
+      "step": 19010
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000642537091357395,
+      "loss": 0.7048,
+      "step": 19020
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006424520681885814,
+      "loss": 0.6528,
+      "step": 19030
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006423670450197679,
+      "loss": 0.7149,
+      "step": 19040
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006422820218509544,
+      "loss": 0.6163,
+      "step": 19050
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006421969986821408,
+      "loss": 0.6334,
+      "step": 19060
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006421119755133275,
+      "loss": 0.6026,
+      "step": 19070
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000642026952344514,
+      "loss": 0.6228,
+      "step": 19080
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0006419419291757004,
+      "loss": 0.6441,
+      "step": 19090
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006418569060068869,
+      "loss": 0.5912,
+      "step": 19100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006417718828380734,
+      "loss": 0.6376,
+      "step": 19110
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006416868596692599,
+      "loss": 0.6579,
+      "step": 19120
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006416018365004464,
+      "loss": 0.5704,
+      "step": 19130
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006415168133316329,
+      "loss": 0.5325,
+      "step": 19140
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006414317901628194,
+      "loss": 0.5912,
+      "step": 19150
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.000641346766994006,
+      "loss": 0.6748,
+      "step": 19160
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006412617438251924,
+      "loss": 0.6224,
+      "step": 19170
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006411767206563789,
+      "loss": 0.6072,
+      "step": 19180
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006410916974875654,
+      "loss": 0.6658,
+      "step": 19190
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006410066743187518,
+      "loss": 0.6073,
+      "step": 19200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0006409216511499385,
+      "loss": 0.6554,
+      "step": 19210
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006408366279811249,
+      "loss": 0.5585,
+      "step": 19220
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006407516048123114,
+      "loss": 0.6024,
+      "step": 19230
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006406665816434979,
+      "loss": 0.5465,
+      "step": 19240
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006405815584746843,
+      "loss": 0.6253,
+      "step": 19250
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006404965353058709,
+      "loss": 0.6311,
+      "step": 19260
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006404115121370574,
+      "loss": 0.6117,
+      "step": 19270
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006403264889682439,
+      "loss": 0.6687,
+      "step": 19280
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006402414657994304,
+      "loss": 0.6523,
+      "step": 19290
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006401564426306169,
+      "loss": 0.5379,
+      "step": 19300
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006400714194618034,
+      "loss": 0.632,
+      "step": 19310
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006399863962929899,
+      "loss": 0.683,
+      "step": 19320
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0006399013731241763,
+      "loss": 0.6054,
+      "step": 19330
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006398163499553628,
+      "loss": 0.6257,
+      "step": 19340
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006397313267865494,
+      "loss": 0.5951,
+      "step": 19350
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006396463036177359,
+      "loss": 0.6355,
+      "step": 19360
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006395612804489224,
+      "loss": 0.6807,
+      "step": 19370
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006394762572801088,
+      "loss": 0.6161,
+      "step": 19380
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006393912341112953,
+      "loss": 0.5931,
+      "step": 19390
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006393062109424819,
+      "loss": 0.6676,
+      "step": 19400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006392211877736683,
+      "loss": 0.6383,
+      "step": 19410
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006391361646048548,
+      "loss": 0.6157,
+      "step": 19420
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006390511414360413,
+      "loss": 0.5588,
+      "step": 19430
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006389661182672279,
+      "loss": 0.5594,
+      "step": 19440
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0006388810950984144,
+      "loss": 0.6619,
+      "step": 19450
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006387960719296008,
+      "loss": 0.6144,
+      "step": 19460
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006387110487607873,
+      "loss": 0.6651,
+      "step": 19470
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006386260255919738,
+      "loss": 0.6224,
+      "step": 19480
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006385410024231603,
+      "loss": 0.5932,
+      "step": 19490
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006384559792543469,
+      "loss": 0.6517,
+      "step": 19500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006383709560855334,
+      "loss": 0.6266,
+      "step": 19510
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006382859329167198,
+      "loss": 0.6808,
+      "step": 19520
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006382009097479064,
+      "loss": 0.6079,
+      "step": 19530
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006381158865790928,
+      "loss": 0.6317,
+      "step": 19540
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006380308634102793,
+      "loss": 0.719,
+      "step": 19550
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0006379458402414658,
+      "loss": 0.6171,
+      "step": 19560
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006378608170726523,
+      "loss": 0.5994,
+      "step": 19570
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006377757939038389,
+      "loss": 0.6439,
+      "step": 19580
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006376907707350254,
+      "loss": 0.5717,
+      "step": 19590
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006376057475662118,
+      "loss": 0.663,
+      "step": 19600
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006375207243973983,
+      "loss": 0.5721,
+      "step": 19610
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006374357012285849,
+      "loss": 0.6466,
+      "step": 19620
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006373506780597713,
+      "loss": 0.6977,
+      "step": 19630
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006372656548909578,
+      "loss": 0.5732,
+      "step": 19640
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006371806317221443,
+      "loss": 0.647,
+      "step": 19650
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006370956085533308,
+      "loss": 0.6357,
+      "step": 19660
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006370105853845174,
+      "loss": 0.6748,
+      "step": 19670
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0006369255622157038,
+      "loss": 0.5674,
+      "step": 19680
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006368405390468903,
+      "loss": 0.7033,
+      "step": 19690
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006367555158780768,
+      "loss": 0.5643,
+      "step": 19700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006366704927092632,
+      "loss": 0.5864,
+      "step": 19710
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006365854695404499,
+      "loss": 0.6351,
+      "step": 19720
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006365004463716363,
+      "loss": 0.5874,
+      "step": 19730
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006364154232028228,
+      "loss": 0.6028,
+      "step": 19740
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006363304000340093,
+      "loss": 0.6514,
+      "step": 19750
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006362453768651958,
+      "loss": 0.6332,
+      "step": 19760
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006361603536963823,
+      "loss": 0.5827,
+      "step": 19770
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006360753305275688,
+      "loss": 0.5984,
+      "step": 19780
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006359903073587553,
+      "loss": 0.6715,
+      "step": 19790
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0006359052841899418,
+      "loss": 0.6056,
+      "step": 19800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006358202610211283,
+      "loss": 0.6282,
+      "step": 19810
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006357352378523148,
+      "loss": 0.6524,
+      "step": 19820
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006356502146835013,
+      "loss": 0.5829,
+      "step": 19830
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006355651915146877,
+      "loss": 0.644,
+      "step": 19840
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006354801683458742,
+      "loss": 0.6471,
+      "step": 19850
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006353951451770608,
+      "loss": 0.6411,
+      "step": 19860
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006353101220082473,
+      "loss": 0.6192,
+      "step": 19870
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006352250988394338,
+      "loss": 0.6535,
+      "step": 19880
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006351400756706202,
+      "loss": 0.6164,
+      "step": 19890
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006350550525018068,
+      "loss": 0.5644,
+      "step": 19900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006349700293329933,
+      "loss": 0.5941,
+      "step": 19910
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0006348850061641797,
+      "loss": 0.5944,
+      "step": 19920
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006347999829953662,
+      "loss": 0.6703,
+      "step": 19930
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006347149598265529,
+      "loss": 0.563,
+      "step": 19940
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006346299366577393,
+      "loss": 0.5852,
+      "step": 19950
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006345449134889258,
+      "loss": 0.6133,
+      "step": 19960
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006344598903201122,
+      "loss": 0.6841,
+      "step": 19970
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006343748671512987,
+      "loss": 0.6275,
+      "step": 19980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006342898439824853,
+      "loss": 0.6054,
+      "step": 19990
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006342048208136717,
+      "loss": 0.6285,
+      "step": 20000
+    },
+    {
+      "epoch": 1.69,
+      "eval_loss": 0.6085468530654907,
+      "eval_runtime": 312.7843,
+      "eval_samples_per_second": 16.801,
+      "eval_steps_per_second": 2.1,
+      "step": 20000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006341197976448583,
+      "loss": 0.6362,
+      "step": 20010
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006340347744760448,
+      "loss": 0.6222,
+      "step": 20020
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006339497513072312,
+      "loss": 0.6666,
+      "step": 20030
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0006338647281384178,
+      "loss": 0.6152,
+      "step": 20040
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006337797049696043,
+      "loss": 0.608,
+      "step": 20050
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006336946818007907,
+      "loss": 0.7048,
+      "step": 20060
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006336096586319772,
+      "loss": 0.7125,
+      "step": 20070
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006335246354631638,
+      "loss": 0.6633,
+      "step": 20080
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006334396122943503,
+      "loss": 0.5025,
+      "step": 20090
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006333545891255368,
+      "loss": 0.6475,
+      "step": 20100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006332695659567232,
+      "loss": 0.6241,
+      "step": 20110
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006331845427879097,
+      "loss": 0.6452,
+      "step": 20120
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006330995196190963,
+      "loss": 0.6668,
+      "step": 20130
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006330144964502827,
+      "loss": 0.6494,
+      "step": 20140
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0006329294732814692,
+      "loss": 0.6065,
+      "step": 20150
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006328444501126557,
+      "loss": 0.5738,
+      "step": 20160
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006327594269438422,
+      "loss": 0.6155,
+      "step": 20170
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006326744037750288,
+      "loss": 0.6367,
+      "step": 20180
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006325893806062152,
+      "loss": 0.5738,
+      "step": 20190
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006325043574374017,
+      "loss": 0.5168,
+      "step": 20200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006324193342685882,
+      "loss": 0.6013,
+      "step": 20210
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006323343110997747,
+      "loss": 0.5642,
+      "step": 20220
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006322492879309613,
+      "loss": 0.5858,
+      "step": 20230
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006321642647621477,
+      "loss": 0.6007,
+      "step": 20240
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006320792415933342,
+      "loss": 0.6001,
+      "step": 20250
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006319942184245207,
+      "loss": 0.633,
+      "step": 20260
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0006319091952557072,
+      "loss": 0.5867,
+      "step": 20270
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006318241720868937,
+      "loss": 0.5851,
+      "step": 20280
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006317391489180802,
+      "loss": 0.6096,
+      "step": 20290
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006316541257492667,
+      "loss": 0.5936,
+      "step": 20300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006315691025804533,
+      "loss": 0.6935,
+      "step": 20310
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006314840794116397,
+      "loss": 0.517,
+      "step": 20320
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006313990562428262,
+      "loss": 0.6081,
+      "step": 20330
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006313140330740127,
+      "loss": 0.5282,
+      "step": 20340
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006312290099051991,
+      "loss": 0.6574,
+      "step": 20350
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006311439867363857,
+      "loss": 0.6277,
+      "step": 20360
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006310589635675722,
+      "loss": 0.6201,
+      "step": 20370
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006309739403987587,
+      "loss": 0.6131,
+      "step": 20380
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0006308889172299452,
+      "loss": 0.5985,
+      "step": 20390
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006308038940611316,
+      "loss": 0.6179,
+      "step": 20400
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006307188708923182,
+      "loss": 0.6775,
+      "step": 20410
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006306338477235047,
+      "loss": 0.6289,
+      "step": 20420
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006305488245546911,
+      "loss": 0.6725,
+      "step": 20430
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006304638013858776,
+      "loss": 0.589,
+      "step": 20440
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006303787782170643,
+      "loss": 0.6242,
+      "step": 20450
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006302937550482507,
+      "loss": 0.5873,
+      "step": 20460
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006302087318794372,
+      "loss": 0.594,
+      "step": 20470
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006301237087106236,
+      "loss": 0.653,
+      "step": 20480
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006300386855418101,
+      "loss": 0.6522,
+      "step": 20490
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006299536623729967,
+      "loss": 0.674,
+      "step": 20500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0006298686392041831,
+      "loss": 0.6369,
+      "step": 20510
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006297836160353697,
+      "loss": 0.6676,
+      "step": 20520
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006296985928665562,
+      "loss": 0.5991,
+      "step": 20530
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006296135696977427,
+      "loss": 0.6549,
+      "step": 20540
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006295285465289292,
+      "loss": 0.5819,
+      "step": 20550
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006294435233601157,
+      "loss": 0.6228,
+      "step": 20560
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006293585001913021,
+      "loss": 0.6272,
+      "step": 20570
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006292734770224886,
+      "loss": 0.6104,
+      "step": 20580
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006291884538536752,
+      "loss": 0.5872,
+      "step": 20590
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006291034306848617,
+      "loss": 0.6275,
+      "step": 20600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006290184075160482,
+      "loss": 0.5759,
+      "step": 20610
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006289333843472346,
+      "loss": 0.5862,
+      "step": 20620
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0006288483611784211,
+      "loss": 0.638,
+      "step": 20630
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006287633380096077,
+      "loss": 0.6055,
+      "step": 20640
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006286783148407941,
+      "loss": 0.5945,
+      "step": 20650
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006285932916719807,
+      "loss": 0.5903,
+      "step": 20660
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006285082685031671,
+      "loss": 0.5984,
+      "step": 20670
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006284232453343537,
+      "loss": 0.6236,
+      "step": 20680
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006283382221655402,
+      "loss": 0.5569,
+      "step": 20690
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006282531989967266,
+      "loss": 0.5384,
+      "step": 20700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006281681758279131,
+      "loss": 0.6644,
+      "step": 20710
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006280831526590996,
+      "loss": 0.5616,
+      "step": 20720
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006279981294902861,
+      "loss": 0.6251,
+      "step": 20730
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006279131063214727,
+      "loss": 0.6889,
+      "step": 20740
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0006278280831526591,
+      "loss": 0.634,
+      "step": 20750
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006277430599838456,
+      "loss": 0.5885,
+      "step": 20760
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006276580368150322,
+      "loss": 0.6706,
+      "step": 20770
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006275730136462186,
+      "loss": 0.5591,
+      "step": 20780
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006274879904774051,
+      "loss": 0.6242,
+      "step": 20790
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006274029673085916,
+      "loss": 0.6043,
+      "step": 20800
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006273179441397781,
+      "loss": 0.6505,
+      "step": 20810
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006272329209709647,
+      "loss": 0.5711,
+      "step": 20820
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006271478978021511,
+      "loss": 0.5764,
+      "step": 20830
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006270628746333376,
+      "loss": 0.5759,
+      "step": 20840
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006269778514645241,
+      "loss": 0.6829,
+      "step": 20850
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0006268928282957105,
+      "loss": 0.6412,
+      "step": 20860
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006268078051268971,
+      "loss": 0.6023,
+      "step": 20870
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006267227819580837,
+      "loss": 0.6269,
+      "step": 20880
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006266377587892701,
+      "loss": 0.6234,
+      "step": 20890
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006265527356204566,
+      "loss": 0.7171,
+      "step": 20900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006264677124516431,
+      "loss": 0.6421,
+      "step": 20910
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006263826892828296,
+      "loss": 0.68,
+      "step": 20920
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006262976661140161,
+      "loss": 0.6571,
+      "step": 20930
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006262126429452025,
+      "loss": 0.652,
+      "step": 20940
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.000626127619776389,
+      "loss": 0.5967,
+      "step": 20950
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006260425966075757,
+      "loss": 0.5557,
+      "step": 20960
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006259575734387621,
+      "loss": 0.605,
+      "step": 20970
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0006258725502699486,
+      "loss": 0.6368,
+      "step": 20980
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006257875271011351,
+      "loss": 0.6264,
+      "step": 20990
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006257025039323216,
+      "loss": 0.6421,
+      "step": 21000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006256174807635081,
+      "loss": 0.6169,
+      "step": 21010
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006255324575946945,
+      "loss": 0.5591,
+      "step": 21020
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006254474344258811,
+      "loss": 0.7132,
+      "step": 21030
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006253624112570676,
+      "loss": 0.6339,
+      "step": 21040
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006252773880882541,
+      "loss": 0.6099,
+      "step": 21050
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006251923649194406,
+      "loss": 0.6916,
+      "step": 21060
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006251073417506271,
+      "loss": 0.6152,
+      "step": 21070
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006250223185818135,
+      "loss": 0.6826,
+      "step": 21080
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.000624937295413,
+      "loss": 0.6092,
+      "step": 21090
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0006248522722441866,
+      "loss": 0.5963,
+      "step": 21100
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006247672490753731,
+      "loss": 0.6013,
+      "step": 21110
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006246822259065596,
+      "loss": 0.5932,
+      "step": 21120
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.000624597202737746,
+      "loss": 0.6593,
+      "step": 21130
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006245121795689326,
+      "loss": 0.5453,
+      "step": 21140
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006244271564001191,
+      "loss": 0.6457,
+      "step": 21150
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006243421332313055,
+      "loss": 0.6201,
+      "step": 21160
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006242571100624921,
+      "loss": 0.6955,
+      "step": 21170
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006241720868936785,
+      "loss": 0.6634,
+      "step": 21180
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006240870637248651,
+      "loss": 0.6414,
+      "step": 21190
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006240020405560516,
+      "loss": 0.677,
+      "step": 21200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.000623917017387238,
+      "loss": 0.5863,
+      "step": 21210
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0006238319942184245,
+      "loss": 0.5528,
+      "step": 21220
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006237469710496111,
+      "loss": 0.688,
+      "step": 21230
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006236619478807975,
+      "loss": 0.6343,
+      "step": 21240
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006235769247119841,
+      "loss": 0.5734,
+      "step": 21250
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006234919015431705,
+      "loss": 0.6381,
+      "step": 21260
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.000623406878374357,
+      "loss": 0.6742,
+      "step": 21270
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006233218552055436,
+      "loss": 0.6292,
+      "step": 21280
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00062323683203673,
+      "loss": 0.6273,
+      "step": 21290
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006231518088679165,
+      "loss": 0.6551,
+      "step": 21300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.000623066785699103,
+      "loss": 0.6025,
+      "step": 21310
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006229817625302896,
+      "loss": 0.601,
+      "step": 21320
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006228967393614761,
+      "loss": 0.6389,
+      "step": 21330
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006228117161926625,
+      "loss": 0.5949,
+      "step": 21340
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.000622726693023849,
+      "loss": 0.5879,
+      "step": 21350
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006226416698550355,
+      "loss": 0.5906,
+      "step": 21360
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.000622556646686222,
+      "loss": 0.6646,
+      "step": 21370
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006224716235174085,
+      "loss": 0.5935,
+      "step": 21380
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006223866003485951,
+      "loss": 0.7168,
+      "step": 21390
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006223015771797815,
+      "loss": 0.5607,
+      "step": 21400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.000622216554010968,
+      "loss": 0.5916,
+      "step": 21410
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006221315308421546,
+      "loss": 0.6827,
+      "step": 21420
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.000622046507673341,
+      "loss": 0.6082,
+      "step": 21430
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006219614845045275,
+      "loss": 0.6453,
+      "step": 21440
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006218764613357139,
+      "loss": 0.5844,
+      "step": 21450
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006217914381669006,
+      "loss": 0.6085,
+      "step": 21460
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006217064149980871,
+      "loss": 0.6164,
+      "step": 21470
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006216213918292735,
+      "loss": 0.5937,
+      "step": 21480
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00062153636866046,
+      "loss": 0.6266,
+      "step": 21490
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006214513454916465,
+      "loss": 0.6488,
+      "step": 21500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000621366322322833,
+      "loss": 0.6797,
+      "step": 21510
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006212812991540195,
+      "loss": 0.5986,
+      "step": 21520
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000621196275985206,
+      "loss": 0.5917,
+      "step": 21530
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006211112528163925,
+      "loss": 0.6843,
+      "step": 21540
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006210262296475791,
+      "loss": 0.6081,
+      "step": 21550
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006209412064787655,
+      "loss": 0.6167,
+      "step": 21560
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000620856183309952,
+      "loss": 0.6463,
+      "step": 21570
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006207711601411385,
+      "loss": 0.5896,
+      "step": 21580
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006206861369723249,
+      "loss": 0.6482,
+      "step": 21590
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006206011138035115,
+      "loss": 0.5741,
+      "step": 21600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000620516090634698,
+      "loss": 0.6225,
+      "step": 21610
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006204310674658845,
+      "loss": 0.6823,
+      "step": 21620
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000620346044297071,
+      "loss": 0.6362,
+      "step": 21630
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006202610211282574,
+      "loss": 0.6249,
+      "step": 21640
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000620175997959444,
+      "loss": 0.6621,
+      "step": 21650
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006200909747906305,
+      "loss": 0.6011,
+      "step": 21660
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006200059516218169,
+      "loss": 0.6328,
+      "step": 21670
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006199209284530035,
+      "loss": 0.5975,
+      "step": 21680
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00061983590528419,
+      "loss": 0.6585,
+      "step": 21690
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006197508821153765,
+      "loss": 0.5901,
+      "step": 21700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.000619665858946563,
+      "loss": 0.6895,
+      "step": 21710
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006195808357777494,
+      "loss": 0.6058,
+      "step": 21720
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006194958126089359,
+      "loss": 0.6626,
+      "step": 21730
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006194107894401225,
+      "loss": 0.5945,
+      "step": 21740
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.000619325766271309,
+      "loss": 0.626,
+      "step": 21750
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006192407431024955,
+      "loss": 0.5456,
+      "step": 21760
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006191557199336819,
+      "loss": 0.6143,
+      "step": 21770
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006190706967648685,
+      "loss": 0.5895,
+      "step": 21780
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.000618985673596055,
+      "loss": 0.6891,
+      "step": 21790
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006189006504272414,
+      "loss": 0.618,
+      "step": 21800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006188156272584279,
+      "loss": 0.6552,
+      "step": 21810
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006187306040896144,
+      "loss": 0.5392,
+      "step": 21820
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000618645580920801,
+      "loss": 0.7507,
+      "step": 21830
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006185605577519875,
+      "loss": 0.7057,
+      "step": 21840
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006184755345831739,
+      "loss": 0.6401,
+      "step": 21850
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006183905114143604,
+      "loss": 0.5897,
+      "step": 21860
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000618305488245547,
+      "loss": 0.6449,
+      "step": 21870
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006182204650767334,
+      "loss": 0.6624,
+      "step": 21880
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006181354419079199,
+      "loss": 0.6533,
+      "step": 21890
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006180504187391065,
+      "loss": 0.5979,
+      "step": 21900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006179653955702929,
+      "loss": 0.5958,
+      "step": 21910
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006178803724014795,
+      "loss": 0.5527,
+      "step": 21920
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000617795349232666,
+      "loss": 0.616,
+      "step": 21930
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006177103260638524,
+      "loss": 0.703,
+      "step": 21940
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006176253028950389,
+      "loss": 0.5839,
+      "step": 21950
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006175402797262253,
+      "loss": 0.6244,
+      "step": 21960
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.000617455256557412,
+      "loss": 0.5782,
+      "step": 21970
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006173702333885985,
+      "loss": 0.6263,
+      "step": 21980
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006172852102197849,
+      "loss": 0.6053,
+      "step": 21990
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006172001870509714,
+      "loss": 0.5943,
+      "step": 22000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.000617115163882158,
+      "loss": 0.6424,
+      "step": 22010
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006170301407133444,
+      "loss": 0.6835,
+      "step": 22020
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006169451175445309,
+      "loss": 0.6002,
+      "step": 22030
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006168600943757174,
+      "loss": 0.6052,
+      "step": 22040
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006167750712069039,
+      "loss": 0.5593,
+      "step": 22050
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006166900480380905,
+      "loss": 0.6144,
+      "step": 22060
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006166050248692769,
+      "loss": 0.6115,
+      "step": 22070
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006165200017004634,
+      "loss": 0.596,
+      "step": 22080
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006164349785316499,
+      "loss": 0.7178,
+      "step": 22090
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006163499553628363,
+      "loss": 0.6589,
+      "step": 22100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006162649321940229,
+      "loss": 0.6093,
+      "step": 22110
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006161799090252094,
+      "loss": 0.6254,
+      "step": 22120
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006160948858563959,
+      "loss": 0.6287,
+      "step": 22130
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006160098626875824,
+      "loss": 0.627,
+      "step": 22140
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006159248395187689,
+      "loss": 0.5832,
+      "step": 22150
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006158398163499554,
+      "loss": 0.6681,
+      "step": 22160
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006157547931811419,
+      "loss": 0.5781,
+      "step": 22170
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006156697700123283,
+      "loss": 0.6817,
+      "step": 22180
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000615584746843515,
+      "loss": 0.6843,
+      "step": 22190
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006154997236747014,
+      "loss": 0.578,
+      "step": 22200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006154147005058879,
+      "loss": 0.5888,
+      "step": 22210
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006153296773370744,
+      "loss": 0.5969,
+      "step": 22220
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006152446541682608,
+      "loss": 0.6346,
+      "step": 22230
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006151596309994474,
+      "loss": 0.6701,
+      "step": 22240
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006150746078306339,
+      "loss": 0.6089,
+      "step": 22250
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006149895846618204,
+      "loss": 0.6062,
+      "step": 22260
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006149045614930069,
+      "loss": 0.58,
+      "step": 22270
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006148195383241933,
+      "loss": 0.7047,
+      "step": 22280
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006147345151553799,
+      "loss": 0.6245,
+      "step": 22290
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006146494919865664,
+      "loss": 0.5705,
+      "step": 22300
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006145644688177528,
+      "loss": 0.6109,
+      "step": 22310
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006144794456489393,
+      "loss": 0.6045,
+      "step": 22320
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006143944224801258,
+      "loss": 0.5587,
+      "step": 22330
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006143093993113124,
+      "loss": 0.6,
+      "step": 22340
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006142243761424989,
+      "loss": 0.5838,
+      "step": 22350
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006141393529736854,
+      "loss": 0.5754,
+      "step": 22360
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006140543298048718,
+      "loss": 0.6645,
+      "step": 22370
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006139693066360584,
+      "loss": 0.5993,
+      "step": 22380
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006138842834672448,
+      "loss": 0.5956,
+      "step": 22390
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006137992602984313,
+      "loss": 0.5568,
+      "step": 22400
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006137142371296179,
+      "loss": 0.6731,
+      "step": 22410
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006136292139608043,
+      "loss": 0.6516,
+      "step": 22420
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006135441907919909,
+      "loss": 0.561,
+      "step": 22430
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006134591676231774,
+      "loss": 0.7051,
+      "step": 22440
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006133741444543638,
+      "loss": 0.5935,
+      "step": 22450
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006132891212855503,
+      "loss": 0.6336,
+      "step": 22460
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006132040981167369,
+      "loss": 0.6211,
+      "step": 22470
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006131190749479234,
+      "loss": 0.6594,
+      "step": 22480
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006130340517791099,
+      "loss": 0.5752,
+      "step": 22490
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006129490286102963,
+      "loss": 0.6239,
+      "step": 22500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006128640054414828,
+      "loss": 0.5474,
+      "step": 22510
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006127789822726694,
+      "loss": 0.6514,
+      "step": 22520
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006126939591038558,
+      "loss": 0.6119,
+      "step": 22530
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006126089359350423,
+      "loss": 0.57,
+      "step": 22540
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006125239127662288,
+      "loss": 0.5772,
+      "step": 22550
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006124388895974154,
+      "loss": 0.6754,
+      "step": 22560
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006123538664286019,
+      "loss": 0.6682,
+      "step": 22570
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006122688432597883,
+      "loss": 0.6157,
+      "step": 22580
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006121838200909748,
+      "loss": 0.5694,
+      "step": 22590
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006120987969221613,
+      "loss": 0.6386,
+      "step": 22600
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006120137737533478,
+      "loss": 0.6942,
+      "step": 22610
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006119287505845343,
+      "loss": 0.6362,
+      "step": 22620
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006118437274157208,
+      "loss": 0.6133,
+      "step": 22630
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006117587042469073,
+      "loss": 0.6269,
+      "step": 22640
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006116736810780938,
+      "loss": 0.7284,
+      "step": 22650
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006115886579092803,
+      "loss": 0.5732,
+      "step": 22660
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006115036347404668,
+      "loss": 0.7032,
+      "step": 22670
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006114186115716533,
+      "loss": 0.5922,
+      "step": 22680
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006113335884028397,
+      "loss": 0.6913,
+      "step": 22690
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006112485652340264,
+      "loss": 0.6358,
+      "step": 22700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006111635420652128,
+      "loss": 0.6658,
+      "step": 22710
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006110785188963993,
+      "loss": 0.6415,
+      "step": 22720
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006109934957275858,
+      "loss": 0.678,
+      "step": 22730
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006109084725587722,
+      "loss": 0.6273,
+      "step": 22740
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006108234493899588,
+      "loss": 0.5928,
+      "step": 22750
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006107384262211453,
+      "loss": 0.5904,
+      "step": 22760
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006106534030523318,
+      "loss": 0.6471,
+      "step": 22770
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006105683798835183,
+      "loss": 0.6772,
+      "step": 22780
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006104833567147049,
+      "loss": 0.5929,
+      "step": 22790
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006103983335458913,
+      "loss": 0.6042,
+      "step": 22800
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006103133103770778,
+      "loss": 0.6978,
+      "step": 22810
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006102282872082642,
+      "loss": 0.5676,
+      "step": 22820
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006101432640394507,
+      "loss": 0.6701,
+      "step": 22830
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006100582408706374,
+      "loss": 0.6742,
+      "step": 22840
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006099732177018238,
+      "loss": 0.652,
+      "step": 22850
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006098881945330103,
+      "loss": 0.5942,
+      "step": 22860
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006098031713641968,
+      "loss": 0.5618,
+      "step": 22870
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006097181481953832,
+      "loss": 0.6429,
+      "step": 22880
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006096331250265698,
+      "loss": 0.5703,
+      "step": 22890
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006095481018577563,
+      "loss": 0.5993,
+      "step": 22900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006094630786889427,
+      "loss": 0.6247,
+      "step": 22910
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006093780555201293,
+      "loss": 0.5795,
+      "step": 22920
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006092930323513158,
+      "loss": 0.6186,
+      "step": 22930
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006092080091825023,
+      "loss": 0.6651,
+      "step": 22940
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006091229860136888,
+      "loss": 0.6569,
+      "step": 22950
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006090379628448752,
+      "loss": 0.5566,
+      "step": 22960
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006089529396760617,
+      "loss": 0.6016,
+      "step": 22970
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006088679165072483,
+      "loss": 0.6343,
+      "step": 22980
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006087828933384348,
+      "loss": 0.6661,
+      "step": 22990
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006086978701696213,
+      "loss": 0.7402,
+      "step": 23000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006086128470008077,
+      "loss": 0.598,
+      "step": 23010
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006085278238319943,
+      "loss": 0.6497,
+      "step": 23020
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006084428006631808,
+      "loss": 0.649,
+      "step": 23030
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006083577774943672,
+      "loss": 0.5545,
+      "step": 23040
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006082727543255537,
+      "loss": 0.6519,
+      "step": 23050
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006081877311567402,
+      "loss": 0.6239,
+      "step": 23060
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006081027079879268,
+      "loss": 0.5712,
+      "step": 23070
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006080176848191133,
+      "loss": 0.6137,
+      "step": 23080
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006079326616502997,
+      "loss": 0.6434,
+      "step": 23090
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006078476384814862,
+      "loss": 0.6195,
+      "step": 23100
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006077626153126727,
+      "loss": 0.5966,
+      "step": 23110
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006076775921438592,
+      "loss": 0.6524,
+      "step": 23120
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006075925689750458,
+      "loss": 0.5786,
+      "step": 23130
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006075075458062322,
+      "loss": 0.6043,
+      "step": 23140
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006074225226374187,
+      "loss": 0.5927,
+      "step": 23150
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006073374994686053,
+      "loss": 0.6616,
+      "step": 23160
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006072524762997917,
+      "loss": 0.665,
+      "step": 23170
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006071674531309782,
+      "loss": 0.553,
+      "step": 23180
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006070824299621647,
+      "loss": 0.6271,
+      "step": 23190
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006069974067933511,
+      "loss": 0.6529,
+      "step": 23200
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006069123836245378,
+      "loss": 0.6571,
+      "step": 23210
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006068273604557242,
+      "loss": 0.6254,
+      "step": 23220
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006067423372869107,
+      "loss": 0.5857,
+      "step": 23230
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006066573141180972,
+      "loss": 0.6271,
+      "step": 23240
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006065722909492836,
+      "loss": 0.5542,
+      "step": 23250
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006064872677804702,
+      "loss": 0.6833,
+      "step": 23260
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006064022446116567,
+      "loss": 0.5648,
+      "step": 23270
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006063172214428432,
+      "loss": 0.5817,
+      "step": 23280
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006062321982740297,
+      "loss": 0.635,
+      "step": 23290
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006061471751052163,
+      "loss": 0.5889,
+      "step": 23300
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006060621519364027,
+      "loss": 0.5662,
+      "step": 23310
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006059771287675892,
+      "loss": 0.6224,
+      "step": 23320
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006058921055987756,
+      "loss": 0.6108,
+      "step": 23330
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006058070824299621,
+      "loss": 0.6436,
+      "step": 23340
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006057220592611488,
+      "loss": 0.703,
+      "step": 23350
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006056370360923352,
+      "loss": 0.6283,
+      "step": 23360
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006055520129235217,
+      "loss": 0.6167,
+      "step": 23370
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006054669897547082,
+      "loss": 0.6192,
+      "step": 23380
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006053819665858947,
+      "loss": 0.6512,
+      "step": 23390
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006052969434170812,
+      "loss": 0.6279,
+      "step": 23400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006052119202482677,
+      "loss": 0.637,
+      "step": 23410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006051268970794541,
+      "loss": 0.5675,
+      "step": 23420
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006050418739106407,
+      "loss": 0.6254,
+      "step": 23430
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006049568507418272,
+      "loss": 0.641,
+      "step": 23440
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006048718275730137,
+      "loss": 0.5966,
+      "step": 23450
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006047868044042002,
+      "loss": 0.6115,
+      "step": 23460
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006047017812353866,
+      "loss": 0.6342,
+      "step": 23470
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006046167580665732,
+      "loss": 0.6519,
+      "step": 23480
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006045317348977597,
+      "loss": 0.6648,
+      "step": 23490
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006044467117289462,
+      "loss": 0.6105,
+      "step": 23500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006043616885601327,
+      "loss": 0.664,
+      "step": 23510
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006042766653913191,
+      "loss": 0.6052,
+      "step": 23520
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006041916422225057,
+      "loss": 0.634,
+      "step": 23530
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006041066190536922,
+      "loss": 0.5923,
+      "step": 23540
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006040215958848786,
+      "loss": 0.5784,
+      "step": 23550
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006039365727160651,
+      "loss": 0.6633,
+      "step": 23560
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006038515495472516,
+      "loss": 0.5419,
+      "step": 23570
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006037665263784382,
+      "loss": 0.6279,
+      "step": 23580
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006036815032096247,
+      "loss": 0.632,
+      "step": 23590
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006035964800408111,
+      "loss": 0.6088,
+      "step": 23600
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006035114568719976,
+      "loss": 0.6411,
+      "step": 23610
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006034264337031842,
+      "loss": 0.594,
+      "step": 23620
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006033414105343706,
+      "loss": 0.5942,
+      "step": 23630
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006032563873655572,
+      "loss": 0.6361,
+      "step": 23640
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006031713641967436,
+      "loss": 0.5056,
+      "step": 23650
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006030863410279301,
+      "loss": 0.5275,
+      "step": 23660
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006030013178591167,
+      "loss": 0.6022,
+      "step": 23670
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006029162946903031,
+      "loss": 0.6001,
+      "step": 23680
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006028312715214896,
+      "loss": 0.6231,
+      "step": 23690
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006027462483526761,
+      "loss": 0.5702,
+      "step": 23700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006026612251838625,
+      "loss": 0.5261,
+      "step": 23710
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006025762020150492,
+      "loss": 0.5858,
+      "step": 23720
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006024911788462357,
+      "loss": 0.576,
+      "step": 23730
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006024061556774221,
+      "loss": 0.6171,
+      "step": 23740
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006023211325086086,
+      "loss": 0.5878,
+      "step": 23750
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006022361093397951,
+      "loss": 0.5899,
+      "step": 23760
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006021510861709816,
+      "loss": 0.6829,
+      "step": 23770
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006020660630021681,
+      "loss": 0.6296,
+      "step": 23780
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006019810398333546,
+      "loss": 0.584,
+      "step": 23790
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006018960166645412,
+      "loss": 0.5715,
+      "step": 23800
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006018109934957277,
+      "loss": 0.4999,
+      "step": 23810
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006017259703269141,
+      "loss": 0.6281,
+      "step": 23820
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0006016409471581006,
+      "loss": 0.631,
+      "step": 23830
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0006015559239892871,
+      "loss": 0.5635,
+      "step": 23840
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0006014709008204736,
+      "loss": 0.5687,
+      "step": 23850
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0006013858776516602,
+      "loss": 0.578,
+      "step": 23860
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0006013008544828466,
+      "loss": 0.5381,
+      "step": 23870
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0006012158313140331,
+      "loss": 0.5984,
+      "step": 23880
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0006011308081452196,
+      "loss": 0.6001,
+      "step": 23890
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0006010457849764061,
+      "loss": 0.6764,
+      "step": 23900
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0006009607618075926,
+      "loss": 0.6497,
+      "step": 23910
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0006008757386387791,
+      "loss": 0.5858,
+      "step": 23920
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0006007907154699656,
+      "loss": 0.609,
+      "step": 23930
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0006007056923011522,
+      "loss": 0.5945,
+      "step": 23940
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0006006206691323386,
+      "loss": 0.5318,
+      "step": 23950
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0006005356459635251,
+      "loss": 0.5651,
+      "step": 23960
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0006004506227947116,
+      "loss": 0.5702,
+      "step": 23970
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.000600365599625898,
+      "loss": 0.5379,
+      "step": 23980
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0006002805764570846,
+      "loss": 0.5621,
+      "step": 23990
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0006001955532882711,
+      "loss": 0.637,
+      "step": 24000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0006001105301194576,
+      "loss": 0.615,
+      "step": 24010
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0006000255069506441,
+      "loss": 0.6127,
+      "step": 24020
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005999404837818305,
+      "loss": 0.5707,
+      "step": 24030
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005998554606130171,
+      "loss": 0.5977,
+      "step": 24040
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005997704374442036,
+      "loss": 0.598,
+      "step": 24050
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00059968541427539,
+      "loss": 0.55,
+      "step": 24060
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005996003911065765,
+      "loss": 0.5534,
+      "step": 24070
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005995153679377631,
+      "loss": 0.6108,
+      "step": 24080
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005994303447689496,
+      "loss": 0.5856,
+      "step": 24090
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005993453216001361,
+      "loss": 0.6304,
+      "step": 24100
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005992602984313225,
+      "loss": 0.5998,
+      "step": 24110
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.000599175275262509,
+      "loss": 0.6017,
+      "step": 24120
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005990902520936956,
+      "loss": 0.625,
+      "step": 24130
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.000599005228924882,
+      "loss": 0.6012,
+      "step": 24140
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005989202057560686,
+      "loss": 0.5521,
+      "step": 24150
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.000598835182587255,
+      "loss": 0.5438,
+      "step": 24160
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005987501594184416,
+      "loss": 0.6165,
+      "step": 24170
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0005986651362496281,
+      "loss": 0.4988,
+      "step": 24180
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005985801130808145,
+      "loss": 0.5901,
+      "step": 24190
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.000598495089912001,
+      "loss": 0.6009,
+      "step": 24200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005984100667431875,
+      "loss": 0.5613,
+      "step": 24210
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005983250435743741,
+      "loss": 0.5883,
+      "step": 24220
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005982400204055606,
+      "loss": 0.5559,
+      "step": 24230
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005981549972367471,
+      "loss": 0.6297,
+      "step": 24240
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005980699740679335,
+      "loss": 0.5983,
+      "step": 24250
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00059798495089912,
+      "loss": 0.5928,
+      "step": 24260
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005978999277303065,
+      "loss": 0.6342,
+      "step": 24270
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.000597814904561493,
+      "loss": 0.6152,
+      "step": 24280
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005977298813926795,
+      "loss": 0.628,
+      "step": 24290
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.000597644858223866,
+      "loss": 0.6289,
+      "step": 24300
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005975598350550526,
+      "loss": 0.586,
+      "step": 24310
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005974748118862391,
+      "loss": 0.5865,
+      "step": 24320
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005973897887174255,
+      "loss": 0.5505,
+      "step": 24330
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.000597304765548612,
+      "loss": 0.6233,
+      "step": 24340
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005972197423797985,
+      "loss": 0.5964,
+      "step": 24350
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.000597134719210985,
+      "loss": 0.5703,
+      "step": 24360
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005970496960421716,
+      "loss": 0.5919,
+      "step": 24370
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.000596964672873358,
+      "loss": 0.5608,
+      "step": 24380
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005968796497045445,
+      "loss": 0.5778,
+      "step": 24390
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005967946265357311,
+      "loss": 0.6203,
+      "step": 24400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0005967096033669175,
+      "loss": 0.572,
+      "step": 24410
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.000596624580198104,
+      "loss": 0.5987,
+      "step": 24420
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005965395570292905,
+      "loss": 0.6749,
+      "step": 24430
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.000596454533860477,
+      "loss": 0.4968,
+      "step": 24440
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005963695106916636,
+      "loss": 0.5422,
+      "step": 24450
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00059628448752285,
+      "loss": 0.5976,
+      "step": 24460
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005961994643540365,
+      "loss": 0.6786,
+      "step": 24470
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.000596114441185223,
+      "loss": 0.5085,
+      "step": 24480
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005960294180164094,
+      "loss": 0.6339,
+      "step": 24490
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.000595944394847596,
+      "loss": 0.581,
+      "step": 24500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005958593716787825,
+      "loss": 0.5302,
+      "step": 24510
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.000595774348509969,
+      "loss": 0.5055,
+      "step": 24520
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005956893253411555,
+      "loss": 0.5927,
+      "step": 24530
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000595604302172342,
+      "loss": 0.6277,
+      "step": 24540
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005955192790035285,
+      "loss": 0.6554,
+      "step": 24550
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000595434255834715,
+      "loss": 0.577,
+      "step": 24560
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005953492326659014,
+      "loss": 0.5201,
+      "step": 24570
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005952642094970879,
+      "loss": 0.5673,
+      "step": 24580
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005951791863282745,
+      "loss": 0.5746,
+      "step": 24590
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000595094163159461,
+      "loss": 0.5719,
+      "step": 24600
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005950091399906475,
+      "loss": 0.5589,
+      "step": 24610
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005949241168218339,
+      "loss": 0.5261,
+      "step": 24620
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005948390936530205,
+      "loss": 0.5155,
+      "step": 24630
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000594754070484207,
+      "loss": 0.5531,
+      "step": 24640
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005946690473153934,
+      "loss": 0.591,
+      "step": 24650
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00059458402414658,
+      "loss": 0.5625,
+      "step": 24660
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005944990009777665,
+      "loss": 0.6128,
+      "step": 24670
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.000594413977808953,
+      "loss": 0.5369,
+      "step": 24680
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005943289546401395,
+      "loss": 0.5579,
+      "step": 24690
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005942439314713259,
+      "loss": 0.5783,
+      "step": 24700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005941589083025124,
+      "loss": 0.5985,
+      "step": 24710
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.000594073885133699,
+      "loss": 0.5771,
+      "step": 24720
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005939888619648855,
+      "loss": 0.6436,
+      "step": 24730
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.000593903838796072,
+      "loss": 0.556,
+      "step": 24740
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005938188156272585,
+      "loss": 0.5996,
+      "step": 24750
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005937337924584449,
+      "loss": 0.5774,
+      "step": 24760
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0005936487692896315,
+      "loss": 0.6834,
+      "step": 24770
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.000593563746120818,
+      "loss": 0.601,
+      "step": 24780
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005934787229520044,
+      "loss": 0.6598,
+      "step": 24790
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005933936997831909,
+      "loss": 0.6277,
+      "step": 24800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005933086766143774,
+      "loss": 0.5193,
+      "step": 24810
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.000593223653445564,
+      "loss": 0.6241,
+      "step": 24820
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005931386302767505,
+      "loss": 0.5589,
+      "step": 24830
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005930536071079369,
+      "loss": 0.6184,
+      "step": 24840
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005929685839391234,
+      "loss": 0.5519,
+      "step": 24850
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00059288356077031,
+      "loss": 0.5726,
+      "step": 24860
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005927985376014964,
+      "loss": 0.6095,
+      "step": 24870
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.000592713514432683,
+      "loss": 0.5255,
+      "step": 24880
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005926284912638694,
+      "loss": 0.5815,
+      "step": 24890
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005925434680950559,
+      "loss": 0.6111,
+      "step": 24900
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005924584449262425,
+      "loss": 0.5847,
+      "step": 24910
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005923734217574289,
+      "loss": 0.6478,
+      "step": 24920
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005922883985886154,
+      "loss": 0.5124,
+      "step": 24930
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005922033754198019,
+      "loss": 0.5777,
+      "step": 24940
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005921183522509885,
+      "loss": 0.6325,
+      "step": 24950
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.000592033329082175,
+      "loss": 0.6664,
+      "step": 24960
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005919483059133614,
+      "loss": 0.6052,
+      "step": 24970
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005918632827445479,
+      "loss": 0.6325,
+      "step": 24980
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005917782595757344,
+      "loss": 0.6093,
+      "step": 24990
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0005916932364069209,
+      "loss": 0.5735,
+      "step": 25000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005916082132381074,
+      "loss": 0.6149,
+      "step": 25010
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005915231900692939,
+      "loss": 0.6332,
+      "step": 25020
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005914381669004804,
+      "loss": 0.6713,
+      "step": 25030
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.000591353143731667,
+      "loss": 0.6654,
+      "step": 25040
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005912681205628534,
+      "loss": 0.5999,
+      "step": 25050
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005911830973940399,
+      "loss": 0.6057,
+      "step": 25060
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005910980742252264,
+      "loss": 0.5455,
+      "step": 25070
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005910130510564128,
+      "loss": 0.5631,
+      "step": 25080
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005909280278875994,
+      "loss": 0.5786,
+      "step": 25090
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.000590843004718786,
+      "loss": 0.6255,
+      "step": 25100
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005907579815499724,
+      "loss": 0.5672,
+      "step": 25110
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005906729583811589,
+      "loss": 0.6546,
+      "step": 25120
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005905879352123453,
+      "loss": 0.6052,
+      "step": 25130
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005905029120435319,
+      "loss": 0.6368,
+      "step": 25140
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005904178888747184,
+      "loss": 0.6105,
+      "step": 25150
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005903328657059048,
+      "loss": 0.5961,
+      "step": 25160
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005902478425370914,
+      "loss": 0.5906,
+      "step": 25170
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.000590162819368278,
+      "loss": 0.6395,
+      "step": 25180
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005900777961994644,
+      "loss": 0.5788,
+      "step": 25190
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005899927730306509,
+      "loss": 0.6099,
+      "step": 25200
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005899077498618374,
+      "loss": 0.5215,
+      "step": 25210
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005898227266930238,
+      "loss": 0.6175,
+      "step": 25220
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005897377035242104,
+      "loss": 0.5256,
+      "step": 25230
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0005896526803553969,
+      "loss": 0.5516,
+      "step": 25240
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005895676571865834,
+      "loss": 0.6399,
+      "step": 25250
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005894826340177699,
+      "loss": 0.5707,
+      "step": 25260
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005893976108489563,
+      "loss": 0.6377,
+      "step": 25270
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005893125876801429,
+      "loss": 0.6577,
+      "step": 25280
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005892275645113294,
+      "loss": 0.6471,
+      "step": 25290
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005891425413425158,
+      "loss": 0.5414,
+      "step": 25300
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005890575181737024,
+      "loss": 0.606,
+      "step": 25310
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005889724950048889,
+      "loss": 0.58,
+      "step": 25320
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005888874718360754,
+      "loss": 0.5986,
+      "step": 25330
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005888024486672619,
+      "loss": 0.5511,
+      "step": 25340
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005887174254984483,
+      "loss": 0.557,
+      "step": 25350
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005886324023296348,
+      "loss": 0.6044,
+      "step": 25360
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005885473791608214,
+      "loss": 0.6942,
+      "step": 25370
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005884623559920078,
+      "loss": 0.6233,
+      "step": 25380
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005883773328231944,
+      "loss": 0.6234,
+      "step": 25390
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005882923096543808,
+      "loss": 0.6355,
+      "step": 25400
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005882072864855674,
+      "loss": 0.5406,
+      "step": 25410
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005881222633167539,
+      "loss": 0.5661,
+      "step": 25420
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005880372401479403,
+      "loss": 0.5508,
+      "step": 25430
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005879522169791268,
+      "loss": 0.5862,
+      "step": 25440
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005878671938103133,
+      "loss": 0.6038,
+      "step": 25450
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005877821706414999,
+      "loss": 0.5875,
+      "step": 25460
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005876971474726864,
+      "loss": 0.5317,
+      "step": 25470
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0005876121243038728,
+      "loss": 0.6627,
+      "step": 25480
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005875271011350593,
+      "loss": 0.6192,
+      "step": 25490
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005874420779662458,
+      "loss": 0.6132,
+      "step": 25500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005873570547974323,
+      "loss": 0.5927,
+      "step": 25510
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005872720316286188,
+      "loss": 0.5541,
+      "step": 25520
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005871870084598053,
+      "loss": 0.5903,
+      "step": 25530
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005871019852909918,
+      "loss": 0.5671,
+      "step": 25540
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005870169621221784,
+      "loss": 0.5858,
+      "step": 25550
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005869319389533648,
+      "loss": 0.4872,
+      "step": 25560
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005868469157845513,
+      "loss": 0.6059,
+      "step": 25570
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005867618926157378,
+      "loss": 0.5999,
+      "step": 25580
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005866768694469242,
+      "loss": 0.6094,
+      "step": 25590
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005865918462781109,
+      "loss": 0.5517,
+      "step": 25600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005865068231092974,
+      "loss": 0.5753,
+      "step": 25610
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005864217999404838,
+      "loss": 0.5685,
+      "step": 25620
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005863367767716703,
+      "loss": 0.5115,
+      "step": 25630
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005862517536028567,
+      "loss": 0.5496,
+      "step": 25640
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005861667304340433,
+      "loss": 0.508,
+      "step": 25650
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005860817072652298,
+      "loss": 0.607,
+      "step": 25660
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005859966840964162,
+      "loss": 0.6566,
+      "step": 25670
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005859116609276028,
+      "loss": 0.6454,
+      "step": 25680
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005858266377587894,
+      "loss": 0.579,
+      "step": 25690
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005857416145899758,
+      "loss": 0.628,
+      "step": 25700
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0005856565914211623,
+      "loss": 0.5029,
+      "step": 25710
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005855715682523488,
+      "loss": 0.6671,
+      "step": 25720
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005854865450835352,
+      "loss": 0.6456,
+      "step": 25730
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005854015219147218,
+      "loss": 0.6074,
+      "step": 25740
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005853164987459083,
+      "loss": 0.5524,
+      "step": 25750
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005852314755770948,
+      "loss": 0.5637,
+      "step": 25760
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005851464524082813,
+      "loss": 0.565,
+      "step": 25770
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005850614292394678,
+      "loss": 0.6015,
+      "step": 25780
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005849764060706543,
+      "loss": 0.6018,
+      "step": 25790
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005848913829018408,
+      "loss": 0.583,
+      "step": 25800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005848063597330272,
+      "loss": 0.6373,
+      "step": 25810
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005847213365642139,
+      "loss": 0.5087,
+      "step": 25820
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005846363133954003,
+      "loss": 0.6126,
+      "step": 25830
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005845512902265868,
+      "loss": 0.559,
+      "step": 25840
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005844662670577733,
+      "loss": 0.583,
+      "step": 25850
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005843812438889597,
+      "loss": 0.5646,
+      "step": 25860
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005842962207201463,
+      "loss": 0.5854,
+      "step": 25870
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005842111975513328,
+      "loss": 0.5251,
+      "step": 25880
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005841261743825192,
+      "loss": 0.5745,
+      "step": 25890
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005840411512137058,
+      "loss": 0.5718,
+      "step": 25900
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005839561280448922,
+      "loss": 0.5726,
+      "step": 25910
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005838711048760788,
+      "loss": 0.6269,
+      "step": 25920
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005837860817072653,
+      "loss": 0.5206,
+      "step": 25930
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005837010585384517,
+      "loss": 0.5333,
+      "step": 25940
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005836160353696382,
+      "loss": 0.593,
+      "step": 25950
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005835310122008247,
+      "loss": 0.6341,
+      "step": 25960
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005834459890320113,
+      "loss": 0.4621,
+      "step": 25970
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005833609658631978,
+      "loss": 0.642,
+      "step": 25980
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005832759426943842,
+      "loss": 0.556,
+      "step": 25990
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005831909195255707,
+      "loss": 0.6232,
+      "step": 26000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005831058963567573,
+      "loss": 0.6226,
+      "step": 26010
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005830208731879437,
+      "loss": 0.5856,
+      "step": 26020
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005829358500191302,
+      "loss": 0.62,
+      "step": 26030
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005828508268503168,
+      "loss": 0.5943,
+      "step": 26040
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005827658036815032,
+      "loss": 0.5676,
+      "step": 26050
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005826807805126898,
+      "loss": 0.6544,
+      "step": 26060
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0005825957573438762,
+      "loss": 0.5565,
+      "step": 26070
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005825107341750627,
+      "loss": 0.6448,
+      "step": 26080
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005824257110062492,
+      "loss": 0.6323,
+      "step": 26090
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005823406878374356,
+      "loss": 0.6472,
+      "step": 26100
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005822556646686223,
+      "loss": 0.6077,
+      "step": 26110
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005821706414998088,
+      "loss": 0.5932,
+      "step": 26120
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005820856183309952,
+      "loss": 0.5774,
+      "step": 26130
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005820005951621817,
+      "loss": 0.6203,
+      "step": 26140
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005819155719933683,
+      "loss": 0.5977,
+      "step": 26150
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005818305488245547,
+      "loss": 0.5544,
+      "step": 26160
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005817455256557412,
+      "loss": 0.6177,
+      "step": 26170
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005816605024869276,
+      "loss": 0.6101,
+      "step": 26180
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005815754793181143,
+      "loss": 0.5946,
+      "step": 26190
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005814904561493008,
+      "loss": 0.5765,
+      "step": 26200
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005814054329804872,
+      "loss": 0.6553,
+      "step": 26210
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005813204098116737,
+      "loss": 0.6165,
+      "step": 26220
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005812353866428602,
+      "loss": 0.5666,
+      "step": 26230
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005811503634740467,
+      "loss": 0.5729,
+      "step": 26240
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005810653403052332,
+      "loss": 0.6236,
+      "step": 26250
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005809803171364197,
+      "loss": 0.6305,
+      "step": 26260
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005808952939676062,
+      "loss": 0.6912,
+      "step": 26270
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005808102707987927,
+      "loss": 0.6257,
+      "step": 26280
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005807252476299792,
+      "loss": 0.5933,
+      "step": 26290
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0005806402244611657,
+      "loss": 0.5747,
+      "step": 26300
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005805552012923522,
+      "loss": 0.5783,
+      "step": 26310
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005804701781235386,
+      "loss": 0.635,
+      "step": 26320
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005803851549547253,
+      "loss": 0.5549,
+      "step": 26330
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005803001317859117,
+      "loss": 0.5922,
+      "step": 26340
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005802151086170982,
+      "loss": 0.5426,
+      "step": 26350
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005801300854482847,
+      "loss": 0.6141,
+      "step": 26360
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005800450622794711,
+      "loss": 0.5741,
+      "step": 26370
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005799600391106577,
+      "loss": 0.5389,
+      "step": 26380
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005798750159418442,
+      "loss": 0.5587,
+      "step": 26390
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005797899927730307,
+      "loss": 0.5109,
+      "step": 26400
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005797049696042172,
+      "loss": 0.5963,
+      "step": 26410
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005796199464354036,
+      "loss": 0.645,
+      "step": 26420
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005795349232665902,
+      "loss": 0.5945,
+      "step": 26430
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005794499000977767,
+      "loss": 0.5901,
+      "step": 26440
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005793648769289631,
+      "loss": 0.6215,
+      "step": 26450
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005792798537601496,
+      "loss": 0.6057,
+      "step": 26460
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005791948305913362,
+      "loss": 0.5582,
+      "step": 26470
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005791098074225227,
+      "loss": 0.6035,
+      "step": 26480
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005790247842537092,
+      "loss": 0.5201,
+      "step": 26490
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005789397610848956,
+      "loss": 0.6119,
+      "step": 26500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005788547379160821,
+      "loss": 0.5747,
+      "step": 26510
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005787697147472687,
+      "loss": 0.6075,
+      "step": 26520
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005786846915784551,
+      "loss": 0.6213,
+      "step": 26530
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005785996684096416,
+      "loss": 0.6216,
+      "step": 26540
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005785146452408282,
+      "loss": 0.6323,
+      "step": 26550
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005784296220720147,
+      "loss": 0.6882,
+      "step": 26560
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005783445989032012,
+      "loss": 0.609,
+      "step": 26570
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005782595757343876,
+      "loss": 0.5087,
+      "step": 26580
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005781745525655741,
+      "loss": 0.6169,
+      "step": 26590
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005780895293967606,
+      "loss": 0.5565,
+      "step": 26600
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005780045062279471,
+      "loss": 0.551,
+      "step": 26610
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005779194830591337,
+      "loss": 0.5489,
+      "step": 26620
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005778344598903202,
+      "loss": 0.6136,
+      "step": 26630
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005777494367215066,
+      "loss": 0.5808,
+      "step": 26640
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005776644135526932,
+      "loss": 0.6544,
+      "step": 26650
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005775793903838797,
+      "loss": 0.6522,
+      "step": 26660
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005774943672150661,
+      "loss": 0.6623,
+      "step": 26670
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005774093440462526,
+      "loss": 0.6571,
+      "step": 26680
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005773243208774391,
+      "loss": 0.6861,
+      "step": 26690
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005772392977086257,
+      "loss": 0.5637,
+      "step": 26700
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005771542745398122,
+      "loss": 0.5772,
+      "step": 26710
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005770692513709986,
+      "loss": 0.5365,
+      "step": 26720
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005769842282021851,
+      "loss": 0.5709,
+      "step": 26730
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005768992050333716,
+      "loss": 0.6413,
+      "step": 26740
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005768141818645581,
+      "loss": 0.5397,
+      "step": 26750
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005767291586957446,
+      "loss": 0.6371,
+      "step": 26760
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005766441355269311,
+      "loss": 0.6046,
+      "step": 26770
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0005765591123581176,
+      "loss": 0.6073,
+      "step": 26780
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005764740891893042,
+      "loss": 0.597,
+      "step": 26790
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005763890660204906,
+      "loss": 0.6153,
+      "step": 26800
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005763040428516771,
+      "loss": 0.6958,
+      "step": 26810
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005762190196828636,
+      "loss": 0.5549,
+      "step": 26820
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00057613399651405,
+      "loss": 0.5607,
+      "step": 26830
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005760489733452367,
+      "loss": 0.595,
+      "step": 26840
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005759639501764231,
+      "loss": 0.5275,
+      "step": 26850
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005758789270076096,
+      "loss": 0.5809,
+      "step": 26860
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005757939038387961,
+      "loss": 0.5959,
+      "step": 26870
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005757088806699825,
+      "loss": 0.5617,
+      "step": 26880
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005756238575011691,
+      "loss": 0.6154,
+      "step": 26890
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005755388343323556,
+      "loss": 0.6047,
+      "step": 26900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005754538111635421,
+      "loss": 0.5635,
+      "step": 26910
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005753687879947286,
+      "loss": 0.5574,
+      "step": 26920
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005752837648259151,
+      "loss": 0.5385,
+      "step": 26930
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005751987416571016,
+      "loss": 0.5644,
+      "step": 26940
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005751137184882881,
+      "loss": 0.5455,
+      "step": 26950
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005750286953194745,
+      "loss": 0.5395,
+      "step": 26960
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.000574943672150661,
+      "loss": 0.603,
+      "step": 26970
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005748586489818477,
+      "loss": 0.5545,
+      "step": 26980
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005747736258130341,
+      "loss": 0.5693,
+      "step": 26990
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005746886026442206,
+      "loss": 0.533,
+      "step": 27000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.000574603579475407,
+      "loss": 0.5169,
+      "step": 27010
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005745185563065936,
+      "loss": 0.5954,
+      "step": 27020
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005744335331377801,
+      "loss": 0.6289,
+      "step": 27030
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005743485099689665,
+      "loss": 0.5694,
+      "step": 27040
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.000574263486800153,
+      "loss": 0.5481,
+      "step": 27050
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005741784636313396,
+      "loss": 0.5655,
+      "step": 27060
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005740934404625261,
+      "loss": 0.6001,
+      "step": 27070
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005740084172937126,
+      "loss": 0.5734,
+      "step": 27080
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005739233941248991,
+      "loss": 0.5761,
+      "step": 27090
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005738383709560855,
+      "loss": 0.6066,
+      "step": 27100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.000573753347787272,
+      "loss": 0.6072,
+      "step": 27110
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005736683246184585,
+      "loss": 0.5557,
+      "step": 27120
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0005735833014496451,
+      "loss": 0.582,
+      "step": 27130
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005734982782808316,
+      "loss": 0.6465,
+      "step": 27140
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.000573413255112018,
+      "loss": 0.6989,
+      "step": 27150
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005733282319432046,
+      "loss": 0.6562,
+      "step": 27160
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005732432087743911,
+      "loss": 0.5932,
+      "step": 27170
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005731581856055775,
+      "loss": 0.5726,
+      "step": 27180
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.000573073162436764,
+      "loss": 0.5831,
+      "step": 27190
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005729881392679505,
+      "loss": 0.5433,
+      "step": 27200
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005729031160991371,
+      "loss": 0.6059,
+      "step": 27210
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005728180929303236,
+      "loss": 0.5691,
+      "step": 27220
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00057273306976151,
+      "loss": 0.5998,
+      "step": 27230
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005726480465926965,
+      "loss": 0.629,
+      "step": 27240
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005725630234238831,
+      "loss": 0.5834,
+      "step": 27250
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005724780002550695,
+      "loss": 0.543,
+      "step": 27260
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.000572392977086256,
+      "loss": 0.5389,
+      "step": 27270
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005723079539174425,
+      "loss": 0.633,
+      "step": 27280
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.000572222930748629,
+      "loss": 0.5575,
+      "step": 27290
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005721379075798156,
+      "loss": 0.616,
+      "step": 27300
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.000572052884411002,
+      "loss": 0.5537,
+      "step": 27310
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005719678612421885,
+      "loss": 0.5591,
+      "step": 27320
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.000571882838073375,
+      "loss": 0.6057,
+      "step": 27330
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005717978149045614,
+      "loss": 0.5534,
+      "step": 27340
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005717127917357481,
+      "loss": 0.5772,
+      "step": 27350
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0005716277685669345,
+      "loss": 0.5375,
+      "step": 27360
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.000571542745398121,
+      "loss": 0.6313,
+      "step": 27370
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005714577222293075,
+      "loss": 0.5907,
+      "step": 27380
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.000571372699060494,
+      "loss": 0.5422,
+      "step": 27390
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005712876758916805,
+      "loss": 0.5819,
+      "step": 27400
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.000571202652722867,
+      "loss": 0.5681,
+      "step": 27410
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005711176295540535,
+      "loss": 0.5098,
+      "step": 27420
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00057103260638524,
+      "loss": 0.6364,
+      "step": 27430
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005709475832164265,
+      "loss": 0.6053,
+      "step": 27440
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.000570862560047613,
+      "loss": 0.586,
+      "step": 27450
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005707775368787995,
+      "loss": 0.5941,
+      "step": 27460
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005706925137099859,
+      "loss": 0.5928,
+      "step": 27470
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005706074905411725,
+      "loss": 0.594,
+      "step": 27480
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005705224673723591,
+      "loss": 0.5897,
+      "step": 27490
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005704374442035455,
+      "loss": 0.6504,
+      "step": 27500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.000570352421034732,
+      "loss": 0.5958,
+      "step": 27510
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005702673978659185,
+      "loss": 0.6274,
+      "step": 27520
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.000570182374697105,
+      "loss": 0.6069,
+      "step": 27530
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005700973515282915,
+      "loss": 0.6002,
+      "step": 27540
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005700123283594779,
+      "loss": 0.5856,
+      "step": 27550
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005699273051906644,
+      "loss": 0.6148,
+      "step": 27560
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005698422820218511,
+      "loss": 0.5308,
+      "step": 27570
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005697572588530375,
+      "loss": 0.5819,
+      "step": 27580
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.000569672235684224,
+      "loss": 0.5971,
+      "step": 27590
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005695872125154105,
+      "loss": 0.6703,
+      "step": 27600
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005695021893465969,
+      "loss": 0.571,
+      "step": 27610
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005694171661777835,
+      "loss": 0.6262,
+      "step": 27620
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00056933214300897,
+      "loss": 0.5858,
+      "step": 27630
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005692471198401565,
+      "loss": 0.6062,
+      "step": 27640
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.000569162096671343,
+      "loss": 0.6133,
+      "step": 27650
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005690770735025294,
+      "loss": 0.5016,
+      "step": 27660
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.000568992050333716,
+      "loss": 0.5778,
+      "step": 27670
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005689070271649025,
+      "loss": 0.6538,
+      "step": 27680
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005688220039960889,
+      "loss": 0.6068,
+      "step": 27690
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005687369808272754,
+      "loss": 0.585,
+      "step": 27700
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.000568651957658462,
+      "loss": 0.6254,
+      "step": 27710
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005685669344896485,
+      "loss": 0.6282,
+      "step": 27720
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.000568481911320835,
+      "loss": 0.6344,
+      "step": 27730
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005683968881520214,
+      "loss": 0.6082,
+      "step": 27740
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005683118649832079,
+      "loss": 0.5797,
+      "step": 27750
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005682268418143945,
+      "loss": 0.6375,
+      "step": 27760
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005681418186455809,
+      "loss": 0.6187,
+      "step": 27770
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005680567954767675,
+      "loss": 0.5479,
+      "step": 27780
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005679717723079539,
+      "loss": 0.5968,
+      "step": 27790
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005678867491391405,
+      "loss": 0.6376,
+      "step": 27800
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.000567801725970327,
+      "loss": 0.7204,
+      "step": 27810
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005677167028015134,
+      "loss": 0.5756,
+      "step": 27820
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005676316796326999,
+      "loss": 0.5412,
+      "step": 27830
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0005675466564638864,
+      "loss": 0.6691,
+      "step": 27840
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005674616332950729,
+      "loss": 0.5688,
+      "step": 27850
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005673766101262595,
+      "loss": 0.6343,
+      "step": 27860
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005672915869574459,
+      "loss": 0.6042,
+      "step": 27870
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005672065637886324,
+      "loss": 0.604,
+      "step": 27880
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.000567121540619819,
+      "loss": 0.5512,
+      "step": 27890
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005670365174510054,
+      "loss": 0.6269,
+      "step": 27900
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005669514942821919,
+      "loss": 0.594,
+      "step": 27910
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005668664711133784,
+      "loss": 0.5841,
+      "step": 27920
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005667814479445649,
+      "loss": 0.6177,
+      "step": 27930
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005666964247757515,
+      "loss": 0.577,
+      "step": 27940
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005666114016069379,
+      "loss": 0.6301,
+      "step": 27950
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005665263784381244,
+      "loss": 0.5732,
+      "step": 27960
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005664413552693109,
+      "loss": 0.6321,
+      "step": 27970
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005663563321004973,
+      "loss": 0.6007,
+      "step": 27980
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005662713089316839,
+      "loss": 0.6286,
+      "step": 27990
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005661862857628705,
+      "loss": 0.6022,
+      "step": 28000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005661012625940569,
+      "loss": 0.6057,
+      "step": 28010
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005660162394252434,
+      "loss": 0.5817,
+      "step": 28020
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00056593121625643,
+      "loss": 0.5678,
+      "step": 28030
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005658461930876164,
+      "loss": 0.5605,
+      "step": 28040
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005657611699188029,
+      "loss": 0.5902,
+      "step": 28050
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005656761467499893,
+      "loss": 0.5406,
+      "step": 28060
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005655911235811759,
+      "loss": 0.5965,
+      "step": 28070
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0005655061004123625,
+      "loss": 0.559,
+      "step": 28080
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005654210772435489,
+      "loss": 0.5955,
+      "step": 28090
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005653360540747354,
+      "loss": 0.5838,
+      "step": 28100
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005652510309059219,
+      "loss": 0.5743,
+      "step": 28110
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005651660077371083,
+      "loss": 0.6413,
+      "step": 28120
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005650809845682949,
+      "loss": 0.6121,
+      "step": 28130
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005649959613994814,
+      "loss": 0.5601,
+      "step": 28140
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005649109382306679,
+      "loss": 0.5156,
+      "step": 28150
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005648259150618544,
+      "loss": 0.6066,
+      "step": 28160
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005647408918930409,
+      "loss": 0.6408,
+      "step": 28170
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005646558687242274,
+      "loss": 0.6113,
+      "step": 28180
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005645708455554139,
+      "loss": 0.6156,
+      "step": 28190
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0005644858223866003,
+      "loss": 0.5207,
+      "step": 28200
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005644007992177868,
+      "loss": 0.5984,
+      "step": 28210
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005643157760489734,
+      "loss": 0.5723,
+      "step": 28220
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005642307528801599,
+      "loss": 0.5728,
+      "step": 28230
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005641457297113464,
+      "loss": 0.5689,
+      "step": 28240
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005640607065425328,
+      "loss": 0.5262,
+      "step": 28250
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005639756833737194,
+      "loss": 0.581,
+      "step": 28260
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005638906602049059,
+      "loss": 0.6018,
+      "step": 28270
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005638056370360923,
+      "loss": 0.6088,
+      "step": 28280
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005637206138672789,
+      "loss": 0.5081,
+      "step": 28290
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005636355906984653,
+      "loss": 0.5853,
+      "step": 28300
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005635505675296519,
+      "loss": 0.5365,
+      "step": 28310
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005634655443608384,
+      "loss": 0.6114,
+      "step": 28320
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005633805211920248,
+      "loss": 0.4823,
+      "step": 28330
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005632954980232113,
+      "loss": 0.6025,
+      "step": 28340
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005632104748543979,
+      "loss": 0.5828,
+      "step": 28350
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005631254516855843,
+      "loss": 0.6329,
+      "step": 28360
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005630404285167709,
+      "loss": 0.5888,
+      "step": 28370
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005629554053479573,
+      "loss": 0.6352,
+      "step": 28380
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005628703821791438,
+      "loss": 0.552,
+      "step": 28390
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005627853590103304,
+      "loss": 0.5849,
+      "step": 28400
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005627003358415168,
+      "loss": 0.5685,
+      "step": 28410
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005626153126727033,
+      "loss": 0.5987,
+      "step": 28420
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0005625302895038898,
+      "loss": 0.5937,
+      "step": 28430
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005624452663350763,
+      "loss": 0.5676,
+      "step": 28440
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005623602431662629,
+      "loss": 0.5929,
+      "step": 28450
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005622752199974494,
+      "loss": 0.6278,
+      "step": 28460
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005621901968286358,
+      "loss": 0.5358,
+      "step": 28470
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005621051736598223,
+      "loss": 0.6409,
+      "step": 28480
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005620201504910087,
+      "loss": 0.6216,
+      "step": 28490
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005619351273221953,
+      "loss": 0.5251,
+      "step": 28500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005618501041533819,
+      "loss": 0.5844,
+      "step": 28510
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005617650809845683,
+      "loss": 0.5704,
+      "step": 28520
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005616800578157548,
+      "loss": 0.6306,
+      "step": 28530
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005615950346469414,
+      "loss": 0.6445,
+      "step": 28540
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005615100114781278,
+      "loss": 0.6491,
+      "step": 28550
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005614249883093143,
+      "loss": 0.5493,
+      "step": 28560
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005613399651405008,
+      "loss": 0.6009,
+      "step": 28570
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005612549419716874,
+      "loss": 0.5586,
+      "step": 28580
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005611699188028739,
+      "loss": 0.574,
+      "step": 28590
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005610848956340603,
+      "loss": 0.6073,
+      "step": 28600
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005609998724652468,
+      "loss": 0.6082,
+      "step": 28610
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005609148492964333,
+      "loss": 0.5053,
+      "step": 28620
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005608298261276198,
+      "loss": 0.6309,
+      "step": 28630
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005607448029588063,
+      "loss": 0.5598,
+      "step": 28640
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005606597797899928,
+      "loss": 0.6712,
+      "step": 28650
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005605747566211793,
+      "loss": 0.5514,
+      "step": 28660
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005604897334523659,
+      "loss": 0.5762,
+      "step": 28670
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005604047102835523,
+      "loss": 0.6287,
+      "step": 28680
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005603196871147388,
+      "loss": 0.5413,
+      "step": 28690
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005602346639459253,
+      "loss": 0.651,
+      "step": 28700
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005601496407771117,
+      "loss": 0.5835,
+      "step": 28710
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005600646176082983,
+      "loss": 0.5953,
+      "step": 28720
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005599795944394848,
+      "loss": 0.6222,
+      "step": 28730
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005598945712706713,
+      "loss": 0.6037,
+      "step": 28740
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005598095481018578,
+      "loss": 0.6791,
+      "step": 28750
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005597245249330442,
+      "loss": 0.5404,
+      "step": 28760
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005596395017642308,
+      "loss": 0.5544,
+      "step": 28770
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005595544785954173,
+      "loss": 0.6123,
+      "step": 28780
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0005594694554266037,
+      "loss": 0.5672,
+      "step": 28790
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005593844322577903,
+      "loss": 0.5428,
+      "step": 28800
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005592994090889768,
+      "loss": 0.6404,
+      "step": 28810
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005592143859201633,
+      "loss": 0.6109,
+      "step": 28820
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005591293627513498,
+      "loss": 0.6147,
+      "step": 28830
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005590443395825362,
+      "loss": 0.6114,
+      "step": 28840
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005589593164137227,
+      "loss": 0.6031,
+      "step": 28850
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005588742932449093,
+      "loss": 0.5738,
+      "step": 28860
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005587892700760958,
+      "loss": 0.5925,
+      "step": 28870
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005587042469072823,
+      "loss": 0.606,
+      "step": 28880
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005586192237384687,
+      "loss": 0.5423,
+      "step": 28890
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0005585342005696552,
+      "loss": 0.526,
+      "step": 28900
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005584491774008418,
+      "loss": 0.5697,
+      "step": 28910
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005583641542320282,
+      "loss": 0.5993,
+      "step": 28920
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005582791310632147,
+      "loss": 0.5802,
+      "step": 28930
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005581941078944012,
+      "loss": 0.6259,
+      "step": 28940
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005581090847255878,
+      "loss": 0.651,
+      "step": 28950
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005580240615567743,
+      "loss": 0.632,
+      "step": 28960
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005579390383879608,
+      "loss": 0.6099,
+      "step": 28970
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005578540152191472,
+      "loss": 0.5974,
+      "step": 28980
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005577689920503337,
+      "loss": 0.516,
+      "step": 28990
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005576839688815202,
+      "loss": 0.5527,
+      "step": 29000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005575989457127067,
+      "loss": 0.6033,
+      "step": 29010
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005575139225438933,
+      "loss": 0.5369,
+      "step": 29020
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005574288993750797,
+      "loss": 0.5936,
+      "step": 29030
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005573438762062663,
+      "loss": 0.512,
+      "step": 29040
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005572588530374528,
+      "loss": 0.5901,
+      "step": 29050
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005571738298686392,
+      "loss": 0.6749,
+      "step": 29060
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005570888066998257,
+      "loss": 0.6128,
+      "step": 29070
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005570037835310122,
+      "loss": 0.5695,
+      "step": 29080
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005569187603621988,
+      "loss": 0.6192,
+      "step": 29090
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005568337371933853,
+      "loss": 0.6068,
+      "step": 29100
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005567487140245717,
+      "loss": 0.6305,
+      "step": 29110
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005566636908557582,
+      "loss": 0.5766,
+      "step": 29120
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005565786676869448,
+      "loss": 0.6102,
+      "step": 29130
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0005564936445181312,
+      "loss": 0.571,
+      "step": 29140
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005564086213493177,
+      "loss": 0.5557,
+      "step": 29150
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005563235981805042,
+      "loss": 0.5474,
+      "step": 29160
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005562385750116907,
+      "loss": 0.5589,
+      "step": 29170
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005561535518428773,
+      "loss": 0.6249,
+      "step": 29180
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005560685286740637,
+      "loss": 0.5487,
+      "step": 29190
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005559835055052502,
+      "loss": 0.5895,
+      "step": 29200
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005558984823364367,
+      "loss": 0.5721,
+      "step": 29210
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005558134591676231,
+      "loss": 0.581,
+      "step": 29220
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005557284359988097,
+      "loss": 0.5773,
+      "step": 29230
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005556434128299962,
+      "loss": 0.6188,
+      "step": 29240
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005555583896611827,
+      "loss": 0.5575,
+      "step": 29250
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0005554733664923692,
+      "loss": 0.5732,
+      "step": 29260
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005553883433235556,
+      "loss": 0.5768,
+      "step": 29270
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005553033201547422,
+      "loss": 0.5805,
+      "step": 29280
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005552182969859287,
+      "loss": 0.5964,
+      "step": 29290
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005551332738171151,
+      "loss": 0.5442,
+      "step": 29300
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005550482506483017,
+      "loss": 0.5534,
+      "step": 29310
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005549632274794882,
+      "loss": 0.5668,
+      "step": 29320
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005548782043106747,
+      "loss": 0.6209,
+      "step": 29330
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005547931811418612,
+      "loss": 0.5558,
+      "step": 29340
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005547081579730476,
+      "loss": 0.6584,
+      "step": 29350
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005546231348042341,
+      "loss": 0.554,
+      "step": 29360
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005545381116354207,
+      "loss": 0.6074,
+      "step": 29370
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005544530884666072,
+      "loss": 0.5264,
+      "step": 29380
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005543680652977937,
+      "loss": 0.5675,
+      "step": 29390
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005542830421289802,
+      "loss": 0.5884,
+      "step": 29400
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005541980189601667,
+      "loss": 0.6023,
+      "step": 29410
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005541129957913532,
+      "loss": 0.6472,
+      "step": 29420
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005540279726225396,
+      "loss": 0.6512,
+      "step": 29430
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005539429494537261,
+      "loss": 0.5944,
+      "step": 29440
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005538579262849128,
+      "loss": 0.576,
+      "step": 29450
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005537729031160992,
+      "loss": 0.5154,
+      "step": 29460
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005536878799472857,
+      "loss": 0.6218,
+      "step": 29470
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005536028567784722,
+      "loss": 0.6302,
+      "step": 29480
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005535178336096586,
+      "loss": 0.6455,
+      "step": 29490
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0005534328104408452,
+      "loss": 0.5947,
+      "step": 29500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005533477872720317,
+      "loss": 0.627,
+      "step": 29510
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005532627641032181,
+      "loss": 0.6177,
+      "step": 29520
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005531777409344047,
+      "loss": 0.5902,
+      "step": 29530
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005530927177655911,
+      "loss": 0.5762,
+      "step": 29540
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005530076945967777,
+      "loss": 0.6046,
+      "step": 29550
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005529226714279642,
+      "loss": 0.7058,
+      "step": 29560
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005528376482591506,
+      "loss": 0.6322,
+      "step": 29570
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005527526250903371,
+      "loss": 0.6022,
+      "step": 29580
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005526676019215237,
+      "loss": 0.5969,
+      "step": 29590
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005525825787527102,
+      "loss": 0.5906,
+      "step": 29600
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005524975555838967,
+      "loss": 0.5781,
+      "step": 29610
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005524125324150831,
+      "loss": 0.5837,
+      "step": 29620
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005523275092462696,
+      "loss": 0.6134,
+      "step": 29630
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005522424860774562,
+      "loss": 0.5957,
+      "step": 29640
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005521574629086426,
+      "loss": 0.6132,
+      "step": 29650
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005520724397398291,
+      "loss": 0.5454,
+      "step": 29660
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005519874165710156,
+      "loss": 0.6293,
+      "step": 29670
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005519023934022021,
+      "loss": 0.6296,
+      "step": 29680
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005518173702333887,
+      "loss": 0.6038,
+      "step": 29690
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005517323470645751,
+      "loss": 0.5699,
+      "step": 29700
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005516473238957616,
+      "loss": 0.5548,
+      "step": 29710
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005515623007269481,
+      "loss": 0.5626,
+      "step": 29720
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005514772775581345,
+      "loss": 0.5468,
+      "step": 29730
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005513922543893211,
+      "loss": 0.5527,
+      "step": 29740
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005513072312205076,
+      "loss": 0.5981,
+      "step": 29750
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005512222080516941,
+      "loss": 0.5856,
+      "step": 29760
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005511371848828806,
+      "loss": 0.6318,
+      "step": 29770
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005510521617140671,
+      "loss": 0.5298,
+      "step": 29780
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005509671385452536,
+      "loss": 0.6158,
+      "step": 29790
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005508821153764401,
+      "loss": 0.5707,
+      "step": 29800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005507970922076265,
+      "loss": 0.5882,
+      "step": 29810
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005507120690388132,
+      "loss": 0.637,
+      "step": 29820
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005506270458699997,
+      "loss": 0.5953,
+      "step": 29830
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005505420227011861,
+      "loss": 0.5559,
+      "step": 29840
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005504569995323726,
+      "loss": 0.5167,
+      "step": 29850
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.000550371976363559,
+      "loss": 0.5862,
+      "step": 29860
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005502869531947456,
+      "loss": 0.6403,
+      "step": 29870
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005502019300259321,
+      "loss": 0.5953,
+      "step": 29880
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005501169068571186,
+      "loss": 0.6316,
+      "step": 29890
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005500318836883051,
+      "loss": 0.6499,
+      "step": 29900
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005499468605194917,
+      "loss": 0.5339,
+      "step": 29910
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005498618373506781,
+      "loss": 0.5672,
+      "step": 29920
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005497768141818646,
+      "loss": 0.6234,
+      "step": 29930
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005496917910130511,
+      "loss": 0.6846,
+      "step": 29940
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005496067678442375,
+      "loss": 0.5995,
+      "step": 29950
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005495217446754242,
+      "loss": 0.6455,
+      "step": 29960
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0005494367215066106,
+      "loss": 0.5822,
+      "step": 29970
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005493516983377971,
+      "loss": 0.5233,
+      "step": 29980
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005492666751689836,
+      "loss": 0.5268,
+      "step": 29990
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00054918165200017,
+      "loss": 0.6163,
+      "step": 30000
+    },
+    {
+      "epoch": 2.54,
+      "eval_loss": 0.5969247817993164,
+      "eval_runtime": 312.0685,
+      "eval_samples_per_second": 16.839,
+      "eval_steps_per_second": 2.105,
+      "step": 30000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005490966288313566,
+      "loss": 0.5615,
+      "step": 30010
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005490116056625431,
+      "loss": 0.5893,
+      "step": 30020
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005489265824937295,
+      "loss": 0.548,
+      "step": 30030
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005488415593249161,
+      "loss": 0.6237,
+      "step": 30040
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005487565361561025,
+      "loss": 0.5534,
+      "step": 30050
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005486715129872891,
+      "loss": 0.62,
+      "step": 30060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005485864898184756,
+      "loss": 0.5549,
+      "step": 30070
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.000548501466649662,
+      "loss": 0.6158,
+      "step": 30080
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005484164434808485,
+      "loss": 0.5454,
+      "step": 30090
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005483314203120351,
+      "loss": 0.5757,
+      "step": 30100
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005482463971432216,
+      "loss": 0.5797,
+      "step": 30110
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005481613739744081,
+      "loss": 0.5438,
+      "step": 30120
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005480763508055945,
+      "loss": 0.6078,
+      "step": 30130
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.000547991327636781,
+      "loss": 0.6315,
+      "step": 30140
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005479063044679676,
+      "loss": 0.582,
+      "step": 30150
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.000547821281299154,
+      "loss": 0.5957,
+      "step": 30160
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005477362581303405,
+      "loss": 0.5492,
+      "step": 30170
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.000547651234961527,
+      "loss": 0.5906,
+      "step": 30180
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005475662117927136,
+      "loss": 0.6231,
+      "step": 30190
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005474811886239001,
+      "loss": 0.5051,
+      "step": 30200
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0005473961654550865,
+      "loss": 0.5223,
+      "step": 30210
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.000547311142286273,
+      "loss": 0.5338,
+      "step": 30220
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005472261191174595,
+      "loss": 0.5295,
+      "step": 30230
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.000547141095948646,
+      "loss": 0.5557,
+      "step": 30240
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005470560727798326,
+      "loss": 0.5543,
+      "step": 30250
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.000546971049611019,
+      "loss": 0.5742,
+      "step": 30260
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005468860264422055,
+      "loss": 0.5954,
+      "step": 30270
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.000546801003273392,
+      "loss": 0.5405,
+      "step": 30280
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005467159801045785,
+      "loss": 0.6113,
+      "step": 30290
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.000546630956935765,
+      "loss": 0.5584,
+      "step": 30300
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005465459337669515,
+      "loss": 0.566,
+      "step": 30310
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0005464609105981379,
+      "loss": 0.5884,
+      "step": 30320
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005463758874293246,
+      "loss": 0.6078,
+      "step": 30330
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005462908642605111,
+      "loss": 0.5252,
+      "step": 30340
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005462058410916975,
+      "loss": 0.5245,
+      "step": 30350
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.000546120817922884,
+      "loss": 0.5767,
+      "step": 30360
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005460357947540704,
+      "loss": 0.6536,
+      "step": 30370
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.000545950771585257,
+      "loss": 0.551,
+      "step": 30380
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005458657484164435,
+      "loss": 0.5116,
+      "step": 30390
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00054578072524763,
+      "loss": 0.5559,
+      "step": 30400
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005456957020788165,
+      "loss": 0.6088,
+      "step": 30410
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005456106789100031,
+      "loss": 0.5783,
+      "step": 30420
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005455256557411895,
+      "loss": 0.5809,
+      "step": 30430
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.000545440632572376,
+      "loss": 0.6167,
+      "step": 30440
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005453556094035625,
+      "loss": 0.5032,
+      "step": 30450
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005452705862347489,
+      "loss": 0.6301,
+      "step": 30460
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005451855630659356,
+      "loss": 0.6059,
+      "step": 30470
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000545100539897122,
+      "loss": 0.6085,
+      "step": 30480
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005450155167283085,
+      "loss": 0.5793,
+      "step": 30490
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000544930493559495,
+      "loss": 0.5611,
+      "step": 30500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005448454703906814,
+      "loss": 0.5512,
+      "step": 30510
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000544760447221868,
+      "loss": 0.5957,
+      "step": 30520
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005446754240530545,
+      "loss": 0.5312,
+      "step": 30530
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000544590400884241,
+      "loss": 0.5311,
+      "step": 30540
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0005445053777154275,
+      "loss": 0.5513,
+      "step": 30550
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000544420354546614,
+      "loss": 0.535,
+      "step": 30560
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005443353313778005,
+      "loss": 0.5625,
+      "step": 30570
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000544250308208987,
+      "loss": 0.5994,
+      "step": 30580
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005441652850401734,
+      "loss": 0.6661,
+      "step": 30590
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00054408026187136,
+      "loss": 0.5818,
+      "step": 30600
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005439952387025465,
+      "loss": 0.6113,
+      "step": 30610
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000543910215533733,
+      "loss": 0.5865,
+      "step": 30620
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005438251923649195,
+      "loss": 0.6298,
+      "step": 30630
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005437401691961059,
+      "loss": 0.626,
+      "step": 30640
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005436551460272925,
+      "loss": 0.6,
+      "step": 30650
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000543570122858479,
+      "loss": 0.6425,
+      "step": 30660
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005434850996896654,
+      "loss": 0.6396,
+      "step": 30670
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005434000765208519,
+      "loss": 0.5826,
+      "step": 30680
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005433150533520384,
+      "loss": 0.5237,
+      "step": 30690
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000543230030183225,
+      "loss": 0.5762,
+      "step": 30700
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005431450070144115,
+      "loss": 0.5579,
+      "step": 30710
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005430599838455979,
+      "loss": 0.5718,
+      "step": 30720
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005429749606767844,
+      "loss": 0.6364,
+      "step": 30730
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000542889937507971,
+      "loss": 0.5772,
+      "step": 30740
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005428049143391574,
+      "loss": 0.5877,
+      "step": 30750
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000542719891170344,
+      "loss": 0.6354,
+      "step": 30760
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005426348680015305,
+      "loss": 0.5952,
+      "step": 30770
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005425498448327169,
+      "loss": 0.6179,
+      "step": 30780
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005424648216639035,
+      "loss": 0.6307,
+      "step": 30790
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005423797984950899,
+      "loss": 0.5566,
+      "step": 30800
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005422947753262764,
+      "loss": 0.4819,
+      "step": 30810
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005422097521574629,
+      "loss": 0.5989,
+      "step": 30820
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005421247289886494,
+      "loss": 0.5717,
+      "step": 30830
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.000542039705819836,
+      "loss": 0.5678,
+      "step": 30840
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005419546826510225,
+      "loss": 0.5516,
+      "step": 30850
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005418696594822089,
+      "loss": 0.5738,
+      "step": 30860
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005417846363133954,
+      "loss": 0.5925,
+      "step": 30870
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.000541699613144582,
+      "loss": 0.6215,
+      "step": 30880
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005416145899757684,
+      "loss": 0.6115,
+      "step": 30890
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005415295668069549,
+      "loss": 0.5694,
+      "step": 30900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005414445436381414,
+      "loss": 0.601,
+      "step": 30910
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.000541359520469328,
+      "loss": 0.532,
+      "step": 30920
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005412744973005145,
+      "loss": 0.6018,
+      "step": 30930
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005411894741317009,
+      "loss": 0.6195,
+      "step": 30940
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005411044509628874,
+      "loss": 0.5853,
+      "step": 30950
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005410194277940739,
+      "loss": 0.6501,
+      "step": 30960
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005409344046252603,
+      "loss": 0.5987,
+      "step": 30970
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.000540849381456447,
+      "loss": 0.5863,
+      "step": 30980
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005407643582876334,
+      "loss": 0.5534,
+      "step": 30990
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005406793351188199,
+      "loss": 0.6379,
+      "step": 31000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005405943119500064,
+      "loss": 0.6175,
+      "step": 31010
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005405092887811929,
+      "loss": 0.5799,
+      "step": 31020
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0005404242656123794,
+      "loss": 0.6482,
+      "step": 31030
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005403392424435659,
+      "loss": 0.569,
+      "step": 31040
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005402542192747524,
+      "loss": 0.5655,
+      "step": 31050
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000540169196105939,
+      "loss": 0.5508,
+      "step": 31060
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005400841729371254,
+      "loss": 0.578,
+      "step": 31070
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005399991497683119,
+      "loss": 0.5813,
+      "step": 31080
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005399141265994984,
+      "loss": 0.6233,
+      "step": 31090
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005398291034306848,
+      "loss": 0.5669,
+      "step": 31100
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005397440802618714,
+      "loss": 0.5694,
+      "step": 31110
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005396590570930579,
+      "loss": 0.7198,
+      "step": 31120
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005395740339242444,
+      "loss": 0.5708,
+      "step": 31130
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005394890107554309,
+      "loss": 0.5526,
+      "step": 31140
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005394039875866173,
+      "loss": 0.5757,
+      "step": 31150
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005393189644178039,
+      "loss": 0.5792,
+      "step": 31160
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005392339412489904,
+      "loss": 0.6258,
+      "step": 31170
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005391489180801768,
+      "loss": 0.5823,
+      "step": 31180
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005390638949113633,
+      "loss": 0.5918,
+      "step": 31190
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00053897887174255,
+      "loss": 0.572,
+      "step": 31200
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005388938485737364,
+      "loss": 0.5467,
+      "step": 31210
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005388088254049229,
+      "loss": 0.584,
+      "step": 31220
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005387238022361093,
+      "loss": 0.5265,
+      "step": 31230
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005386387790672958,
+      "loss": 0.6071,
+      "step": 31240
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005385537558984824,
+      "loss": 0.5863,
+      "step": 31250
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005384687327296688,
+      "loss": 0.625,
+      "step": 31260
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0005383837095608554,
+      "loss": 0.5974,
+      "step": 31270
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005382986863920419,
+      "loss": 0.6316,
+      "step": 31280
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005382136632232283,
+      "loss": 0.6083,
+      "step": 31290
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005381286400544149,
+      "loss": 0.5304,
+      "step": 31300
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005380436168856014,
+      "loss": 0.5879,
+      "step": 31310
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005379585937167878,
+      "loss": 0.5749,
+      "step": 31320
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005378735705479743,
+      "loss": 0.5657,
+      "step": 31330
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005377885473791609,
+      "loss": 0.5919,
+      "step": 31340
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005377035242103474,
+      "loss": 0.5321,
+      "step": 31350
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005376185010415339,
+      "loss": 0.6001,
+      "step": 31360
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005375334778727203,
+      "loss": 0.5958,
+      "step": 31370
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005374484547039068,
+      "loss": 0.571,
+      "step": 31380
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0005373634315350934,
+      "loss": 0.594,
+      "step": 31390
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005372784083662798,
+      "loss": 0.4853,
+      "step": 31400
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005371933851974663,
+      "loss": 0.5605,
+      "step": 31410
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005371083620286528,
+      "loss": 0.5553,
+      "step": 31420
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005370233388598394,
+      "loss": 0.6335,
+      "step": 31430
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005369383156910259,
+      "loss": 0.5549,
+      "step": 31440
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005368532925222123,
+      "loss": 0.5944,
+      "step": 31450
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005367682693533988,
+      "loss": 0.5748,
+      "step": 31460
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005366832461845853,
+      "loss": 0.6028,
+      "step": 31470
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005365982230157718,
+      "loss": 0.6443,
+      "step": 31480
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005365131998469584,
+      "loss": 0.6068,
+      "step": 31490
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005364281766781448,
+      "loss": 0.5664,
+      "step": 31500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0005363431535093313,
+      "loss": 0.5431,
+      "step": 31510
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005362581303405179,
+      "loss": 0.5688,
+      "step": 31520
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005361731071717043,
+      "loss": 0.5362,
+      "step": 31530
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005360880840028908,
+      "loss": 0.5464,
+      "step": 31540
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005360030608340773,
+      "loss": 0.5975,
+      "step": 31550
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005359180376652638,
+      "loss": 0.582,
+      "step": 31560
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005358330144964504,
+      "loss": 0.5673,
+      "step": 31570
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005357479913276368,
+      "loss": 0.5448,
+      "step": 31580
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005356629681588233,
+      "loss": 0.5639,
+      "step": 31590
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005355779449900098,
+      "loss": 0.557,
+      "step": 31600
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005354929218211962,
+      "loss": 0.6886,
+      "step": 31610
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005354078986523828,
+      "loss": 0.6024,
+      "step": 31620
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005353228754835693,
+      "loss": 0.5148,
+      "step": 31630
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005352378523147558,
+      "loss": 0.5836,
+      "step": 31640
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005351528291459423,
+      "loss": 0.504,
+      "step": 31650
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005350678059771288,
+      "loss": 0.6442,
+      "step": 31660
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005349827828083153,
+      "loss": 0.5076,
+      "step": 31670
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005348977596395018,
+      "loss": 0.571,
+      "step": 31680
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005348127364706882,
+      "loss": 0.5807,
+      "step": 31690
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005347277133018747,
+      "loss": 0.5682,
+      "step": 31700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005346426901330614,
+      "loss": 0.5402,
+      "step": 31710
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005345576669642478,
+      "loss": 0.5603,
+      "step": 31720
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005344726437954343,
+      "loss": 0.5495,
+      "step": 31730
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005343876206266207,
+      "loss": 0.5619,
+      "step": 31740
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005343025974578072,
+      "loss": 0.6579,
+      "step": 31750
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005342175742889938,
+      "loss": 0.558,
+      "step": 31760
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005341325511201802,
+      "loss": 0.6289,
+      "step": 31770
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005340475279513668,
+      "loss": 0.5721,
+      "step": 31780
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005339625047825533,
+      "loss": 0.5816,
+      "step": 31790
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005338774816137398,
+      "loss": 0.5629,
+      "step": 31800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005337924584449263,
+      "loss": 0.6387,
+      "step": 31810
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005337074352761128,
+      "loss": 0.6487,
+      "step": 31820
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005336224121072992,
+      "loss": 0.5711,
+      "step": 31830
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005335373889384857,
+      "loss": 0.4914,
+      "step": 31840
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005334523657696723,
+      "loss": 0.6288,
+      "step": 31850
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0005333673426008588,
+      "loss": 0.63,
+      "step": 31860
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005332823194320453,
+      "loss": 0.5873,
+      "step": 31870
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005331972962632317,
+      "loss": 0.5759,
+      "step": 31880
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005331122730944183,
+      "loss": 0.5808,
+      "step": 31890
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005330272499256048,
+      "loss": 0.6273,
+      "step": 31900
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005329422267567912,
+      "loss": 0.5536,
+      "step": 31910
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005328572035879778,
+      "loss": 0.606,
+      "step": 31920
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005327721804191642,
+      "loss": 0.599,
+      "step": 31930
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005326871572503508,
+      "loss": 0.5635,
+      "step": 31940
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005326021340815373,
+      "loss": 0.5995,
+      "step": 31950
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005325171109127237,
+      "loss": 0.527,
+      "step": 31960
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005324320877439102,
+      "loss": 0.5495,
+      "step": 31970
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005323470645750968,
+      "loss": 0.6285,
+      "step": 31980
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005322620414062832,
+      "loss": 0.5957,
+      "step": 31990
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005321770182374698,
+      "loss": 0.6044,
+      "step": 32000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005320919950686562,
+      "loss": 0.5307,
+      "step": 32010
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005320069718998427,
+      "loss": 0.6451,
+      "step": 32020
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005319219487310293,
+      "loss": 0.67,
+      "step": 32030
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005318369255622157,
+      "loss": 0.6291,
+      "step": 32040
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005317519023934022,
+      "loss": 0.5575,
+      "step": 32050
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005316668792245887,
+      "loss": 0.5761,
+      "step": 32060
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005315818560557752,
+      "loss": 0.5518,
+      "step": 32070
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005314968328869618,
+      "loss": 0.6267,
+      "step": 32080
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005314118097181482,
+      "loss": 0.6,
+      "step": 32090
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0005313267865493347,
+      "loss": 0.5779,
+      "step": 32100
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005312417633805212,
+      "loss": 0.5768,
+      "step": 32110
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005311567402117077,
+      "loss": 0.6149,
+      "step": 32120
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005310717170428942,
+      "loss": 0.5841,
+      "step": 32130
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005309866938740808,
+      "loss": 0.6421,
+      "step": 32140
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005309016707052672,
+      "loss": 0.5639,
+      "step": 32150
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005308166475364537,
+      "loss": 0.5814,
+      "step": 32160
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005307316243676402,
+      "loss": 0.5692,
+      "step": 32170
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005306466011988267,
+      "loss": 0.5506,
+      "step": 32180
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005305615780300132,
+      "loss": 0.6015,
+      "step": 32190
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005304765548611996,
+      "loss": 0.5785,
+      "step": 32200
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005303915316923861,
+      "loss": 0.606,
+      "step": 32210
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005303065085235728,
+      "loss": 0.5942,
+      "step": 32220
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005302214853547592,
+      "loss": 0.6145,
+      "step": 32230
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005301364621859457,
+      "loss": 0.5734,
+      "step": 32240
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005300514390171322,
+      "loss": 0.5569,
+      "step": 32250
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005299664158483187,
+      "loss": 0.5675,
+      "step": 32260
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005298813926795052,
+      "loss": 0.598,
+      "step": 32270
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005297963695106916,
+      "loss": 0.5995,
+      "step": 32280
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005297113463418782,
+      "loss": 0.6014,
+      "step": 32290
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005296263231730648,
+      "loss": 0.5664,
+      "step": 32300
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005295413000042512,
+      "loss": 0.5824,
+      "step": 32310
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005294562768354377,
+      "loss": 0.5497,
+      "step": 32320
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0005293712536666242,
+      "loss": 0.6011,
+      "step": 32330
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005292862304978106,
+      "loss": 0.6264,
+      "step": 32340
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005292012073289972,
+      "loss": 0.5397,
+      "step": 32350
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005291161841601837,
+      "loss": 0.5841,
+      "step": 32360
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005290311609913702,
+      "loss": 0.5956,
+      "step": 32370
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005289461378225567,
+      "loss": 0.6264,
+      "step": 32380
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005288611146537431,
+      "loss": 0.6513,
+      "step": 32390
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005287760914849297,
+      "loss": 0.6146,
+      "step": 32400
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005286910683161162,
+      "loss": 0.5363,
+      "step": 32410
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005286060451473026,
+      "loss": 0.6175,
+      "step": 32420
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005285210219784892,
+      "loss": 0.5812,
+      "step": 32430
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005284359988096757,
+      "loss": 0.6231,
+      "step": 32440
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005283509756408622,
+      "loss": 0.54,
+      "step": 32450
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005282659524720487,
+      "loss": 0.5916,
+      "step": 32460
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005281809293032351,
+      "loss": 0.6195,
+      "step": 32470
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005280959061344216,
+      "loss": 0.6131,
+      "step": 32480
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005280108829656082,
+      "loss": 0.656,
+      "step": 32490
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005279258597967946,
+      "loss": 0.6014,
+      "step": 32500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005278408366279812,
+      "loss": 0.5509,
+      "step": 32510
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005277558134591676,
+      "loss": 0.601,
+      "step": 32520
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005276707902903541,
+      "loss": 0.5599,
+      "step": 32530
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005275857671215407,
+      "loss": 0.5492,
+      "step": 32540
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005275007439527271,
+      "loss": 0.5613,
+      "step": 32550
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005274157207839136,
+      "loss": 0.5833,
+      "step": 32560
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0005273306976151001,
+      "loss": 0.5052,
+      "step": 32570
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005272456744462867,
+      "loss": 0.624,
+      "step": 32580
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005271606512774732,
+      "loss": 0.5701,
+      "step": 32590
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005270756281086596,
+      "loss": 0.6066,
+      "step": 32600
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005269906049398461,
+      "loss": 0.5969,
+      "step": 32610
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005269055817710326,
+      "loss": 0.5519,
+      "step": 32620
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005268205586022191,
+      "loss": 0.6368,
+      "step": 32630
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005267355354334056,
+      "loss": 0.5457,
+      "step": 32640
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005266505122645922,
+      "loss": 0.5546,
+      "step": 32650
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005265654890957786,
+      "loss": 0.5646,
+      "step": 32660
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005264804659269652,
+      "loss": 0.555,
+      "step": 32670
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005263954427581516,
+      "loss": 0.5959,
+      "step": 32680
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005263104195893381,
+      "loss": 0.6747,
+      "step": 32690
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005262253964205246,
+      "loss": 0.5533,
+      "step": 32700
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.000526140373251711,
+      "loss": 0.5992,
+      "step": 32710
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005260553500828977,
+      "loss": 0.5688,
+      "step": 32720
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005259703269140842,
+      "loss": 0.5552,
+      "step": 32730
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005258853037452706,
+      "loss": 0.5848,
+      "step": 32740
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005258002805764571,
+      "loss": 0.6228,
+      "step": 32750
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005257152574076437,
+      "loss": 0.6016,
+      "step": 32760
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005256302342388301,
+      "loss": 0.5513,
+      "step": 32770
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005255452110700166,
+      "loss": 0.5863,
+      "step": 32780
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.000525460187901203,
+      "loss": 0.5528,
+      "step": 32790
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005253751647323896,
+      "loss": 0.5657,
+      "step": 32800
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005252901415635762,
+      "loss": 0.6199,
+      "step": 32810
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005252051183947626,
+      "loss": 0.6228,
+      "step": 32820
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005251200952259491,
+      "loss": 0.6029,
+      "step": 32830
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005250350720571356,
+      "loss": 0.5947,
+      "step": 32840
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.000524950048888322,
+      "loss": 0.4873,
+      "step": 32850
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005248650257195086,
+      "loss": 0.6219,
+      "step": 32860
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005247800025506951,
+      "loss": 0.5916,
+      "step": 32870
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005246949793818816,
+      "loss": 0.5039,
+      "step": 32880
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005246099562130681,
+      "loss": 0.6043,
+      "step": 32890
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005245249330442546,
+      "loss": 0.695,
+      "step": 32900
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005244399098754411,
+      "loss": 0.5936,
+      "step": 32910
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0005243548867066276,
+      "loss": 0.6035,
+      "step": 32920
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.000524269863537814,
+      "loss": 0.5293,
+      "step": 32930
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005241848403690006,
+      "loss": 0.6991,
+      "step": 32940
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005240998172001871,
+      "loss": 0.5335,
+      "step": 32950
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005240147940313736,
+      "loss": 0.6077,
+      "step": 32960
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005239297708625601,
+      "loss": 0.5829,
+      "step": 32970
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005238447476937465,
+      "loss": 0.5992,
+      "step": 32980
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.000523759724524933,
+      "loss": 0.5356,
+      "step": 32990
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005236747013561196,
+      "loss": 0.6286,
+      "step": 33000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005235896781873061,
+      "loss": 0.5443,
+      "step": 33010
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005235046550184926,
+      "loss": 0.5966,
+      "step": 33020
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.000523419631849679,
+      "loss": 0.5987,
+      "step": 33030
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005233346086808656,
+      "loss": 0.6477,
+      "step": 33040
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005232495855120521,
+      "loss": 0.6186,
+      "step": 33050
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005231645623432385,
+      "loss": 0.5772,
+      "step": 33060
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.000523079539174425,
+      "loss": 0.5402,
+      "step": 33070
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005229945160056115,
+      "loss": 0.5725,
+      "step": 33080
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005229094928367981,
+      "loss": 0.5971,
+      "step": 33090
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005228244696679846,
+      "loss": 0.599,
+      "step": 33100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.000522739446499171,
+      "loss": 0.5771,
+      "step": 33110
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005226544233303575,
+      "loss": 0.5545,
+      "step": 33120
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005225694001615441,
+      "loss": 0.5817,
+      "step": 33130
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005224843769927305,
+      "loss": 0.5953,
+      "step": 33140
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.000522399353823917,
+      "loss": 0.6006,
+      "step": 33150
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0005223143306551036,
+      "loss": 0.6513,
+      "step": 33160
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00052222930748629,
+      "loss": 0.6858,
+      "step": 33170
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005221442843174766,
+      "loss": 0.609,
+      "step": 33180
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005220592611486631,
+      "loss": 0.625,
+      "step": 33190
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005219742379798495,
+      "loss": 0.539,
+      "step": 33200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.000521889214811036,
+      "loss": 0.5673,
+      "step": 33210
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005218041916422224,
+      "loss": 0.4888,
+      "step": 33220
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005217191684734091,
+      "loss": 0.5731,
+      "step": 33230
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005216341453045956,
+      "loss": 0.597,
+      "step": 33240
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.000521549122135782,
+      "loss": 0.5944,
+      "step": 33250
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005214640989669685,
+      "loss": 0.5949,
+      "step": 33260
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005213790757981551,
+      "loss": 0.5902,
+      "step": 33270
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005212940526293415,
+      "loss": 0.5977,
+      "step": 33280
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.000521209029460528,
+      "loss": 0.5548,
+      "step": 33290
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005211240062917145,
+      "loss": 0.5623,
+      "step": 33300
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.000521038983122901,
+      "loss": 0.5805,
+      "step": 33310
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005209539599540876,
+      "loss": 0.6504,
+      "step": 33320
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.000520868936785274,
+      "loss": 0.5231,
+      "step": 33330
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005207839136164605,
+      "loss": 0.6729,
+      "step": 33340
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.000520698890447647,
+      "loss": 0.5728,
+      "step": 33350
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005206138672788335,
+      "loss": 0.5483,
+      "step": 33360
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00052052884411002,
+      "loss": 0.6245,
+      "step": 33370
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005204438209412065,
+      "loss": 0.6113,
+      "step": 33380
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.000520358797772393,
+      "loss": 0.5434,
+      "step": 33390
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005202737746035795,
+      "loss": 0.6608,
+      "step": 33400
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.000520188751434766,
+      "loss": 0.5318,
+      "step": 33410
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005201037282659525,
+      "loss": 0.5534,
+      "step": 33420
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.000520018705097139,
+      "loss": 0.6291,
+      "step": 33430
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005199336819283254,
+      "loss": 0.6257,
+      "step": 33440
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005198486587595121,
+      "loss": 0.6005,
+      "step": 33450
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005197636355906985,
+      "loss": 0.5679,
+      "step": 33460
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.000519678612421885,
+      "loss": 0.5997,
+      "step": 33470
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005195935892530715,
+      "loss": 0.6207,
+      "step": 33480
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005195085660842579,
+      "loss": 0.5806,
+      "step": 33490
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005194235429154445,
+      "loss": 0.5413,
+      "step": 33500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.000519338519746631,
+      "loss": 0.5675,
+      "step": 33510
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005192534965778175,
+      "loss": 0.5578,
+      "step": 33520
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.000519168473409004,
+      "loss": 0.5933,
+      "step": 33530
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005190834502401904,
+      "loss": 0.5745,
+      "step": 33540
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.000518998427071377,
+      "loss": 0.5578,
+      "step": 33550
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005189134039025635,
+      "loss": 0.5617,
+      "step": 33560
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005188283807337499,
+      "loss": 0.5717,
+      "step": 33570
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005187433575649364,
+      "loss": 0.6271,
+      "step": 33580
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.000518658334396123,
+      "loss": 0.6001,
+      "step": 33590
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005185733112273095,
+      "loss": 0.5133,
+      "step": 33600
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.000518488288058496,
+      "loss": 0.5847,
+      "step": 33610
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005184032648896825,
+      "loss": 0.6078,
+      "step": 33620
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0005183182417208689,
+      "loss": 0.5788,
+      "step": 33630
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005182332185520555,
+      "loss": 0.5653,
+      "step": 33640
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005181481953832419,
+      "loss": 0.5724,
+      "step": 33650
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005180631722144284,
+      "loss": 0.5317,
+      "step": 33660
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000517978149045615,
+      "loss": 0.5626,
+      "step": 33670
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005178931258768015,
+      "loss": 0.5963,
+      "step": 33680
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000517808102707988,
+      "loss": 0.6127,
+      "step": 33690
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005177230795391745,
+      "loss": 0.5781,
+      "step": 33700
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005176380563703609,
+      "loss": 0.5951,
+      "step": 33710
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005175530332015474,
+      "loss": 0.5334,
+      "step": 33720
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000517468010032734,
+      "loss": 0.5765,
+      "step": 33730
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005173829868639205,
+      "loss": 0.6039,
+      "step": 33740
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000517297963695107,
+      "loss": 0.5696,
+      "step": 33750
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005172129405262934,
+      "loss": 0.5856,
+      "step": 33760
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00051712791735748,
+      "loss": 0.4993,
+      "step": 33770
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005170428941886665,
+      "loss": 0.6118,
+      "step": 33780
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005169578710198529,
+      "loss": 0.5697,
+      "step": 33790
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005168728478510394,
+      "loss": 0.6069,
+      "step": 33800
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005167878246822259,
+      "loss": 0.5848,
+      "step": 33810
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005167028015134125,
+      "loss": 0.4706,
+      "step": 33820
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.000516617778344599,
+      "loss": 0.5725,
+      "step": 33830
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005165327551757854,
+      "loss": 0.5517,
+      "step": 33840
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005164477320069719,
+      "loss": 0.6172,
+      "step": 33850
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005163627088381584,
+      "loss": 0.5352,
+      "step": 33860
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0005162776856693449,
+      "loss": 0.5198,
+      "step": 33870
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005161926625005314,
+      "loss": 0.5811,
+      "step": 33880
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005161076393317179,
+      "loss": 0.5347,
+      "step": 33890
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005160226161629044,
+      "loss": 0.6308,
+      "step": 33900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.000515937592994091,
+      "loss": 0.562,
+      "step": 33910
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005158525698252774,
+      "loss": 0.6204,
+      "step": 33920
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005157675466564639,
+      "loss": 0.5973,
+      "step": 33930
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005156825234876504,
+      "loss": 0.62,
+      "step": 33940
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005155975003188368,
+      "loss": 0.6064,
+      "step": 33950
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005155124771500235,
+      "loss": 0.5479,
+      "step": 33960
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005154274539812099,
+      "loss": 0.6152,
+      "step": 33970
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005153424308123964,
+      "loss": 0.5326,
+      "step": 33980
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0005152574076435829,
+      "loss": 0.5387,
+      "step": 33990
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005151723844747693,
+      "loss": 0.5632,
+      "step": 34000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005150873613059559,
+      "loss": 0.5844,
+      "step": 34010
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005150023381371424,
+      "loss": 0.5358,
+      "step": 34020
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005149173149683289,
+      "loss": 0.599,
+      "step": 34030
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005148322917995154,
+      "loss": 0.5822,
+      "step": 34040
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005147472686307019,
+      "loss": 0.5872,
+      "step": 34050
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005146622454618884,
+      "loss": 0.6023,
+      "step": 34060
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005145772222930749,
+      "loss": 0.5856,
+      "step": 34070
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005144921991242613,
+      "loss": 0.5829,
+      "step": 34080
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005144071759554478,
+      "loss": 0.5928,
+      "step": 34090
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005143221527866345,
+      "loss": 0.5845,
+      "step": 34100
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005142371296178209,
+      "loss": 0.5709,
+      "step": 34110
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005141521064490074,
+      "loss": 0.5861,
+      "step": 34120
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005140670832801939,
+      "loss": 0.6042,
+      "step": 34130
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005139820601113804,
+      "loss": 0.5582,
+      "step": 34140
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005138970369425669,
+      "loss": 0.5752,
+      "step": 34150
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005138120137737533,
+      "loss": 0.6232,
+      "step": 34160
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005137269906049398,
+      "loss": 0.5861,
+      "step": 34170
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005136419674361264,
+      "loss": 0.6184,
+      "step": 34180
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005135569442673129,
+      "loss": 0.5981,
+      "step": 34190
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005134719210984994,
+      "loss": 0.617,
+      "step": 34200
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005133868979296859,
+      "loss": 0.5705,
+      "step": 34210
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005133018747608723,
+      "loss": 0.5307,
+      "step": 34220
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0005132168515920588,
+      "loss": 0.6538,
+      "step": 34230
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005131318284232454,
+      "loss": 0.6033,
+      "step": 34240
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005130468052544319,
+      "loss": 0.5725,
+      "step": 34250
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005129617820856184,
+      "loss": 0.5939,
+      "step": 34260
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005128767589168048,
+      "loss": 0.6041,
+      "step": 34270
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005127917357479914,
+      "loss": 0.6197,
+      "step": 34280
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005127067125791779,
+      "loss": 0.6248,
+      "step": 34290
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005126216894103643,
+      "loss": 0.608,
+      "step": 34300
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005125366662415508,
+      "loss": 0.556,
+      "step": 34310
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005124516430727373,
+      "loss": 0.62,
+      "step": 34320
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005123666199039239,
+      "loss": 0.5377,
+      "step": 34330
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005122815967351104,
+      "loss": 0.6057,
+      "step": 34340
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005121965735662968,
+      "loss": 0.5754,
+      "step": 34350
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005121115503974833,
+      "loss": 0.475,
+      "step": 34360
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005120265272286699,
+      "loss": 0.5928,
+      "step": 34370
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005119415040598563,
+      "loss": 0.5912,
+      "step": 34380
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005118564808910429,
+      "loss": 0.5788,
+      "step": 34390
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005117714577222293,
+      "loss": 0.5717,
+      "step": 34400
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005116864345534158,
+      "loss": 0.5793,
+      "step": 34410
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005116014113846024,
+      "loss": 0.5801,
+      "step": 34420
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005115163882157888,
+      "loss": 0.5761,
+      "step": 34430
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005114313650469753,
+      "loss": 0.5247,
+      "step": 34440
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005113463418781618,
+      "loss": 0.6307,
+      "step": 34450
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005112613187093482,
+      "loss": 0.5259,
+      "step": 34460
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005111762955405349,
+      "loss": 0.6529,
+      "step": 34470
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005110912723717213,
+      "loss": 0.6077,
+      "step": 34480
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005110062492029078,
+      "loss": 0.6149,
+      "step": 34490
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005109212260340943,
+      "loss": 0.5439,
+      "step": 34500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005108362028652808,
+      "loss": 0.6103,
+      "step": 34510
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005107511796964673,
+      "loss": 0.5687,
+      "step": 34520
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005106661565276538,
+      "loss": 0.6153,
+      "step": 34530
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005105811333588403,
+      "loss": 0.6182,
+      "step": 34540
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005104961101900268,
+      "loss": 0.559,
+      "step": 34550
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005104110870212134,
+      "loss": 0.6478,
+      "step": 34560
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005103260638523998,
+      "loss": 0.5906,
+      "step": 34570
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005102410406835863,
+      "loss": 0.595,
+      "step": 34580
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005101560175147727,
+      "loss": 0.5359,
+      "step": 34590
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005100709943459592,
+      "loss": 0.6037,
+      "step": 34600
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005099859711771459,
+      "loss": 0.5979,
+      "step": 34610
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005099009480083323,
+      "loss": 0.5233,
+      "step": 34620
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005098159248395188,
+      "loss": 0.6074,
+      "step": 34630
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005097309016707053,
+      "loss": 0.6212,
+      "step": 34640
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005096458785018918,
+      "loss": 0.6237,
+      "step": 34650
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005095608553330783,
+      "loss": 0.5803,
+      "step": 34660
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005094758321642648,
+      "loss": 0.5583,
+      "step": 34670
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005093908089954512,
+      "loss": 0.6309,
+      "step": 34680
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005093057858266379,
+      "loss": 0.6031,
+      "step": 34690
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0005092207626578243,
+      "loss": 0.5708,
+      "step": 34700
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005091357394890108,
+      "loss": 0.564,
+      "step": 34710
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005090507163201973,
+      "loss": 0.6229,
+      "step": 34720
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005089656931513837,
+      "loss": 0.5586,
+      "step": 34730
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005088806699825703,
+      "loss": 0.6075,
+      "step": 34740
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005087956468137568,
+      "loss": 0.6275,
+      "step": 34750
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005087106236449433,
+      "loss": 0.5637,
+      "step": 34760
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005086256004761298,
+      "loss": 0.5825,
+      "step": 34770
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005085405773073162,
+      "loss": 0.5954,
+      "step": 34780
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005084555541385028,
+      "loss": 0.553,
+      "step": 34790
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005083705309696893,
+      "loss": 0.5919,
+      "step": 34800
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005082855078008757,
+      "loss": 0.5573,
+      "step": 34810
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005082004846320622,
+      "loss": 0.6519,
+      "step": 34820
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005081154614632488,
+      "loss": 0.5741,
+      "step": 34830
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005080304382944353,
+      "loss": 0.6397,
+      "step": 34840
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005079454151256218,
+      "loss": 0.5642,
+      "step": 34850
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005078603919568082,
+      "loss": 0.6078,
+      "step": 34860
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005077753687879947,
+      "loss": 0.5482,
+      "step": 34870
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005076903456191813,
+      "loss": 0.6062,
+      "step": 34880
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005076053224503677,
+      "loss": 0.5511,
+      "step": 34890
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005075202992815543,
+      "loss": 0.6131,
+      "step": 34900
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005074352761127407,
+      "loss": 0.4986,
+      "step": 34910
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005073502529439273,
+      "loss": 0.5686,
+      "step": 34920
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0005072652297751138,
+      "loss": 0.506,
+      "step": 34930
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005071802066063002,
+      "loss": 0.5131,
+      "step": 34940
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005070951834374867,
+      "loss": 0.5346,
+      "step": 34950
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005070101602686732,
+      "loss": 0.569,
+      "step": 34960
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005069251370998597,
+      "loss": 0.5762,
+      "step": 34970
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005068401139310463,
+      "loss": 0.6398,
+      "step": 34980
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005067550907622327,
+      "loss": 0.582,
+      "step": 34990
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005066700675934192,
+      "loss": 0.5598,
+      "step": 35000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005065850444246057,
+      "loss": 0.5554,
+      "step": 35010
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005065000212557922,
+      "loss": 0.5502,
+      "step": 35020
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005064149980869787,
+      "loss": 0.5825,
+      "step": 35030
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005063299749181652,
+      "loss": 0.6184,
+      "step": 35040
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0005062449517493517,
+      "loss": 0.5544,
+      "step": 35050
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005061599285805383,
+      "loss": 0.6724,
+      "step": 35060
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005060749054117248,
+      "loss": 0.6183,
+      "step": 35070
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005059898822429112,
+      "loss": 0.5755,
+      "step": 35080
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005059048590740977,
+      "loss": 0.5872,
+      "step": 35090
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005058198359052841,
+      "loss": 0.5648,
+      "step": 35100
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005057348127364707,
+      "loss": 0.5741,
+      "step": 35110
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005056497895676573,
+      "loss": 0.6296,
+      "step": 35120
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005055647663988437,
+      "loss": 0.6126,
+      "step": 35130
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005054797432300302,
+      "loss": 0.5562,
+      "step": 35140
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005053947200612168,
+      "loss": 0.598,
+      "step": 35150
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005053096968924032,
+      "loss": 0.6534,
+      "step": 35160
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005052246737235897,
+      "loss": 0.5369,
+      "step": 35170
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005051396505547762,
+      "loss": 0.5257,
+      "step": 35180
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005050546273859627,
+      "loss": 0.6057,
+      "step": 35190
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005049696042171493,
+      "loss": 0.5858,
+      "step": 35200
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005048845810483357,
+      "loss": 0.605,
+      "step": 35210
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005047995578795222,
+      "loss": 0.5536,
+      "step": 35220
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005047145347107087,
+      "loss": 0.6038,
+      "step": 35230
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005046295115418951,
+      "loss": 0.5898,
+      "step": 35240
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005045444883730817,
+      "loss": 0.5584,
+      "step": 35250
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005044594652042682,
+      "loss": 0.6387,
+      "step": 35260
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005043744420354547,
+      "loss": 0.5889,
+      "step": 35270
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005042894188666412,
+      "loss": 0.5761,
+      "step": 35280
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0005042043956978277,
+      "loss": 0.6452,
+      "step": 35290
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005041193725290142,
+      "loss": 0.6301,
+      "step": 35300
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005040343493602007,
+      "loss": 0.6529,
+      "step": 35310
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005039493261913871,
+      "loss": 0.5273,
+      "step": 35320
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005038643030225736,
+      "loss": 0.5794,
+      "step": 35330
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005037792798537602,
+      "loss": 0.562,
+      "step": 35340
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005036942566849467,
+      "loss": 0.6273,
+      "step": 35350
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005036092335161332,
+      "loss": 0.5927,
+      "step": 35360
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005035242103473196,
+      "loss": 0.6171,
+      "step": 35370
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005034391871785061,
+      "loss": 0.5679,
+      "step": 35380
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005033541640096927,
+      "loss": 0.5428,
+      "step": 35390
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005032691408408791,
+      "loss": 0.5248,
+      "step": 35400
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005031841176720657,
+      "loss": 0.5848,
+      "step": 35410
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005030990945032521,
+      "loss": 0.5954,
+      "step": 35420
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005030140713344387,
+      "loss": 0.5442,
+      "step": 35430
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005029290481656252,
+      "loss": 0.6049,
+      "step": 35440
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005028440249968116,
+      "loss": 0.5756,
+      "step": 35450
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005027590018279981,
+      "loss": 0.5512,
+      "step": 35460
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005026739786591846,
+      "loss": 0.6241,
+      "step": 35470
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005025889554903712,
+      "loss": 0.6122,
+      "step": 35480
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005025039323215577,
+      "loss": 0.5625,
+      "step": 35490
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005024189091527442,
+      "loss": 0.5433,
+      "step": 35500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005023338859839306,
+      "loss": 0.5254,
+      "step": 35510
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005022488628151172,
+      "loss": 0.5263,
+      "step": 35520
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005021638396463036,
+      "loss": 0.5386,
+      "step": 35530
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005020788164774901,
+      "loss": 0.5371,
+      "step": 35540
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005019937933086766,
+      "loss": 0.567,
+      "step": 35550
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005019087701398631,
+      "loss": 0.5123,
+      "step": 35560
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005018237469710497,
+      "loss": 0.5576,
+      "step": 35570
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005017387238022362,
+      "loss": 0.5431,
+      "step": 35580
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005016537006334226,
+      "loss": 0.4914,
+      "step": 35590
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005015686774646091,
+      "loss": 0.5304,
+      "step": 35600
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005014836542957957,
+      "loss": 0.5033,
+      "step": 35610
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005013986311269821,
+      "loss": 0.5362,
+      "step": 35620
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005013136079581687,
+      "loss": 0.5405,
+      "step": 35630
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005012285847893551,
+      "loss": 0.5318,
+      "step": 35640
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0005011435616205416,
+      "loss": 0.5392,
+      "step": 35650
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0005010585384517282,
+      "loss": 0.6021,
+      "step": 35660
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0005009735152829146,
+      "loss": 0.5378,
+      "step": 35670
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0005008884921141011,
+      "loss": 0.5024,
+      "step": 35680
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0005008034689452876,
+      "loss": 0.5644,
+      "step": 35690
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0005007184457764742,
+      "loss": 0.5629,
+      "step": 35700
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0005006334226076607,
+      "loss": 0.5827,
+      "step": 35710
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0005005483994388471,
+      "loss": 0.6501,
+      "step": 35720
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0005004633762700336,
+      "loss": 0.532,
+      "step": 35730
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0005003783531012201,
+      "loss": 0.4894,
+      "step": 35740
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0005002933299324066,
+      "loss": 0.4958,
+      "step": 35750
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0005002083067635931,
+      "loss": 0.5682,
+      "step": 35760
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0005001232835947796,
+      "loss": 0.5716,
+      "step": 35770
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0005000382604259661,
+      "loss": 0.5139,
+      "step": 35780
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004999532372571526,
+      "loss": 0.5502,
+      "step": 35790
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004998682140883391,
+      "loss": 0.5897,
+      "step": 35800
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004997831909195256,
+      "loss": 0.5634,
+      "step": 35810
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004996981677507121,
+      "loss": 0.5658,
+      "step": 35820
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004996131445818985,
+      "loss": 0.5512,
+      "step": 35830
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.000499528121413085,
+      "loss": 0.5675,
+      "step": 35840
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004994430982442716,
+      "loss": 0.5931,
+      "step": 35850
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004993580750754581,
+      "loss": 0.543,
+      "step": 35860
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0004992730519066446,
+      "loss": 0.5616,
+      "step": 35870
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.000499188028737831,
+      "loss": 0.5247,
+      "step": 35880
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004991030055690176,
+      "loss": 0.5606,
+      "step": 35890
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004990179824002041,
+      "loss": 0.5314,
+      "step": 35900
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004989329592313905,
+      "loss": 0.5572,
+      "step": 35910
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004988479360625771,
+      "loss": 0.5784,
+      "step": 35920
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004987629128937637,
+      "loss": 0.6385,
+      "step": 35930
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004986778897249501,
+      "loss": 0.5736,
+      "step": 35940
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004985928665561366,
+      "loss": 0.4909,
+      "step": 35950
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.000498507843387323,
+      "loss": 0.6082,
+      "step": 35960
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004984228202185095,
+      "loss": 0.5706,
+      "step": 35970
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004983377970496961,
+      "loss": 0.575,
+      "step": 35980
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004982527738808826,
+      "loss": 0.5017,
+      "step": 35990
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0004981677507120691,
+      "loss": 0.5778,
+      "step": 36000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004980827275432556,
+      "loss": 0.5503,
+      "step": 36010
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.000497997704374442,
+      "loss": 0.5048,
+      "step": 36020
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004979126812056286,
+      "loss": 0.5527,
+      "step": 36030
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004978276580368151,
+      "loss": 0.561,
+      "step": 36040
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004977426348680015,
+      "loss": 0.6699,
+      "step": 36050
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.000497657611699188,
+      "loss": 0.6163,
+      "step": 36060
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004975725885303746,
+      "loss": 0.5328,
+      "step": 36070
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004974875653615611,
+      "loss": 0.5611,
+      "step": 36080
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004974025421927476,
+      "loss": 0.561,
+      "step": 36090
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.000497317519023934,
+      "loss": 0.5241,
+      "step": 36100
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004972324958551205,
+      "loss": 0.513,
+      "step": 36110
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0004971474726863071,
+      "loss": 0.6017,
+      "step": 36120
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004970624495174935,
+      "loss": 0.5103,
+      "step": 36130
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004969774263486801,
+      "loss": 0.5429,
+      "step": 36140
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004968924031798665,
+      "loss": 0.5858,
+      "step": 36150
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.000496807380011053,
+      "loss": 0.5673,
+      "step": 36160
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004967223568422396,
+      "loss": 0.5846,
+      "step": 36170
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.000496637333673426,
+      "loss": 0.5408,
+      "step": 36180
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004965523105046125,
+      "loss": 0.5722,
+      "step": 36190
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.000496467287335799,
+      "loss": 0.5637,
+      "step": 36200
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004963822641669856,
+      "loss": 0.543,
+      "step": 36210
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004962972409981721,
+      "loss": 0.5284,
+      "step": 36220
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0004962122178293585,
+      "loss": 0.5595,
+      "step": 36230
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.000496127194660545,
+      "loss": 0.5536,
+      "step": 36240
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004960421714917315,
+      "loss": 0.5995,
+      "step": 36250
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.000495957148322918,
+      "loss": 0.5461,
+      "step": 36260
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004958721251541045,
+      "loss": 0.5957,
+      "step": 36270
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.000495787101985291,
+      "loss": 0.5764,
+      "step": 36280
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004957020788164775,
+      "loss": 0.5585,
+      "step": 36290
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004956170556476641,
+      "loss": 0.5723,
+      "step": 36300
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004955320324788505,
+      "loss": 0.5398,
+      "step": 36310
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.000495447009310037,
+      "loss": 0.6201,
+      "step": 36320
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004953619861412235,
+      "loss": 0.5426,
+      "step": 36330
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004952769629724099,
+      "loss": 0.5947,
+      "step": 36340
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0004951919398035965,
+      "loss": 0.5604,
+      "step": 36350
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.000495106916634783,
+      "loss": 0.5751,
+      "step": 36360
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004950218934659695,
+      "loss": 0.5894,
+      "step": 36370
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.000494936870297156,
+      "loss": 0.4944,
+      "step": 36380
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004948518471283424,
+      "loss": 0.5382,
+      "step": 36390
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.000494766823959529,
+      "loss": 0.5462,
+      "step": 36400
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004946818007907155,
+      "loss": 0.5411,
+      "step": 36410
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004945967776219019,
+      "loss": 0.5805,
+      "step": 36420
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004945117544530885,
+      "loss": 0.5002,
+      "step": 36430
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004944267312842751,
+      "loss": 0.5409,
+      "step": 36440
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004943417081154615,
+      "loss": 0.5098,
+      "step": 36450
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.000494256684946648,
+      "loss": 0.4989,
+      "step": 36460
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0004941716617778344,
+      "loss": 0.5645,
+      "step": 36470
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004940866386090209,
+      "loss": 0.5989,
+      "step": 36480
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004940016154402075,
+      "loss": 0.569,
+      "step": 36490
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.000493916592271394,
+      "loss": 0.5793,
+      "step": 36500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004938315691025805,
+      "loss": 0.5654,
+      "step": 36510
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.000493746545933767,
+      "loss": 0.5124,
+      "step": 36520
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004936615227649535,
+      "loss": 0.5033,
+      "step": 36530
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00049357649959614,
+      "loss": 0.636,
+      "step": 36540
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004934914764273265,
+      "loss": 0.5724,
+      "step": 36550
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004934064532585129,
+      "loss": 0.5866,
+      "step": 36560
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004933214300896995,
+      "loss": 0.5438,
+      "step": 36570
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.000493236406920886,
+      "loss": 0.5876,
+      "step": 36580
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0004931513837520725,
+      "loss": 0.5591,
+      "step": 36590
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.000493066360583259,
+      "loss": 0.5316,
+      "step": 36600
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004929813374144454,
+      "loss": 0.5467,
+      "step": 36610
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.000492896314245632,
+      "loss": 0.4878,
+      "step": 36620
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004928112910768185,
+      "loss": 0.5483,
+      "step": 36630
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004927262679080049,
+      "loss": 0.5115,
+      "step": 36640
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004926412447391915,
+      "loss": 0.4976,
+      "step": 36650
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004925562215703779,
+      "loss": 0.5328,
+      "step": 36660
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004924711984015645,
+      "loss": 0.5377,
+      "step": 36670
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.000492386175232751,
+      "loss": 0.5529,
+      "step": 36680
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004923011520639374,
+      "loss": 0.5596,
+      "step": 36690
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004922161288951239,
+      "loss": 0.5436,
+      "step": 36700
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0004921311057263104,
+      "loss": 0.5102,
+      "step": 36710
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.000492046082557497,
+      "loss": 0.573,
+      "step": 36720
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004919610593886835,
+      "loss": 0.5208,
+      "step": 36730
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004918760362198699,
+      "loss": 0.527,
+      "step": 36740
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004917910130510564,
+      "loss": 0.5151,
+      "step": 36750
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.000491705989882243,
+      "loss": 0.5273,
+      "step": 36760
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004916209667134294,
+      "loss": 0.5563,
+      "step": 36770
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004915359435446159,
+      "loss": 0.5253,
+      "step": 36780
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004914509203758024,
+      "loss": 0.5428,
+      "step": 36790
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004913658972069889,
+      "loss": 0.5543,
+      "step": 36800
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004912808740381755,
+      "loss": 0.5602,
+      "step": 36810
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004911958508693619,
+      "loss": 0.5926,
+      "step": 36820
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0004911108277005484,
+      "loss": 0.5467,
+      "step": 36830
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004910258045317349,
+      "loss": 0.4709,
+      "step": 36840
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004909407813629213,
+      "loss": 0.5633,
+      "step": 36850
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.000490855758194108,
+      "loss": 0.5102,
+      "step": 36860
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004907707350252945,
+      "loss": 0.5556,
+      "step": 36870
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004906857118564809,
+      "loss": 0.5059,
+      "step": 36880
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004906006886876674,
+      "loss": 0.518,
+      "step": 36890
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004905156655188539,
+      "loss": 0.5012,
+      "step": 36900
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004904306423500404,
+      "loss": 0.5684,
+      "step": 36910
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004903456191812269,
+      "loss": 0.5449,
+      "step": 36920
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004902605960124133,
+      "loss": 0.494,
+      "step": 36930
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004901755728436,
+      "loss": 0.5747,
+      "step": 36940
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0004900905496747865,
+      "loss": 0.5168,
+      "step": 36950
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004900055265059729,
+      "loss": 0.5165,
+      "step": 36960
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004899205033371594,
+      "loss": 0.5204,
+      "step": 36970
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004898354801683459,
+      "loss": 0.5769,
+      "step": 36980
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004897504569995324,
+      "loss": 0.6196,
+      "step": 36990
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004896654338307189,
+      "loss": 0.5574,
+      "step": 37000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004895804106619054,
+      "loss": 0.5996,
+      "step": 37010
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004894953874930919,
+      "loss": 0.5934,
+      "step": 37020
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004894103643242784,
+      "loss": 0.596,
+      "step": 37030
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004893253411554649,
+      "loss": 0.5418,
+      "step": 37040
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004892403179866514,
+      "loss": 0.5598,
+      "step": 37050
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0004891552948178379,
+      "loss": 0.5441,
+      "step": 37060
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004890702716490243,
+      "loss": 0.5637,
+      "step": 37070
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.000488985248480211,
+      "loss": 0.5632,
+      "step": 37080
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004889002253113974,
+      "loss": 0.5377,
+      "step": 37090
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004888152021425839,
+      "loss": 0.594,
+      "step": 37100
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004887301789737704,
+      "loss": 0.5624,
+      "step": 37110
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004886451558049568,
+      "loss": 0.5677,
+      "step": 37120
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004885601326361434,
+      "loss": 0.5071,
+      "step": 37130
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004884751094673299,
+      "loss": 0.6116,
+      "step": 37140
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004883900862985163,
+      "loss": 0.5441,
+      "step": 37150
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004883050631297029,
+      "loss": 0.5558,
+      "step": 37160
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0004882200399608894,
+      "loss": 0.5911,
+      "step": 37170
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00048813501679207587,
+      "loss": 0.5695,
+      "step": 37180
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00048804999362326234,
+      "loss": 0.5506,
+      "step": 37190
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00048796497045444886,
+      "loss": 0.5944,
+      "step": 37200
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00048787994728563533,
+      "loss": 0.5233,
+      "step": 37210
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004877949241168218,
+      "loss": 0.5237,
+      "step": 37220
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004877099009480084,
+      "loss": 0.5924,
+      "step": 37230
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004876248777791949,
+      "loss": 0.5539,
+      "step": 37240
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00048753985461038137,
+      "loss": 0.5613,
+      "step": 37250
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00048745483144156784,
+      "loss": 0.5828,
+      "step": 37260
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004873698082727543,
+      "loss": 0.5994,
+      "step": 37270
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00048728478510394084,
+      "loss": 0.5944,
+      "step": 37280
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004871997619351273,
+      "loss": 0.5645,
+      "step": 37290
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0004871147387663139,
+      "loss": 0.5831,
+      "step": 37300
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048702971559750035,
+      "loss": 0.5436,
+      "step": 37310
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0004869446924286869,
+      "loss": 0.5575,
+      "step": 37320
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048685966925987335,
+      "loss": 0.5394,
+      "step": 37330
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0004867746460910598,
+      "loss": 0.5498,
+      "step": 37340
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048668962292224634,
+      "loss": 0.526,
+      "step": 37350
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0004866045997534328,
+      "loss": 0.5029,
+      "step": 37360
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0004865195765846194,
+      "loss": 0.5367,
+      "step": 37370
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048643455341580586,
+      "loss": 0.5196,
+      "step": 37380
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0004863495302469923,
+      "loss": 0.5361,
+      "step": 37390
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00048626450707817885,
+      "loss": 0.547,
+      "step": 37400
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0004861794839093653,
+      "loss": 0.5326,
+      "step": 37410
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0004860944607405518,
+      "loss": 0.5341,
+      "step": 37420
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004860094375717383,
+      "loss": 0.5024,
+      "step": 37430
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004859244144029248,
+      "loss": 0.4971,
+      "step": 37440
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00048583939123411136,
+      "loss": 0.5185,
+      "step": 37450
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00048575436806529783,
+      "loss": 0.5013,
+      "step": 37460
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004856693448964843,
+      "loss": 0.5583,
+      "step": 37470
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004855843217276708,
+      "loss": 0.5478,
+      "step": 37480
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004854992985588573,
+      "loss": 0.606,
+      "step": 37490
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00048541427539004376,
+      "loss": 0.6085,
+      "step": 37500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004853292522212303,
+      "loss": 0.5437,
+      "step": 37510
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00048524422905241686,
+      "loss": 0.5221,
+      "step": 37520
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00048515920588360333,
+      "loss": 0.5283,
+      "step": 37530
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0004850741827147898,
+      "loss": 0.5026,
+      "step": 37540
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004849891595459763,
+      "loss": 0.5223,
+      "step": 37550
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004849041363771628,
+      "loss": 0.5439,
+      "step": 37560
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00048481911320834926,
+      "loss": 0.6105,
+      "step": 37570
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00048473409003953573,
+      "loss": 0.5886,
+      "step": 37580
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004846490668707223,
+      "loss": 0.5964,
+      "step": 37590
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00048456404370190884,
+      "loss": 0.548,
+      "step": 37600
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004844790205330953,
+      "loss": 0.571,
+      "step": 37610
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004843939973642818,
+      "loss": 0.6031,
+      "step": 37620
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0004843089741954683,
+      "loss": 0.5725,
+      "step": 37630
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00048422395102665477,
+      "loss": 0.5036,
+      "step": 37640
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00048413892785784124,
+      "loss": 0.6371,
+      "step": 37650
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004840539046890278,
+      "loss": 0.5479,
+      "step": 37660
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004839688815202143,
+      "loss": 0.5038,
+      "step": 37670
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004838838583514008,
+      "loss": 0.5826,
+      "step": 37680
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004837988351825873,
+      "loss": 0.5716,
+      "step": 37690
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00048371381201377375,
+      "loss": 0.5571,
+      "step": 37700
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00048362878884496027,
+      "loss": 0.5423,
+      "step": 37710
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00048354376567614674,
+      "loss": 0.5478,
+      "step": 37720
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004834587425073332,
+      "loss": 0.5603,
+      "step": 37730
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004833737193385198,
+      "loss": 0.5422,
+      "step": 37740
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004832886961697063,
+      "loss": 0.6093,
+      "step": 37750
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0004832036730008928,
+      "loss": 0.5067,
+      "step": 37760
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00048311864983207925,
+      "loss": 0.5743,
+      "step": 37770
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004830336266632657,
+      "loss": 0.5997,
+      "step": 37780
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00048294860349445224,
+      "loss": 0.585,
+      "step": 37790
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004828635803256387,
+      "loss": 0.5485,
+      "step": 37800
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004827785571568253,
+      "loss": 0.549,
+      "step": 37810
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00048269353398801176,
+      "loss": 0.5698,
+      "step": 37820
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004826085108191983,
+      "loss": 0.5583,
+      "step": 37830
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00048252348765038476,
+      "loss": 0.5385,
+      "step": 37840
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004824384644815712,
+      "loss": 0.5901,
+      "step": 37850
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00048235344131275775,
+      "loss": 0.4879,
+      "step": 37860
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004822684181439442,
+      "loss": 0.5642,
+      "step": 37870
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0004821833949751308,
+      "loss": 0.5034,
+      "step": 37880
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00048209837180631727,
+      "loss": 0.5932,
+      "step": 37890
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00048201334863750374,
+      "loss": 0.6489,
+      "step": 37900
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00048192832546869026,
+      "loss": 0.5747,
+      "step": 37910
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00048184330229987673,
+      "loss": 0.5005,
+      "step": 37920
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0004817582791310632,
+      "loss": 0.5336,
+      "step": 37930
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0004816732559622497,
+      "loss": 0.6437,
+      "step": 37940
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0004815882327934363,
+      "loss": 0.5781,
+      "step": 37950
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00048150320962462277,
+      "loss": 0.538,
+      "step": 37960
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00048141818645580924,
+      "loss": 0.4495,
+      "step": 37970
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00048133316328699576,
+      "loss": 0.5054,
+      "step": 37980
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00048124814011818223,
+      "loss": 0.5341,
+      "step": 37990
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0004811631169493687,
+      "loss": 0.5845,
+      "step": 38000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00048107809378055517,
+      "loss": 0.5122,
+      "step": 38010
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004809930706117417,
+      "loss": 0.5798,
+      "step": 38020
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00048090804744292827,
+      "loss": 0.5384,
+      "step": 38030
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00048082302427411474,
+      "loss": 0.5897,
+      "step": 38040
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004807380011053012,
+      "loss": 0.4783,
+      "step": 38050
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00048065297793648774,
+      "loss": 0.5493,
+      "step": 38060
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004805679547676742,
+      "loss": 0.5024,
+      "step": 38070
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004804829315988607,
+      "loss": 0.6134,
+      "step": 38080
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004803979084300472,
+      "loss": 0.5194,
+      "step": 38090
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004803128852612337,
+      "loss": 0.5686,
+      "step": 38100
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00048022786209242025,
+      "loss": 0.5362,
+      "step": 38110
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004801428389236067,
+      "loss": 0.5097,
+      "step": 38120
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0004800578157547932,
+      "loss": 0.5213,
+      "step": 38130
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004799727925859797,
+      "loss": 0.5345,
+      "step": 38140
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004798877694171662,
+      "loss": 0.5302,
+      "step": 38150
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00047980274624835265,
+      "loss": 0.4868,
+      "step": 38160
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004797177230795392,
+      "loss": 0.6282,
+      "step": 38170
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00047963269991072575,
+      "loss": 0.5159,
+      "step": 38180
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004795476767419122,
+      "loss": 0.5472,
+      "step": 38190
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004794626535730987,
+      "loss": 0.5783,
+      "step": 38200
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00047937763040428516,
+      "loss": 0.5616,
+      "step": 38210
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004792926072354717,
+      "loss": 0.4924,
+      "step": 38220
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00047920758406665815,
+      "loss": 0.5614,
+      "step": 38230
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00047912256089784473,
+      "loss": 0.5034,
+      "step": 38240
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0004790375377290312,
+      "loss": 0.5356,
+      "step": 38250
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004789525145602177,
+      "loss": 0.567,
+      "step": 38260
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004788674913914042,
+      "loss": 0.5843,
+      "step": 38270
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00047878246822259066,
+      "loss": 0.5309,
+      "step": 38280
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004786974450537772,
+      "loss": 0.5491,
+      "step": 38290
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00047861242188496365,
+      "loss": 0.5874,
+      "step": 38300
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004785273987161501,
+      "loss": 0.5714,
+      "step": 38310
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004784423755473367,
+      "loss": 0.5431,
+      "step": 38320
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00047835735237852317,
+      "loss": 0.5996,
+      "step": 38330
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0004782723292097097,
+      "loss": 0.5567,
+      "step": 38340
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00047818730604089616,
+      "loss": 0.5067,
+      "step": 38350
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00047810228287208263,
+      "loss": 0.564,
+      "step": 38360
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047801725970326916,
+      "loss": 0.471,
+      "step": 38370
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047793223653445563,
+      "loss": 0.5645,
+      "step": 38380
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0004778472133656422,
+      "loss": 0.5217,
+      "step": 38390
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0004777621901968287,
+      "loss": 0.6002,
+      "step": 38400
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047767716702801514,
+      "loss": 0.5496,
+      "step": 38410
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047759214385920167,
+      "loss": 0.5726,
+      "step": 38420
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047750712069038814,
+      "loss": 0.4549,
+      "step": 38430
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0004774220975215746,
+      "loss": 0.5652,
+      "step": 38440
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047733707435276113,
+      "loss": 0.5938,
+      "step": 38450
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0004772520511839477,
+      "loss": 0.5223,
+      "step": 38460
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0004771670280151342,
+      "loss": 0.5672,
+      "step": 38470
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00047708200484632065,
+      "loss": 0.6081,
+      "step": 38480
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00047699698167750717,
+      "loss": 0.5047,
+      "step": 38490
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00047691195850869364,
+      "loss": 0.5115,
+      "step": 38500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004768269353398801,
+      "loss": 0.5651,
+      "step": 38510
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004767419121710666,
+      "loss": 0.5345,
+      "step": 38520
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004766568890022531,
+      "loss": 0.602,
+      "step": 38530
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004765718658334397,
+      "loss": 0.5213,
+      "step": 38540
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00047648684266462615,
+      "loss": 0.475,
+      "step": 38550
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004764018194958126,
+      "loss": 0.5693,
+      "step": 38560
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00047631679632699914,
+      "loss": 0.5753,
+      "step": 38570
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004762317731581856,
+      "loss": 0.5383,
+      "step": 38580
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004761467499893721,
+      "loss": 0.4899,
+      "step": 38590
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0004760617268205586,
+      "loss": 0.5523,
+      "step": 38600
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004759767036517452,
+      "loss": 0.5301,
+      "step": 38610
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00047589168048293166,
+      "loss": 0.521,
+      "step": 38620
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004758066573141181,
+      "loss": 0.5609,
+      "step": 38630
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004757216341453046,
+      "loss": 0.5783,
+      "step": 38640
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004756366109764911,
+      "loss": 0.5099,
+      "step": 38650
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004755515878076776,
+      "loss": 0.574,
+      "step": 38660
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00047546656463886406,
+      "loss": 0.5556,
+      "step": 38670
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00047538154147005064,
+      "loss": 0.5532,
+      "step": 38680
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00047529651830123716,
+      "loss": 0.5791,
+      "step": 38690
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00047521149513242363,
+      "loss": 0.5301,
+      "step": 38700
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004751264719636101,
+      "loss": 0.5299,
+      "step": 38710
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0004750414487947966,
+      "loss": 0.5444,
+      "step": 38720
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004749564256259831,
+      "loss": 0.5443,
+      "step": 38730
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00047487140245716956,
+      "loss": 0.5663,
+      "step": 38740
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00047478637928835614,
+      "loss": 0.5478,
+      "step": 38750
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004747013561195426,
+      "loss": 0.5534,
+      "step": 38760
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00047461633295072913,
+      "loss": 0.5025,
+      "step": 38770
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004745313097819156,
+      "loss": 0.5928,
+      "step": 38780
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00047444628661310207,
+      "loss": 0.5131,
+      "step": 38790
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004743612634442886,
+      "loss": 0.6107,
+      "step": 38800
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00047427624027547506,
+      "loss": 0.6017,
+      "step": 38810
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00047419121710666153,
+      "loss": 0.5173,
+      "step": 38820
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004741061939378481,
+      "loss": 0.6249,
+      "step": 38830
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0004740211707690346,
+      "loss": 0.5437,
+      "step": 38840
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0004739361476002211,
+      "loss": 0.5993,
+      "step": 38850
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0004738511244314076,
+      "loss": 0.5257,
+      "step": 38860
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00047376610126259404,
+      "loss": 0.5874,
+      "step": 38870
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00047368107809378057,
+      "loss": 0.5649,
+      "step": 38880
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00047359605492496704,
+      "loss": 0.5651,
+      "step": 38890
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0004735110317561536,
+      "loss": 0.5848,
+      "step": 38900
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0004734260085873401,
+      "loss": 0.5448,
+      "step": 38910
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0004733409854185266,
+      "loss": 0.5659,
+      "step": 38920
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0004732559622497131,
+      "loss": 0.6023,
+      "step": 38930
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00047317093908089955,
+      "loss": 0.5711,
+      "step": 38940
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.000473085915912086,
+      "loss": 0.5587,
+      "step": 38950
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00047300089274327254,
+      "loss": 0.5297,
+      "step": 38960
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0004729158695744591,
+      "loss": 0.5095,
+      "step": 38970
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0004728308464056456,
+      "loss": 0.5497,
+      "step": 38980
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047274582323683206,
+      "loss": 0.5674,
+      "step": 38990
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0004726608000680186,
+      "loss": 0.6173,
+      "step": 39000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047257577689920505,
+      "loss": 0.5448,
+      "step": 39010
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0004724907537303915,
+      "loss": 0.5282,
+      "step": 39020
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047240573056157804,
+      "loss": 0.572,
+      "step": 39030
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047232070739276457,
+      "loss": 0.6081,
+      "step": 39040
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0004722356842239511,
+      "loss": 0.5471,
+      "step": 39050
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047215066105513756,
+      "loss": 0.5806,
+      "step": 39060
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00047206563788632403,
+      "loss": 0.4967,
+      "step": 39070
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00047198061471751055,
+      "loss": 0.5837,
+      "step": 39080
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.000471895591548697,
+      "loss": 0.515,
+      "step": 39090
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0004718105683798835,
+      "loss": 0.5325,
+      "step": 39100
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00047172554521107,
+      "loss": 0.569,
+      "step": 39110
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0004716405220422566,
+      "loss": 0.5828,
+      "step": 39120
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00047155549887344306,
+      "loss": 0.5891,
+      "step": 39130
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00047147047570462953,
+      "loss": 0.5633,
+      "step": 39140
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.000471385452535816,
+      "loss": 0.5491,
+      "step": 39150
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00047130042936700253,
+      "loss": 0.4806,
+      "step": 39160
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.000471215406198189,
+      "loss": 0.573,
+      "step": 39170
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00047113038302937547,
+      "loss": 0.6105,
+      "step": 39180
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00047104535986056204,
+      "loss": 0.5456,
+      "step": 39190
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00047096033669174857,
+      "loss": 0.5371,
+      "step": 39200
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00047087531352293504,
+      "loss": 0.5432,
+      "step": 39210
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0004707902903541215,
+      "loss": 0.5982,
+      "step": 39220
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00047070526718530803,
+      "loss": 0.5583,
+      "step": 39230
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0004706202440164945,
+      "loss": 0.5468,
+      "step": 39240
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00047053522084768097,
+      "loss": 0.5712,
+      "step": 39250
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00047045019767886755,
+      "loss": 0.5556,
+      "step": 39260
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.000470365174510054,
+      "loss": 0.5103,
+      "step": 39270
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00047028015134124054,
+      "loss": 0.5642,
+      "step": 39280
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.000470195128172427,
+      "loss": 0.5905,
+      "step": 39290
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0004701101050036135,
+      "loss": 0.6041,
+      "step": 39300
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0004700250818348,
+      "loss": 0.5078,
+      "step": 39310
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0004699400586659865,
+      "loss": 0.5781,
+      "step": 39320
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00046985503549717305,
+      "loss": 0.6037,
+      "step": 39330
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0004697700123283595,
+      "loss": 0.5842,
+      "step": 39340
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00046968498915954604,
+      "loss": 0.5533,
+      "step": 39350
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0004695999659907325,
+      "loss": 0.5462,
+      "step": 39360
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.000469514942821919,
+      "loss": 0.6005,
+      "step": 39370
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00046942991965310545,
+      "loss": 0.5553,
+      "step": 39380
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.000469344896484292,
+      "loss": 0.5169,
+      "step": 39390
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00046925987331547845,
+      "loss": 0.5765,
+      "step": 39400
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.000469174850146665,
+      "loss": 0.5864,
+      "step": 39410
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0004690898269778515,
+      "loss": 0.5879,
+      "step": 39420
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.000469004803809038,
+      "loss": 0.5164,
+      "step": 39430
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0004689197806402245,
+      "loss": 0.5371,
+      "step": 39440
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046883475747141096,
+      "loss": 0.5299,
+      "step": 39450
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0004687497343025975,
+      "loss": 0.5738,
+      "step": 39460
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046866471113378395,
+      "loss": 0.562,
+      "step": 39470
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046857968796497053,
+      "loss": 0.5096,
+      "step": 39480
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.000468494664796157,
+      "loss": 0.5842,
+      "step": 39490
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046840964162734347,
+      "loss": 0.5803,
+      "step": 39500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046832461845853,
+      "loss": 0.5923,
+      "step": 39510
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046823959528971646,
+      "loss": 0.5272,
+      "step": 39520
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046815457212090293,
+      "loss": 0.5081,
+      "step": 39530
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046806954895208945,
+      "loss": 0.4875,
+      "step": 39540
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00046798452578327603,
+      "loss": 0.5218,
+      "step": 39550
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0004678995026144625,
+      "loss": 0.5059,
+      "step": 39560
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046781447944564897,
+      "loss": 0.5434,
+      "step": 39570
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046772945627683544,
+      "loss": 0.5573,
+      "step": 39580
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046764443310802196,
+      "loss": 0.5801,
+      "step": 39590
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046755940993920843,
+      "loss": 0.5406,
+      "step": 39600
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0004674743867703949,
+      "loss": 0.5622,
+      "step": 39610
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0004673893636015814,
+      "loss": 0.5571,
+      "step": 39620
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.000467304340432768,
+      "loss": 0.5334,
+      "step": 39630
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0004672193172639545,
+      "loss": 0.5726,
+      "step": 39640
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046713429409514094,
+      "loss": 0.6027,
+      "step": 39650
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00046704927092632747,
+      "loss": 0.5175,
+      "step": 39660
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046696424775751394,
+      "loss": 0.6304,
+      "step": 39670
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004668792245887004,
+      "loss": 0.5433,
+      "step": 39680
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004667942014198869,
+      "loss": 0.5834,
+      "step": 39690
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046670917825107345,
+      "loss": 0.5621,
+      "step": 39700
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046662415508226,
+      "loss": 0.5577,
+      "step": 39710
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046653913191344645,
+      "loss": 0.5286,
+      "step": 39720
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004664541087446329,
+      "loss": 0.5428,
+      "step": 39730
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046636908557581944,
+      "loss": 0.6144,
+      "step": 39740
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004662840624070059,
+      "loss": 0.5372,
+      "step": 39750
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0004661990392381924,
+      "loss": 0.495,
+      "step": 39760
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046611401606937896,
+      "loss": 0.5214,
+      "step": 39770
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00046602899290056543,
+      "loss": 0.5313,
+      "step": 39780
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00046594396973175195,
+      "loss": 0.5848,
+      "step": 39790
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004658589465629384,
+      "loss": 0.5971,
+      "step": 39800
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004657739233941249,
+      "loss": 0.5591,
+      "step": 39810
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004656889002253114,
+      "loss": 0.5347,
+      "step": 39820
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004656038770564979,
+      "loss": 0.5742,
+      "step": 39830
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00046551885388768446,
+      "loss": 0.5103,
+      "step": 39840
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00046543383071887093,
+      "loss": 0.5288,
+      "step": 39850
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00046534880755005745,
+      "loss": 0.6228,
+      "step": 39860
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004652637843812439,
+      "loss": 0.5627,
+      "step": 39870
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004651787612124304,
+      "loss": 0.5162,
+      "step": 39880
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00046509373804361686,
+      "loss": 0.5728,
+      "step": 39890
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0004650087148748034,
+      "loss": 0.5602,
+      "step": 39900
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046492369170598986,
+      "loss": 0.5114,
+      "step": 39910
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046483866853717643,
+      "loss": 0.4901,
+      "step": 39920
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004647536453683629,
+      "loss": 0.4761,
+      "step": 39930
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046466862219954943,
+      "loss": 0.5432,
+      "step": 39940
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004645835990307359,
+      "loss": 0.4984,
+      "step": 39950
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046449857586192237,
+      "loss": 0.5672,
+      "step": 39960
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004644135526931089,
+      "loss": 0.5994,
+      "step": 39970
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046432852952429536,
+      "loss": 0.528,
+      "step": 39980
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00046424350635548194,
+      "loss": 0.51,
+      "step": 39990
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004641584831866684,
+      "loss": 0.5268,
+      "step": 40000
+    },
+    {
+      "epoch": 3.38,
+      "eval_loss": 0.5884655714035034,
+      "eval_runtime": 312.1444,
+      "eval_samples_per_second": 16.835,
+      "eval_steps_per_second": 2.105,
+      "step": 40000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004640734600178549,
+      "loss": 0.5079,
+      "step": 40010
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004639884368490414,
+      "loss": 0.5385,
+      "step": 40020
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00046390341368022787,
+      "loss": 0.4994,
+      "step": 40030
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00046381839051141434,
+      "loss": 0.5415,
+      "step": 40040
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00046373336734260086,
+      "loss": 0.6306,
+      "step": 40050
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00046364834417378744,
+      "loss": 0.5124,
+      "step": 40060
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004635633210049739,
+      "loss": 0.5875,
+      "step": 40070
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004634782978361604,
+      "loss": 0.5842,
+      "step": 40080
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004633932746673469,
+      "loss": 0.6121,
+      "step": 40090
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004633082514985334,
+      "loss": 0.5606,
+      "step": 40100
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00046322322832971984,
+      "loss": 0.5404,
+      "step": 40110
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004631382051609063,
+      "loss": 0.4507,
+      "step": 40120
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004630531819920929,
+      "loss": 0.5704,
+      "step": 40130
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004629681588232794,
+      "loss": 0.588,
+      "step": 40140
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004628831356544659,
+      "loss": 0.5466,
+      "step": 40150
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00046279811248565235,
+      "loss": 0.6003,
+      "step": 40160
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004627130893168389,
+      "loss": 0.5557,
+      "step": 40170
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00046262806614802535,
+      "loss": 0.5526,
+      "step": 40180
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004625430429792118,
+      "loss": 0.5564,
+      "step": 40190
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00046245801981039834,
+      "loss": 0.4804,
+      "step": 40200
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00046237299664158486,
+      "loss": 0.5515,
+      "step": 40210
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004622879734727714,
+      "loss": 0.5476,
+      "step": 40220
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00046220295030395786,
+      "loss": 0.5271,
+      "step": 40230
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004621179271351443,
+      "loss": 0.4908,
+      "step": 40240
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00046203290396633085,
+      "loss": 0.5,
+      "step": 40250
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004619478807975173,
+      "loss": 0.5183,
+      "step": 40260
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004618628576287038,
+      "loss": 0.5646,
+      "step": 40270
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00046177783445989037,
+      "loss": 0.5592,
+      "step": 40280
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004616928112910769,
+      "loss": 0.5061,
+      "step": 40290
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00046160778812226336,
+      "loss": 0.5292,
+      "step": 40300
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00046152276495344983,
+      "loss": 0.5406,
+      "step": 40310
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004614377417846363,
+      "loss": 0.6087,
+      "step": 40320
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004613527186158228,
+      "loss": 0.51,
+      "step": 40330
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004612676954470093,
+      "loss": 0.4998,
+      "step": 40340
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00046118267227819587,
+      "loss": 0.5961,
+      "step": 40350
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00046109764910938234,
+      "loss": 0.5655,
+      "step": 40360
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00046101262594056886,
+      "loss": 0.6251,
+      "step": 40370
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00046092760277175533,
+      "loss": 0.5797,
+      "step": 40380
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004608425796029418,
+      "loss": 0.5423,
+      "step": 40390
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004607575564341283,
+      "loss": 0.581,
+      "step": 40400
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004606725332653148,
+      "loss": 0.5297,
+      "step": 40410
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004605875100965014,
+      "loss": 0.5224,
+      "step": 40420
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00046050248692768784,
+      "loss": 0.5376,
+      "step": 40430
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004604174637588743,
+      "loss": 0.5467,
+      "step": 40440
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00046033244059006084,
+      "loss": 0.4867,
+      "step": 40450
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004602474174212473,
+      "loss": 0.5241,
+      "step": 40460
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004601623942524338,
+      "loss": 0.5404,
+      "step": 40470
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004600773710836203,
+      "loss": 0.5097,
+      "step": 40480
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00045999234791480677,
+      "loss": 0.589,
+      "step": 40490
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00045990732474599335,
+      "loss": 0.5182,
+      "step": 40500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004598223015771798,
+      "loss": 0.5227,
+      "step": 40510
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004597372784083663,
+      "loss": 0.5604,
+      "step": 40520
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004596522552395528,
+      "loss": 0.5685,
+      "step": 40530
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004595672320707393,
+      "loss": 0.5684,
+      "step": 40540
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00045948220890192575,
+      "loss": 0.5775,
+      "step": 40550
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004593971857331123,
+      "loss": 0.532,
+      "step": 40560
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00045931216256429885,
+      "loss": 0.5445,
+      "step": 40570
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004592271393954853,
+      "loss": 0.5801,
+      "step": 40580
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004591421162266718,
+      "loss": 0.5501,
+      "step": 40590
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004590570930578583,
+      "loss": 0.5273,
+      "step": 40600
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004589720698890448,
+      "loss": 0.5327,
+      "step": 40610
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00045888704672023125,
+      "loss": 0.5646,
+      "step": 40620
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004588020235514177,
+      "loss": 0.5482,
+      "step": 40630
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004587170003826043,
+      "loss": 0.534,
+      "step": 40640
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004586319772137908,
+      "loss": 0.5243,
+      "step": 40650
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004585469540449773,
+      "loss": 0.4856,
+      "step": 40660
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00045846193087616376,
+      "loss": 0.512,
+      "step": 40670
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004583769077073503,
+      "loss": 0.5684,
+      "step": 40680
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00045829188453853676,
+      "loss": 0.5856,
+      "step": 40690
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004582068613697232,
+      "loss": 0.5429,
+      "step": 40700
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004581218382009098,
+      "loss": 0.5531,
+      "step": 40710
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004580368150320963,
+      "loss": 0.5437,
+      "step": 40720
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004579517918632828,
+      "loss": 0.6066,
+      "step": 40730
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00045786676869446927,
+      "loss": 0.5893,
+      "step": 40740
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00045778174552565574,
+      "loss": 0.5611,
+      "step": 40750
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00045769672235684226,
+      "loss": 0.5865,
+      "step": 40760
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00045761169918802873,
+      "loss": 0.4996,
+      "step": 40770
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004575266760192152,
+      "loss": 0.6028,
+      "step": 40780
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004574416528504018,
+      "loss": 0.5739,
+      "step": 40790
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004573566296815883,
+      "loss": 0.5352,
+      "step": 40800
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00045727160651277477,
+      "loss": 0.5659,
+      "step": 40810
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00045718658334396124,
+      "loss": 0.5624,
+      "step": 40820
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004571015601751477,
+      "loss": 0.587,
+      "step": 40830
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00045701653700633423,
+      "loss": 0.5396,
+      "step": 40840
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004569315138375207,
+      "loss": 0.5425,
+      "step": 40850
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004568464906687073,
+      "loss": 0.5103,
+      "step": 40860
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045676146749989375,
+      "loss": 0.5563,
+      "step": 40870
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004566764443310803,
+      "loss": 0.5567,
+      "step": 40880
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045659142116226674,
+      "loss": 0.6015,
+      "step": 40890
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004565063979934532,
+      "loss": 0.5445,
+      "step": 40900
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045642137482463974,
+      "loss": 0.5454,
+      "step": 40910
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004563363516558262,
+      "loss": 0.6219,
+      "step": 40920
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004562513284870128,
+      "loss": 0.5857,
+      "step": 40930
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045616630531819925,
+      "loss": 0.5477,
+      "step": 40940
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004560812821493857,
+      "loss": 0.6048,
+      "step": 40950
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045599625898057225,
+      "loss": 0.5408,
+      "step": 40960
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004559112358117587,
+      "loss": 0.6473,
+      "step": 40970
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004558262126429452,
+      "loss": 0.591,
+      "step": 40980
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004557411894741317,
+      "loss": 0.5411,
+      "step": 40990
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004556561663053182,
+      "loss": 0.5613,
+      "step": 41000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045557114313650476,
+      "loss": 0.5344,
+      "step": 41010
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004554861199676912,
+      "loss": 0.5139,
+      "step": 41020
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045540109679887775,
+      "loss": 0.564,
+      "step": 41030
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004553160736300642,
+      "loss": 0.5184,
+      "step": 41040
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004552310504612507,
+      "loss": 0.5972,
+      "step": 41050
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045514602729243716,
+      "loss": 0.5041,
+      "step": 41060
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004550610041236237,
+      "loss": 0.569,
+      "step": 41070
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045497598095481026,
+      "loss": 0.5673,
+      "step": 41080
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00045489095778599673,
+      "loss": 0.5286,
+      "step": 41090
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004548059346171832,
+      "loss": 0.5863,
+      "step": 41100
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004547209114483697,
+      "loss": 0.562,
+      "step": 41110
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004546358882795562,
+      "loss": 0.5456,
+      "step": 41120
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00045455086511074266,
+      "loss": 0.5611,
+      "step": 41130
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004544658419419292,
+      "loss": 0.5475,
+      "step": 41140
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004543808187731157,
+      "loss": 0.597,
+      "step": 41150
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00045429579560430223,
+      "loss": 0.5617,
+      "step": 41160
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004542107724354887,
+      "loss": 0.5356,
+      "step": 41170
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00045412574926667517,
+      "loss": 0.5463,
+      "step": 41180
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004540407260978617,
+      "loss": 0.5829,
+      "step": 41190
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00045395570292904817,
+      "loss": 0.5239,
+      "step": 41200
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00045387067976023464,
+      "loss": 0.5601,
+      "step": 41210
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004537856565914212,
+      "loss": 0.5244,
+      "step": 41220
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00045370063342260774,
+      "loss": 0.6159,
+      "step": 41230
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004536156102537942,
+      "loss": 0.5648,
+      "step": 41240
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004535305870849807,
+      "loss": 0.5398,
+      "step": 41250
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00045344556391616715,
+      "loss": 0.5854,
+      "step": 41260
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00045336054074735367,
+      "loss": 0.5301,
+      "step": 41270
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00045327551757854014,
+      "loss": 0.5672,
+      "step": 41280
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004531904944097266,
+      "loss": 0.6609,
+      "step": 41290
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004531054712409132,
+      "loss": 0.597,
+      "step": 41300
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004530204480720997,
+      "loss": 0.5507,
+      "step": 41310
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004529354249032862,
+      "loss": 0.5926,
+      "step": 41320
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00045285040173447265,
+      "loss": 0.5492,
+      "step": 41330
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004527653785656592,
+      "loss": 0.5999,
+      "step": 41340
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00045268035539684564,
+      "loss": 0.5447,
+      "step": 41350
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004525953322280321,
+      "loss": 0.5357,
+      "step": 41360
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004525103090592187,
+      "loss": 0.612,
+      "step": 41370
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00045242528589040516,
+      "loss": 0.5467,
+      "step": 41380
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004523402627215917,
+      "loss": 0.4791,
+      "step": 41390
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00045225523955277815,
+      "loss": 0.5635,
+      "step": 41400
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004521702163839646,
+      "loss": 0.595,
+      "step": 41410
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00045208519321515115,
+      "loss": 0.5359,
+      "step": 41420
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004520001700463376,
+      "loss": 0.5555,
+      "step": 41430
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004519151468775242,
+      "loss": 0.5395,
+      "step": 41440
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00045183012370871066,
+      "loss": 0.527,
+      "step": 41450
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00045174510053989713,
+      "loss": 0.5711,
+      "step": 41460
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00045166007737108366,
+      "loss": 0.6691,
+      "step": 41470
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004515750542022701,
+      "loss": 0.565,
+      "step": 41480
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004514900310334566,
+      "loss": 0.6025,
+      "step": 41490
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004514050078646431,
+      "loss": 0.575,
+      "step": 41500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004513199846958297,
+      "loss": 0.6005,
+      "step": 41510
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00045123496152701617,
+      "loss": 0.5394,
+      "step": 41520
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00045114993835820264,
+      "loss": 0.5733,
+      "step": 41530
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00045106491518938916,
+      "loss": 0.5599,
+      "step": 41540
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00045097989202057563,
+      "loss": 0.4956,
+      "step": 41550
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004508948688517621,
+      "loss": 0.5548,
+      "step": 41560
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00045080984568294857,
+      "loss": 0.5003,
+      "step": 41570
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004507248225141351,
+      "loss": 0.5627,
+      "step": 41580
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00045063979934532167,
+      "loss": 0.5829,
+      "step": 41590
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00045055477617650814,
+      "loss": 0.5538,
+      "step": 41600
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004504697530076946,
+      "loss": 0.5796,
+      "step": 41610
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00045038472983888113,
+      "loss": 0.5585,
+      "step": 41620
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004502997066700676,
+      "loss": 0.5365,
+      "step": 41630
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00045021468350125407,
+      "loss": 0.521,
+      "step": 41640
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004501296603324406,
+      "loss": 0.5393,
+      "step": 41650
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004500446371636272,
+      "loss": 0.5331,
+      "step": 41660
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00044995961399481364,
+      "loss": 0.5882,
+      "step": 41670
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004498745908260001,
+      "loss": 0.556,
+      "step": 41680
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004497895676571866,
+      "loss": 0.5928,
+      "step": 41690
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004497045444883731,
+      "loss": 0.5638,
+      "step": 41700
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004496195213195596,
+      "loss": 0.4748,
+      "step": 41710
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00044953449815074605,
+      "loss": 0.5775,
+      "step": 41720
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004494494749819326,
+      "loss": 0.5705,
+      "step": 41730
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00044936445181311915,
+      "loss": 0.664,
+      "step": 41740
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004492794286443056,
+      "loss": 0.5647,
+      "step": 41750
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004491944054754921,
+      "loss": 0.5307,
+      "step": 41760
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004491093823066786,
+      "loss": 0.526,
+      "step": 41770
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004490243591378651,
+      "loss": 0.5605,
+      "step": 41780
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00044893933596905155,
+      "loss": 0.5243,
+      "step": 41790
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004488543128002381,
+      "loss": 0.6285,
+      "step": 41800
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004487692896314246,
+      "loss": 0.5976,
+      "step": 41810
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004486842664626111,
+      "loss": 0.5896,
+      "step": 41820
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004485992432937976,
+      "loss": 0.5158,
+      "step": 41830
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00044851422012498406,
+      "loss": 0.5108,
+      "step": 41840
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004484291969561706,
+      "loss": 0.5849,
+      "step": 41850
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00044834417378735705,
+      "loss": 0.6225,
+      "step": 41860
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004482591506185435,
+      "loss": 0.5782,
+      "step": 41870
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004481741274497301,
+      "loss": 0.5089,
+      "step": 41880
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00044808910428091657,
+      "loss": 0.5778,
+      "step": 41890
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004480040811121031,
+      "loss": 0.5326,
+      "step": 41900
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00044791905794328956,
+      "loss": 0.5308,
+      "step": 41910
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044783403477447603,
+      "loss": 0.5276,
+      "step": 41920
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044774901160566256,
+      "loss": 0.5983,
+      "step": 41930
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.000447663988436849,
+      "loss": 0.5971,
+      "step": 41940
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004475789652680356,
+      "loss": 0.579,
+      "step": 41950
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044749394209922207,
+      "loss": 0.5984,
+      "step": 41960
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004474089189304086,
+      "loss": 0.4954,
+      "step": 41970
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044732389576159507,
+      "loss": 0.5864,
+      "step": 41980
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044723887259278154,
+      "loss": 0.506,
+      "step": 41990
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.000447153849423968,
+      "loss": 0.5068,
+      "step": 42000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00044706882625515453,
+      "loss": 0.5731,
+      "step": 42010
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004469838030863411,
+      "loss": 0.4623,
+      "step": 42020
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004468987799175276,
+      "loss": 0.4488,
+      "step": 42030
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044681375674871405,
+      "loss": 0.541,
+      "step": 42040
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044672873357990057,
+      "loss": 0.5569,
+      "step": 42050
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044664371041108704,
+      "loss": 0.5648,
+      "step": 42060
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004465586872422735,
+      "loss": 0.5513,
+      "step": 42070
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044647366407346003,
+      "loss": 0.5964,
+      "step": 42080
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004463886409046465,
+      "loss": 0.5691,
+      "step": 42090
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004463036177358331,
+      "loss": 0.5671,
+      "step": 42100
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044621859456701955,
+      "loss": 0.5332,
+      "step": 42110
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.000446133571398206,
+      "loss": 0.4945,
+      "step": 42120
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00044604854822939254,
+      "loss": 0.5852,
+      "step": 42130
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.000445963525060579,
+      "loss": 0.5685,
+      "step": 42140
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004458785018917655,
+      "loss": 0.5136,
+      "step": 42150
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.000445793478722952,
+      "loss": 0.5326,
+      "step": 42160
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004457084555541386,
+      "loss": 0.5729,
+      "step": 42170
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00044562343238532505,
+      "loss": 0.5294,
+      "step": 42180
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004455384092165115,
+      "loss": 0.5853,
+      "step": 42190
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.000445453386047698,
+      "loss": 0.5278,
+      "step": 42200
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004453683628788845,
+      "loss": 0.7179,
+      "step": 42210
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.000445283339710071,
+      "loss": 0.5698,
+      "step": 42220
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00044519831654125745,
+      "loss": 0.5208,
+      "step": 42230
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00044511329337244403,
+      "loss": 0.5351,
+      "step": 42240
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00044502827020363056,
+      "loss": 0.5474,
+      "step": 42250
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.000444943247034817,
+      "loss": 0.5516,
+      "step": 42260
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004448582238660035,
+      "loss": 0.5806,
+      "step": 42270
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044477320069719,
+      "loss": 0.5531,
+      "step": 42280
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0004446881775283765,
+      "loss": 0.5449,
+      "step": 42290
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044460315435956296,
+      "loss": 0.5171,
+      "step": 42300
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044451813119074954,
+      "loss": 0.531,
+      "step": 42310
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.000444433108021936,
+      "loss": 0.5382,
+      "step": 42320
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044434808485312253,
+      "loss": 0.5417,
+      "step": 42330
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.000444263061684309,
+      "loss": 0.5882,
+      "step": 42340
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044417803851549547,
+      "loss": 0.6192,
+      "step": 42350
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.000444093015346682,
+      "loss": 0.5594,
+      "step": 42360
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044400799217786846,
+      "loss": 0.4952,
+      "step": 42370
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00044392296900905493,
+      "loss": 0.5898,
+      "step": 42380
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004438379458402415,
+      "loss": 0.5737,
+      "step": 42390
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00044375292267142803,
+      "loss": 0.6027,
+      "step": 42400
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004436678995026145,
+      "loss": 0.5568,
+      "step": 42410
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00044358287633380097,
+      "loss": 0.5611,
+      "step": 42420
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00044349785316498744,
+      "loss": 0.5587,
+      "step": 42430
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00044341282999617397,
+      "loss": 0.551,
+      "step": 42440
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00044332780682736043,
+      "loss": 0.581,
+      "step": 42450
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.000443242783658547,
+      "loss": 0.521,
+      "step": 42460
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004431577604897335,
+      "loss": 0.4932,
+      "step": 42470
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00044307273732092,
+      "loss": 0.6127,
+      "step": 42480
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004429877141521065,
+      "loss": 0.6047,
+      "step": 42490
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00044290269098329295,
+      "loss": 0.622,
+      "step": 42500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044281766781447947,
+      "loss": 0.5794,
+      "step": 42510
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044273264464566594,
+      "loss": 0.5333,
+      "step": 42520
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0004426476214768525,
+      "loss": 0.5624,
+      "step": 42530
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.000442562598308039,
+      "loss": 0.553,
+      "step": 42540
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044247757513922546,
+      "loss": 0.5051,
+      "step": 42550
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.000442392551970412,
+      "loss": 0.5474,
+      "step": 42560
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044230752880159845,
+      "loss": 0.5165,
+      "step": 42570
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0004422225056327849,
+      "loss": 0.5928,
+      "step": 42580
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044213748246397144,
+      "loss": 0.5384,
+      "step": 42590
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.000442052459295158,
+      "loss": 0.5926,
+      "step": 42600
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0004419674361263445,
+      "loss": 0.6124,
+      "step": 42610
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00044188241295753096,
+      "loss": 0.551,
+      "step": 42620
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00044179738978871743,
+      "loss": 0.5353,
+      "step": 42630
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00044171236661990395,
+      "loss": 0.5276,
+      "step": 42640
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004416273434510904,
+      "loss": 0.5396,
+      "step": 42650
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004415423202822769,
+      "loss": 0.5745,
+      "step": 42660
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004414572971134634,
+      "loss": 0.5546,
+      "step": 42670
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00044137227394465,
+      "loss": 0.547,
+      "step": 42680
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00044128725077583646,
+      "loss": 0.534,
+      "step": 42690
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00044120222760702293,
+      "loss": 0.6336,
+      "step": 42700
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00044111720443820946,
+      "loss": 0.5739,
+      "step": 42710
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004410321812693959,
+      "loss": 0.4778,
+      "step": 42720
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004409471581005824,
+      "loss": 0.5498,
+      "step": 42730
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00044086213493176886,
+      "loss": 0.5713,
+      "step": 42740
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00044077711176295544,
+      "loss": 0.5638,
+      "step": 42750
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00044069208859414197,
+      "loss": 0.512,
+      "step": 42760
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00044060706542532844,
+      "loss": 0.5299,
+      "step": 42770
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0004405220422565149,
+      "loss": 0.5361,
+      "step": 42780
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00044043701908770143,
+      "loss": 0.5736,
+      "step": 42790
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0004403519959188879,
+      "loss": 0.5624,
+      "step": 42800
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00044026697275007437,
+      "loss": 0.5552,
+      "step": 42810
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00044018194958126095,
+      "loss": 0.5597,
+      "step": 42820
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0004400969264124474,
+      "loss": 0.5866,
+      "step": 42830
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00044001190324363394,
+      "loss": 0.5845,
+      "step": 42840
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0004399268800748204,
+      "loss": 0.5678,
+      "step": 42850
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0004398418569060069,
+      "loss": 0.5857,
+      "step": 42860
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004397568337371934,
+      "loss": 0.5585,
+      "step": 42870
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00043967181056837987,
+      "loss": 0.585,
+      "step": 42880
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00043958678739956645,
+      "loss": 0.5058,
+      "step": 42890
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004395017642307529,
+      "loss": 0.6143,
+      "step": 42900
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00043941674106193944,
+      "loss": 0.5563,
+      "step": 42910
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004393317178931259,
+      "loss": 0.5026,
+      "step": 42920
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004392466947243124,
+      "loss": 0.5717,
+      "step": 42930
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00043916167155549885,
+      "loss": 0.5928,
+      "step": 42940
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004390766483866854,
+      "loss": 0.5715,
+      "step": 42950
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00043899162521787184,
+      "loss": 0.5717,
+      "step": 42960
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004389066020490584,
+      "loss": 0.5184,
+      "step": 42970
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0004388215788802449,
+      "loss": 0.5183,
+      "step": 42980
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004387365557114314,
+      "loss": 0.5018,
+      "step": 42990
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004386515325426179,
+      "loss": 0.5681,
+      "step": 43000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00043856650937380435,
+      "loss": 0.5571,
+      "step": 43010
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004384814862049909,
+      "loss": 0.5597,
+      "step": 43020
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00043839646303617735,
+      "loss": 0.5477,
+      "step": 43030
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004383114398673639,
+      "loss": 0.4836,
+      "step": 43040
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004382264166985504,
+      "loss": 0.5348,
+      "step": 43050
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00043814139352973687,
+      "loss": 0.5683,
+      "step": 43060
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0004380563703609234,
+      "loss": 0.6,
+      "step": 43070
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00043797134719210986,
+      "loss": 0.5968,
+      "step": 43080
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00043788632402329633,
+      "loss": 0.629,
+      "step": 43090
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043780130085448285,
+      "loss": 0.568,
+      "step": 43100
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043771627768566943,
+      "loss": 0.591,
+      "step": 43110
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0004376312545168559,
+      "loss": 0.5702,
+      "step": 43120
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043754623134804237,
+      "loss": 0.5474,
+      "step": 43130
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043746120817922884,
+      "loss": 0.5222,
+      "step": 43140
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043737618501041536,
+      "loss": 0.4885,
+      "step": 43150
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043729116184160183,
+      "loss": 0.6445,
+      "step": 43160
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0004372061386727883,
+      "loss": 0.5866,
+      "step": 43170
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0004371211155039749,
+      "loss": 0.4991,
+      "step": 43180
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0004370360923351614,
+      "loss": 0.5459,
+      "step": 43190
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043695106916634787,
+      "loss": 0.6005,
+      "step": 43200
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00043686604599753434,
+      "loss": 0.6008,
+      "step": 43210
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043678102282872087,
+      "loss": 0.5966,
+      "step": 43220
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043669599965990734,
+      "loss": 0.5346,
+      "step": 43230
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004366109764910938,
+      "loss": 0.4992,
+      "step": 43240
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004365259533222803,
+      "loss": 0.6275,
+      "step": 43250
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043644093015346685,
+      "loss": 0.5487,
+      "step": 43260
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004363559069846534,
+      "loss": 0.575,
+      "step": 43270
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043627088381583985,
+      "loss": 0.5321,
+      "step": 43280
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004361858606470263,
+      "loss": 0.5755,
+      "step": 43290
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043610083747821284,
+      "loss": 0.5729,
+      "step": 43300
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004360158143093993,
+      "loss": 0.5254,
+      "step": 43310
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0004359307911405858,
+      "loss": 0.5241,
+      "step": 43320
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00043584576797177236,
+      "loss": 0.5337,
+      "step": 43330
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004357607448029589,
+      "loss": 0.5756,
+      "step": 43340
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00043567572163414535,
+      "loss": 0.5115,
+      "step": 43350
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004355906984653318,
+      "loss": 0.5618,
+      "step": 43360
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004355056752965183,
+      "loss": 0.6531,
+      "step": 43370
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004354206521277048,
+      "loss": 0.5407,
+      "step": 43380
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004353356289588913,
+      "loss": 0.5508,
+      "step": 43390
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00043525060579007786,
+      "loss": 0.5295,
+      "step": 43400
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00043516558262126433,
+      "loss": 0.5526,
+      "step": 43410
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00043508055945245085,
+      "loss": 0.6304,
+      "step": 43420
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004349955362836373,
+      "loss": 0.6071,
+      "step": 43430
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004349105131148238,
+      "loss": 0.4851,
+      "step": 43440
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0004348254899460103,
+      "loss": 0.5295,
+      "step": 43450
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004347404667771968,
+      "loss": 0.5384,
+      "step": 43460
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00043465544360838325,
+      "loss": 0.5229,
+      "step": 43470
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00043457042043956983,
+      "loss": 0.619,
+      "step": 43480
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004344853972707563,
+      "loss": 0.5846,
+      "step": 43490
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004344003741019428,
+      "loss": 0.5625,
+      "step": 43500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004343153509331293,
+      "loss": 0.6149,
+      "step": 43510
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00043423032776431576,
+      "loss": 0.6131,
+      "step": 43520
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004341453045955023,
+      "loss": 0.4849,
+      "step": 43530
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00043406028142668876,
+      "loss": 0.5276,
+      "step": 43540
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00043397525825787534,
+      "loss": 0.5887,
+      "step": 43550
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004338902350890618,
+      "loss": 0.5602,
+      "step": 43560
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004338052119202483,
+      "loss": 0.5943,
+      "step": 43570
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0004337201887514348,
+      "loss": 0.5406,
+      "step": 43580
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043363516558262127,
+      "loss": 0.5642,
+      "step": 43590
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043355014241380774,
+      "loss": 0.5906,
+      "step": 43600
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043346511924499426,
+      "loss": 0.5354,
+      "step": 43610
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043338009607618084,
+      "loss": 0.5195,
+      "step": 43620
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0004332950729073673,
+      "loss": 0.5233,
+      "step": 43630
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0004332100497385538,
+      "loss": 0.5367,
+      "step": 43640
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0004331250265697403,
+      "loss": 0.5113,
+      "step": 43650
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043304000340092677,
+      "loss": 0.5474,
+      "step": 43660
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00043295498023211324,
+      "loss": 0.524,
+      "step": 43670
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0004328699570632997,
+      "loss": 0.6019,
+      "step": 43680
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0004327849338944863,
+      "loss": 0.6803,
+      "step": 43690
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0004326999107256728,
+      "loss": 0.5812,
+      "step": 43700
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0004326148875568593,
+      "loss": 0.5258,
+      "step": 43710
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043252986438804575,
+      "loss": 0.4949,
+      "step": 43720
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0004324448412192323,
+      "loss": 0.589,
+      "step": 43730
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043235981805041874,
+      "loss": 0.5562,
+      "step": 43740
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0004322747948816052,
+      "loss": 0.6137,
+      "step": 43750
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043218977171279174,
+      "loss": 0.5444,
+      "step": 43760
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043210474854397826,
+      "loss": 0.5255,
+      "step": 43770
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0004320197253751648,
+      "loss": 0.5366,
+      "step": 43780
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00043193470220635125,
+      "loss": 0.5789,
+      "step": 43790
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0004318496790375377,
+      "loss": 0.6572,
+      "step": 43800
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00043176465586872425,
+      "loss": 0.5772,
+      "step": 43810
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004316796326999107,
+      "loss": 0.5778,
+      "step": 43820
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004315946095310972,
+      "loss": 0.5795,
+      "step": 43830
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00043150958636228377,
+      "loss": 0.5294,
+      "step": 43840
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004314245631934703,
+      "loss": 0.557,
+      "step": 43850
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00043133954002465676,
+      "loss": 0.5156,
+      "step": 43860
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00043125451685584323,
+      "loss": 0.5591,
+      "step": 43870
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004311694936870297,
+      "loss": 0.6337,
+      "step": 43880
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004310844705182162,
+      "loss": 0.6227,
+      "step": 43890
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0004309994473494027,
+      "loss": 0.6187,
+      "step": 43900
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00043091442418058927,
+      "loss": 0.5387,
+      "step": 43910
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00043082940101177574,
+      "loss": 0.584,
+      "step": 43920
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00043074437784296226,
+      "loss": 0.6159,
+      "step": 43930
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00043065935467414873,
+      "loss": 0.5372,
+      "step": 43940
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004305743315053352,
+      "loss": 0.5704,
+      "step": 43950
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004304893083365217,
+      "loss": 0.5324,
+      "step": 43960
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004304042851677082,
+      "loss": 0.6102,
+      "step": 43970
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00043031926199889477,
+      "loss": 0.5656,
+      "step": 43980
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00043023423883008124,
+      "loss": 0.5514,
+      "step": 43990
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004301492156612677,
+      "loss": 0.5909,
+      "step": 44000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00043006419249245424,
+      "loss": 0.521,
+      "step": 44010
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004299791693236407,
+      "loss": 0.5623,
+      "step": 44020
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004298941461548272,
+      "loss": 0.5353,
+      "step": 44030
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0004298091229860137,
+      "loss": 0.5159,
+      "step": 44040
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042972409981720017,
+      "loss": 0.534,
+      "step": 44050
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042963907664838675,
+      "loss": 0.5411,
+      "step": 44060
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004295540534795732,
+      "loss": 0.5455,
+      "step": 44070
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042946903031075974,
+      "loss": 0.5421,
+      "step": 44080
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004293840071419462,
+      "loss": 0.5404,
+      "step": 44090
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004292989839731327,
+      "loss": 0.6334,
+      "step": 44100
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042921396080431915,
+      "loss": 0.5245,
+      "step": 44110
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042912893763550567,
+      "loss": 0.5509,
+      "step": 44120
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00042904391446669225,
+      "loss": 0.5763,
+      "step": 44130
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004289588912978787,
+      "loss": 0.5634,
+      "step": 44140
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004288738681290652,
+      "loss": 0.513,
+      "step": 44150
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004287888449602517,
+      "loss": 0.4968,
+      "step": 44160
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004287038217914382,
+      "loss": 0.6463,
+      "step": 44170
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00042861879862262465,
+      "loss": 0.6254,
+      "step": 44180
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004285337754538112,
+      "loss": 0.6083,
+      "step": 44190
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004284487522849977,
+      "loss": 0.5991,
+      "step": 44200
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004283637291161842,
+      "loss": 0.5512,
+      "step": 44210
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004282787059473707,
+      "loss": 0.4968,
+      "step": 44220
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00042819368277855716,
+      "loss": 0.5847,
+      "step": 44230
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004281086596097437,
+      "loss": 0.5215,
+      "step": 44240
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00042802363644093015,
+      "loss": 0.5984,
+      "step": 44250
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004279386132721166,
+      "loss": 0.5207,
+      "step": 44260
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004278535901033032,
+      "loss": 0.5509,
+      "step": 44270
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0004277685669344897,
+      "loss": 0.5465,
+      "step": 44280
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004276835437656762,
+      "loss": 0.4752,
+      "step": 44290
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042759852059686266,
+      "loss": 0.5445,
+      "step": 44300
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042751349742804913,
+      "loss": 0.4893,
+      "step": 44310
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042742847425923566,
+      "loss": 0.5322,
+      "step": 44320
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042734345109042213,
+      "loss": 0.5526,
+      "step": 44330
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004272584279216086,
+      "loss": 0.4588,
+      "step": 44340
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004271734047527952,
+      "loss": 0.4918,
+      "step": 44350
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004270883815839817,
+      "loss": 0.553,
+      "step": 44360
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042700335841516817,
+      "loss": 0.6207,
+      "step": 44370
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042691833524635464,
+      "loss": 0.5353,
+      "step": 44380
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00042683331207754116,
+      "loss": 0.5273,
+      "step": 44390
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00042674828890872763,
+      "loss": 0.5429,
+      "step": 44400
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0004266632657399141,
+      "loss": 0.5626,
+      "step": 44410
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0004265782425711007,
+      "loss": 0.5587,
+      "step": 44420
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00042649321940228715,
+      "loss": 0.5728,
+      "step": 44430
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00042640819623347367,
+      "loss": 0.5837,
+      "step": 44440
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00042632317306466014,
+      "loss": 0.5723,
+      "step": 44450
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0004262381498958466,
+      "loss": 0.5716,
+      "step": 44460
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00042615312672703313,
+      "loss": 0.5023,
+      "step": 44470
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0004260681035582196,
+      "loss": 0.5939,
+      "step": 44480
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0004259830803894062,
+      "loss": 0.5607,
+      "step": 44490
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00042589805722059265,
+      "loss": 0.5861,
+      "step": 44500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0004258130340517791,
+      "loss": 0.5741,
+      "step": 44510
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00042572801088296564,
+      "loss": 0.5378,
+      "step": 44520
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004256429877141521,
+      "loss": 0.5611,
+      "step": 44530
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004255579645453386,
+      "loss": 0.6109,
+      "step": 44540
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004254729413765251,
+      "loss": 0.5317,
+      "step": 44550
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004253879182077116,
+      "loss": 0.5509,
+      "step": 44560
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00042530289503889815,
+      "loss": 0.5928,
+      "step": 44570
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004252178718700846,
+      "loss": 0.5597,
+      "step": 44580
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00042513284870127115,
+      "loss": 0.6022,
+      "step": 44590
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004250478255324576,
+      "loss": 0.5359,
+      "step": 44600
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004249628023636441,
+      "loss": 0.5827,
+      "step": 44610
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00042487777919483056,
+      "loss": 0.5179,
+      "step": 44620
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004247927560260171,
+      "loss": 0.5333,
+      "step": 44630
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042470773285720366,
+      "loss": 0.5664,
+      "step": 44640
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042462270968839013,
+      "loss": 0.5312,
+      "step": 44650
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0004245376865195766,
+      "loss": 0.6013,
+      "step": 44660
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0004244526633507631,
+      "loss": 0.5574,
+      "step": 44670
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0004243676401819496,
+      "loss": 0.5526,
+      "step": 44680
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042428261701313606,
+      "loss": 0.5445,
+      "step": 44690
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0004241975938443226,
+      "loss": 0.5255,
+      "step": 44700
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042411257067550916,
+      "loss": 0.5796,
+      "step": 44710
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042402754750669563,
+      "loss": 0.5815,
+      "step": 44720
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0004239425243378821,
+      "loss": 0.5768,
+      "step": 44730
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00042385750116906857,
+      "loss": 0.4578,
+      "step": 44740
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0004237724780002551,
+      "loss": 0.5241,
+      "step": 44750
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042368745483144156,
+      "loss": 0.5622,
+      "step": 44760
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042360243166262803,
+      "loss": 0.5698,
+      "step": 44770
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004235174084938146,
+      "loss": 0.4889,
+      "step": 44780
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042343238532500114,
+      "loss": 0.5475,
+      "step": 44790
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004233473621561876,
+      "loss": 0.5172,
+      "step": 44800
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004232623389873741,
+      "loss": 0.5613,
+      "step": 44810
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004231773158185606,
+      "loss": 0.6178,
+      "step": 44820
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042309229264974707,
+      "loss": 0.5747,
+      "step": 44830
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042300726948093354,
+      "loss": 0.5111,
+      "step": 44840
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00042292224631212,
+      "loss": 0.5514,
+      "step": 44850
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004228372231433066,
+      "loss": 0.6105,
+      "step": 44860
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004227521999744931,
+      "loss": 0.5834,
+      "step": 44870
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0004226671768056796,
+      "loss": 0.5476,
+      "step": 44880
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00042258215363686605,
+      "loss": 0.5933,
+      "step": 44890
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00042249713046805257,
+      "loss": 0.5189,
+      "step": 44900
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00042241210729923904,
+      "loss": 0.549,
+      "step": 44910
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0004223270841304255,
+      "loss": 0.5715,
+      "step": 44920
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0004222420609616121,
+      "loss": 0.5986,
+      "step": 44930
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00042215703779279856,
+      "loss": 0.5546,
+      "step": 44940
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0004220720146239851,
+      "loss": 0.5678,
+      "step": 44950
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00042198699145517155,
+      "loss": 0.6043,
+      "step": 44960
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.000421901968286358,
+      "loss": 0.5149,
+      "step": 44970
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00042181694511754454,
+      "loss": 0.539,
+      "step": 44980
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.000421731921948731,
+      "loss": 0.5708,
+      "step": 44990
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004216468987799176,
+      "loss": 0.5163,
+      "step": 45000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00042156187561110406,
+      "loss": 0.5543,
+      "step": 45010
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004214768524422906,
+      "loss": 0.5016,
+      "step": 45020
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00042139182927347705,
+      "loss": 0.5435,
+      "step": 45030
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004213068061046635,
+      "loss": 0.5466,
+      "step": 45040
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00042122178293585,
+      "loss": 0.5601,
+      "step": 45050
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004211367597670365,
+      "loss": 0.5215,
+      "step": 45060
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0004210517365982231,
+      "loss": 0.5838,
+      "step": 45070
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00042096671342940956,
+      "loss": 0.5733,
+      "step": 45080
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00042088169026059603,
+      "loss": 0.5494,
+      "step": 45090
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00042079666709178256,
+      "loss": 0.5683,
+      "step": 45100
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00042071164392296903,
+      "loss": 0.4978,
+      "step": 45110
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0004206266207541555,
+      "loss": 0.5767,
+      "step": 45120
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.000420541597585342,
+      "loss": 0.5448,
+      "step": 45130
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0004204565744165285,
+      "loss": 0.5548,
+      "step": 45140
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00042037155124771507,
+      "loss": 0.5598,
+      "step": 45150
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00042028652807890154,
+      "loss": 0.5691,
+      "step": 45160
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.000420201504910088,
+      "loss": 0.557,
+      "step": 45170
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00042011648174127453,
+      "loss": 0.5293,
+      "step": 45180
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.000420031458572461,
+      "loss": 0.5998,
+      "step": 45190
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00041994643540364747,
+      "loss": 0.4781,
+      "step": 45200
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.000419861412234834,
+      "loss": 0.4716,
+      "step": 45210
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00041977638906602057,
+      "loss": 0.5827,
+      "step": 45220
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00041969136589720704,
+      "loss": 0.5491,
+      "step": 45230
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0004196063427283935,
+      "loss": 0.4838,
+      "step": 45240
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00041952131955958,
+      "loss": 0.621,
+      "step": 45250
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0004194362963907665,
+      "loss": 0.5614,
+      "step": 45260
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.000419351273221953,
+      "loss": 0.5601,
+      "step": 45270
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00041926625005313944,
+      "loss": 0.5558,
+      "step": 45280
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.000419181226884326,
+      "loss": 0.5284,
+      "step": 45290
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00041909620371551254,
+      "loss": 0.5586,
+      "step": 45300
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.000419011180546699,
+      "loss": 0.5793,
+      "step": 45310
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0004189261573778855,
+      "loss": 0.5873,
+      "step": 45320
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.000418841134209072,
+      "loss": 0.6222,
+      "step": 45330
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0004187561110402585,
+      "loss": 0.6189,
+      "step": 45340
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00041867108787144495,
+      "loss": 0.5503,
+      "step": 45350
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004185860647026315,
+      "loss": 0.5322,
+      "step": 45360
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.000418501041533818,
+      "loss": 0.4934,
+      "step": 45370
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004184160183650045,
+      "loss": 0.5282,
+      "step": 45380
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.000418330995196191,
+      "loss": 0.539,
+      "step": 45390
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00041824597202737746,
+      "loss": 0.5889,
+      "step": 45400
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.000418160948858564,
+      "loss": 0.57,
+      "step": 45410
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00041807592568975045,
+      "loss": 0.6169,
+      "step": 45420
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004179909025209369,
+      "loss": 0.5859,
+      "step": 45430
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004179058793521235,
+      "loss": 0.5142,
+      "step": 45440
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00041782085618331,
+      "loss": 0.5412,
+      "step": 45450
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004177358330144965,
+      "loss": 0.5846,
+      "step": 45460
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00041765080984568296,
+      "loss": 0.5423,
+      "step": 45470
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00041756578667686943,
+      "loss": 0.5707,
+      "step": 45480
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00041748076350805595,
+      "loss": 0.5494,
+      "step": 45490
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0004173957403392424,
+      "loss": 0.5455,
+      "step": 45500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.000417310717170429,
+      "loss": 0.5226,
+      "step": 45510
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00041722569400161547,
+      "loss": 0.5223,
+      "step": 45520
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.000417140670832802,
+      "loss": 0.5404,
+      "step": 45530
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00041705564766398846,
+      "loss": 0.6091,
+      "step": 45540
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00041697062449517493,
+      "loss": 0.5588,
+      "step": 45550
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00041688560132636146,
+      "loss": 0.5386,
+      "step": 45560
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0004168005781575479,
+      "loss": 0.5824,
+      "step": 45570
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0004167155549887345,
+      "loss": 0.5163,
+      "step": 45580
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.000416630531819921,
+      "loss": 0.6027,
+      "step": 45590
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041654550865110744,
+      "loss": 0.5236,
+      "step": 45600
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041646048548229397,
+      "loss": 0.5835,
+      "step": 45610
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041637546231348044,
+      "loss": 0.4865,
+      "step": 45620
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0004162904391446669,
+      "loss": 0.5199,
+      "step": 45630
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041620541597585343,
+      "loss": 0.5853,
+      "step": 45640
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041612039280704,
+      "loss": 0.5263,
+      "step": 45650
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0004160353696382265,
+      "loss": 0.5787,
+      "step": 45660
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041595034646941295,
+      "loss": 0.5552,
+      "step": 45670
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0004158653233005994,
+      "loss": 0.5137,
+      "step": 45680
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00041578030013178594,
+      "loss": 0.5027,
+      "step": 45690
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0004156952769629724,
+      "loss": 0.5158,
+      "step": 45700
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0004156102537941589,
+      "loss": 0.538,
+      "step": 45710
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0004155252306253454,
+      "loss": 0.5423,
+      "step": 45720
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.000415440207456532,
+      "loss": 0.5391,
+      "step": 45730
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041535518428771845,
+      "loss": 0.6189,
+      "step": 45740
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0004152701611189049,
+      "loss": 0.5656,
+      "step": 45750
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041518513795009144,
+      "loss": 0.5514,
+      "step": 45760
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0004151001147812779,
+      "loss": 0.5665,
+      "step": 45770
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0004150150916124644,
+      "loss": 0.5325,
+      "step": 45780
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041493006844365085,
+      "loss": 0.5599,
+      "step": 45790
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041484504527483743,
+      "loss": 0.5409,
+      "step": 45800
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00041476002210602395,
+      "loss": 0.5298,
+      "step": 45810
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004146749989372104,
+      "loss": 0.5411,
+      "step": 45820
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004145899757683969,
+      "loss": 0.5742,
+      "step": 45830
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004145049525995834,
+      "loss": 0.5643,
+      "step": 45840
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004144199294307699,
+      "loss": 0.5864,
+      "step": 45850
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00041433490626195636,
+      "loss": 0.5399,
+      "step": 45860
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00041424988309314293,
+      "loss": 0.5179,
+      "step": 45870
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004141648599243294,
+      "loss": 0.5885,
+      "step": 45880
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00041407983675551593,
+      "loss": 0.6716,
+      "step": 45890
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004139948135867024,
+      "loss": 0.5295,
+      "step": 45900
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00041390979041788887,
+      "loss": 0.6333,
+      "step": 45910
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0004138247672490754,
+      "loss": 0.5956,
+      "step": 45920
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00041373974408026186,
+      "loss": 0.5148,
+      "step": 45930
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00041365472091144833,
+      "loss": 0.533,
+      "step": 45940
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004135696977426349,
+      "loss": 0.5742,
+      "step": 45950
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00041348467457382143,
+      "loss": 0.4891,
+      "step": 45960
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004133996514050079,
+      "loss": 0.5427,
+      "step": 45970
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00041331462823619437,
+      "loss": 0.5169,
+      "step": 45980
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00041322960506738084,
+      "loss": 0.5141,
+      "step": 45990
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00041314458189856736,
+      "loss": 0.5588,
+      "step": 46000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00041305955872975383,
+      "loss": 0.578,
+      "step": 46010
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004129745355609404,
+      "loss": 0.557,
+      "step": 46020
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004128895123921269,
+      "loss": 0.5384,
+      "step": 46030
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004128044892233134,
+      "loss": 0.5395,
+      "step": 46040
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0004127194660544999,
+      "loss": 0.5358,
+      "step": 46050
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041263444288568634,
+      "loss": 0.5266,
+      "step": 46060
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041254941971687287,
+      "loss": 0.5763,
+      "step": 46070
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041246439654805934,
+      "loss": 0.4935,
+      "step": 46080
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0004123793733792459,
+      "loss": 0.6095,
+      "step": 46090
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0004122943502104324,
+      "loss": 0.542,
+      "step": 46100
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041220932704161885,
+      "loss": 0.5159,
+      "step": 46110
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0004121243038728054,
+      "loss": 0.5297,
+      "step": 46120
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041203928070399185,
+      "loss": 0.6387,
+      "step": 46130
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0004119542575351783,
+      "loss": 0.5231,
+      "step": 46140
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00041186923436636484,
+      "loss": 0.6043,
+      "step": 46150
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0004117842111975514,
+      "loss": 0.5293,
+      "step": 46160
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0004116991880287379,
+      "loss": 0.6166,
+      "step": 46170
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00041161416485992436,
+      "loss": 0.5868,
+      "step": 46180
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004115291416911108,
+      "loss": 0.6174,
+      "step": 46190
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00041144411852229735,
+      "loss": 0.6012,
+      "step": 46200
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004113590953534838,
+      "loss": 0.5817,
+      "step": 46210
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004112740721846703,
+      "loss": 0.5413,
+      "step": 46220
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004111890490158568,
+      "loss": 0.6067,
+      "step": 46230
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004111040258470434,
+      "loss": 0.5603,
+      "step": 46240
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00041101900267822986,
+      "loss": 0.5631,
+      "step": 46250
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00041093397950941633,
+      "loss": 0.5672,
+      "step": 46260
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00041084895634060285,
+      "loss": 0.5679,
+      "step": 46270
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004107639331717893,
+      "loss": 0.5403,
+      "step": 46280
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0004106789100029758,
+      "loss": 0.5557,
+      "step": 46290
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00041059388683416226,
+      "loss": 0.5515,
+      "step": 46300
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00041050886366534884,
+      "loss": 0.5717,
+      "step": 46310
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00041042384049653536,
+      "loss": 0.548,
+      "step": 46320
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00041033881732772183,
+      "loss": 0.6124,
+      "step": 46330
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0004102537941589083,
+      "loss": 0.5385,
+      "step": 46340
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0004101687709900948,
+      "loss": 0.5027,
+      "step": 46350
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0004100837478212813,
+      "loss": 0.5453,
+      "step": 46360
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00040999872465246777,
+      "loss": 0.526,
+      "step": 46370
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00040991370148365434,
+      "loss": 0.6104,
+      "step": 46380
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00040982867831484087,
+      "loss": 0.5567,
+      "step": 46390
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00040974365514602734,
+      "loss": 0.5132,
+      "step": 46400
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004096586319772138,
+      "loss": 0.5934,
+      "step": 46410
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004095736088084003,
+      "loss": 0.5653,
+      "step": 46420
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004094885856395868,
+      "loss": 0.5379,
+      "step": 46430
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00040940356247077327,
+      "loss": 0.5525,
+      "step": 46440
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00040931853930195985,
+      "loss": 0.5625,
+      "step": 46450
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004092335161331463,
+      "loss": 0.5479,
+      "step": 46460
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00040914849296433284,
+      "loss": 0.5658,
+      "step": 46470
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004090634697955193,
+      "loss": 0.5575,
+      "step": 46480
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004089784466267058,
+      "loss": 0.6363,
+      "step": 46490
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0004088934234578923,
+      "loss": 0.5367,
+      "step": 46500
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00040880840028907877,
+      "loss": 0.6028,
+      "step": 46510
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00040872337712026524,
+      "loss": 0.5326,
+      "step": 46520
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004086383539514518,
+      "loss": 0.5327,
+      "step": 46530
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004085533307826383,
+      "loss": 0.5786,
+      "step": 46540
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004084683076138248,
+      "loss": 0.6032,
+      "step": 46550
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004083832844450113,
+      "loss": 0.5851,
+      "step": 46560
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00040829826127619775,
+      "loss": 0.6131,
+      "step": 46570
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004082132381073843,
+      "loss": 0.5754,
+      "step": 46580
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00040812821493857075,
+      "loss": 0.4812,
+      "step": 46590
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004080431917697573,
+      "loss": 0.5996,
+      "step": 46600
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004079581686009438,
+      "loss": 0.6135,
+      "step": 46610
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00040787314543213026,
+      "loss": 0.52,
+      "step": 46620
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0004077881222633168,
+      "loss": 0.5047,
+      "step": 46630
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00040770309909450326,
+      "loss": 0.6414,
+      "step": 46640
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004076180759256897,
+      "loss": 0.5076,
+      "step": 46650
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040753305275687625,
+      "loss": 0.5545,
+      "step": 46660
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040744802958806283,
+      "loss": 0.5373,
+      "step": 46670
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004073630064192493,
+      "loss": 0.5396,
+      "step": 46680
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040727798325043577,
+      "loss": 0.6177,
+      "step": 46690
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004071929600816223,
+      "loss": 0.5356,
+      "step": 46700
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040710793691280876,
+      "loss": 0.5564,
+      "step": 46710
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040702291374399523,
+      "loss": 0.5586,
+      "step": 46720
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004069378905751817,
+      "loss": 0.5714,
+      "step": 46730
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004068528674063683,
+      "loss": 0.5919,
+      "step": 46740
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0004067678442375548,
+      "loss": 0.6554,
+      "step": 46750
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00040668282106874127,
+      "loss": 0.5681,
+      "step": 46760
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00040659779789992774,
+      "loss": 0.4971,
+      "step": 46770
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00040651277473111426,
+      "loss": 0.5651,
+      "step": 46780
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00040642775156230073,
+      "loss": 0.561,
+      "step": 46790
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004063427283934872,
+      "loss": 0.609,
+      "step": 46800
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004062577052246737,
+      "loss": 0.5427,
+      "step": 46810
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00040617268205586025,
+      "loss": 0.5541,
+      "step": 46820
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004060876588870468,
+      "loss": 0.5954,
+      "step": 46830
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00040600263571823324,
+      "loss": 0.545,
+      "step": 46840
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004059176125494197,
+      "loss": 0.5901,
+      "step": 46850
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00040583258938060624,
+      "loss": 0.5594,
+      "step": 46860
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004057475662117927,
+      "loss": 0.5354,
+      "step": 46870
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0004056625430429792,
+      "loss": 0.5568,
+      "step": 46880
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00040557751987416575,
+      "loss": 0.5015,
+      "step": 46890
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004054924967053523,
+      "loss": 0.5686,
+      "step": 46900
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00040540747353653875,
+      "loss": 0.6679,
+      "step": 46910
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004053224503677252,
+      "loss": 0.5268,
+      "step": 46920
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004052374271989117,
+      "loss": 0.5373,
+      "step": 46930
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004051524040300982,
+      "loss": 0.6443,
+      "step": 46940
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004050673808612847,
+      "loss": 0.5703,
+      "step": 46950
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00040498235769247126,
+      "loss": 0.5351,
+      "step": 46960
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004048973345236577,
+      "loss": 0.5721,
+      "step": 46970
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00040481231135484425,
+      "loss": 0.5328,
+      "step": 46980
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004047272881860307,
+      "loss": 0.53,
+      "step": 46990
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004046422650172172,
+      "loss": 0.4566,
+      "step": 47000
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004045572418484037,
+      "loss": 0.5837,
+      "step": 47010
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004044722186795902,
+      "loss": 0.5203,
+      "step": 47020
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00040438719551077665,
+      "loss": 0.5618,
+      "step": 47030
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00040430217234196323,
+      "loss": 0.506,
+      "step": 47040
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004042171491731497,
+      "loss": 0.5264,
+      "step": 47050
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004041321260043362,
+      "loss": 0.5998,
+      "step": 47060
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004040471028355227,
+      "loss": 0.5472,
+      "step": 47070
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00040396207966670916,
+      "loss": 0.5207,
+      "step": 47080
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0004038770564978957,
+      "loss": 0.49,
+      "step": 47090
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00040379203332908216,
+      "loss": 0.5846,
+      "step": 47100
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00040370701016026873,
+      "loss": 0.553,
+      "step": 47110
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004036219869914552,
+      "loss": 0.524,
+      "step": 47120
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004035369638226417,
+      "loss": 0.5706,
+      "step": 47130
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004034519406538282,
+      "loss": 0.555,
+      "step": 47140
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00040336691748501467,
+      "loss": 0.4921,
+      "step": 47150
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00040328189431620114,
+      "loss": 0.5389,
+      "step": 47160
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00040319687114738766,
+      "loss": 0.5781,
+      "step": 47170
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00040311184797857424,
+      "loss": 0.5521,
+      "step": 47180
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004030268248097607,
+      "loss": 0.4737,
+      "step": 47190
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004029418016409472,
+      "loss": 0.5755,
+      "step": 47200
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0004028567784721337,
+      "loss": 0.5325,
+      "step": 47210
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00040277175530332017,
+      "loss": 0.6178,
+      "step": 47220
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00040268673213450664,
+      "loss": 0.5854,
+      "step": 47230
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040260170896569316,
+      "loss": 0.5533,
+      "step": 47240
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004025166857968797,
+      "loss": 0.5466,
+      "step": 47250
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004024316626280662,
+      "loss": 0.5224,
+      "step": 47260
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004023466394592527,
+      "loss": 0.6487,
+      "step": 47270
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040226161629043915,
+      "loss": 0.5231,
+      "step": 47280
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040217659312162567,
+      "loss": 0.542,
+      "step": 47290
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040209156995281214,
+      "loss": 0.5744,
+      "step": 47300
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004020065467839986,
+      "loss": 0.5541,
+      "step": 47310
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040192152361518514,
+      "loss": 0.575,
+      "step": 47320
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004018365004463717,
+      "loss": 0.504,
+      "step": 47330
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004017514772775582,
+      "loss": 0.5129,
+      "step": 47340
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00040166645410874465,
+      "loss": 0.4983,
+      "step": 47350
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0004015814309399311,
+      "loss": 0.4959,
+      "step": 47360
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00040149640777111765,
+      "loss": 0.5099,
+      "step": 47370
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0004014113846023041,
+      "loss": 0.4561,
+      "step": 47380
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0004013263614334906,
+      "loss": 0.5322,
+      "step": 47390
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00040124133826467716,
+      "loss": 0.5043,
+      "step": 47400
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0004011563150958637,
+      "loss": 0.5553,
+      "step": 47410
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00040107129192705016,
+      "loss": 0.4948,
+      "step": 47420
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0004009862687582366,
+      "loss": 0.5122,
+      "step": 47430
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00040090124558942315,
+      "loss": 0.5433,
+      "step": 47440
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0004008162224206096,
+      "loss": 0.5115,
+      "step": 47450
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0004007311992517961,
+      "loss": 0.5001,
+      "step": 47460
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00040064617608298267,
+      "loss": 0.5886,
+      "step": 47470
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00040056115291416914,
+      "loss": 0.4754,
+      "step": 47480
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00040047612974535566,
+      "loss": 0.5046,
+      "step": 47490
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00040039110657654213,
+      "loss": 0.5278,
+      "step": 47500
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0004003060834077286,
+      "loss": 0.5262,
+      "step": 47510
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0004002210602389151,
+      "loss": 0.5387,
+      "step": 47520
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0004001360370701016,
+      "loss": 0.5272,
+      "step": 47530
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00040005101390128817,
+      "loss": 0.5275,
+      "step": 47540
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003999659907324746,
+      "loss": 0.4707,
+      "step": 47550
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003998809675636611,
+      "loss": 0.5338,
+      "step": 47560
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00039979594439484763,
+      "loss": 0.5407,
+      "step": 47570
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0003997109212260341,
+      "loss": 0.5458,
+      "step": 47580
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00039962589805722057,
+      "loss": 0.5041,
+      "step": 47590
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00039954087488840715,
+      "loss": 0.5596,
+      "step": 47600
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003994558517195936,
+      "loss": 0.5132,
+      "step": 47610
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003993708285507801,
+      "loss": 0.5309,
+      "step": 47620
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003992858053819666,
+      "loss": 0.5108,
+      "step": 47630
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00039920078221315314,
+      "loss": 0.5019,
+      "step": 47640
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003991157590443396,
+      "loss": 0.4778,
+      "step": 47650
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003990307358755261,
+      "loss": 0.4807,
+      "step": 47660
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003989457127067126,
+      "loss": 0.4853,
+      "step": 47670
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003988606895378991,
+      "loss": 0.5051,
+      "step": 47680
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0003987756663690856,
+      "loss": 0.5446,
+      "step": 47690
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00039869064320027206,
+      "loss": 0.4685,
+      "step": 47700
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003986056200314586,
+      "loss": 0.478,
+      "step": 47710
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003985205968626451,
+      "loss": 0.5307,
+      "step": 47720
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003984355736938316,
+      "loss": 0.531,
+      "step": 47730
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003983505505250181,
+      "loss": 0.489,
+      "step": 47740
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00039826552735620457,
+      "loss": 0.4547,
+      "step": 47750
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003981805041873911,
+      "loss": 0.5327,
+      "step": 47760
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00039809548101857757,
+      "loss": 0.5654,
+      "step": 47770
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003980104578497641,
+      "loss": 0.5438,
+      "step": 47780
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00039792543468095056,
+      "loss": 0.4506,
+      "step": 47790
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003978404115121371,
+      "loss": 0.5042,
+      "step": 47800
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00039775538834332355,
+      "loss": 0.5635,
+      "step": 47810
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0003976703651745101,
+      "loss": 0.5536,
+      "step": 47820
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039758534200569655,
+      "loss": 0.4808,
+      "step": 47830
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039750031883688307,
+      "loss": 0.5102,
+      "step": 47840
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0003974152956680696,
+      "loss": 0.4828,
+      "step": 47850
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039733027249925606,
+      "loss": 0.557,
+      "step": 47860
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0003972452493304426,
+      "loss": 0.513,
+      "step": 47870
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039716022616162906,
+      "loss": 0.5421,
+      "step": 47880
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0003970752029928156,
+      "loss": 0.4358,
+      "step": 47890
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039699017982400205,
+      "loss": 0.513,
+      "step": 47900
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039690515665518857,
+      "loss": 0.5378,
+      "step": 47910
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039682013348637504,
+      "loss": 0.5417,
+      "step": 47920
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039673511031756157,
+      "loss": 0.4718,
+      "step": 47930
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00039665008714874804,
+      "loss": 0.4864,
+      "step": 47940
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00039656506397993456,
+      "loss": 0.56,
+      "step": 47950
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0003964800408111211,
+      "loss": 0.4849,
+      "step": 47960
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00039639501764230755,
+      "loss": 0.5097,
+      "step": 47970
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.000396309994473494,
+      "loss": 0.5404,
+      "step": 47980
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00039622497130468055,
+      "loss": 0.5549,
+      "step": 47990
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00039613994813586707,
+      "loss": 0.5138,
+      "step": 48000
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00039605492496705354,
+      "loss": 0.4949,
+      "step": 48010
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00039596990179824,
+      "loss": 0.4631,
+      "step": 48020
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00039588487862942653,
+      "loss": 0.5429,
+      "step": 48030
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00039579985546061306,
+      "loss": 0.53,
+      "step": 48040
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0003957148322917995,
+      "loss": 0.4472,
+      "step": 48050
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.000395629809122986,
+      "loss": 0.5306,
+      "step": 48060
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00039554478595417257,
+      "loss": 0.5157,
+      "step": 48070
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00039545976278535904,
+      "loss": 0.5539,
+      "step": 48080
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0003953747396165455,
+      "loss": 0.549,
+      "step": 48090
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.000395289716447732,
+      "loss": 0.5889,
+      "step": 48100
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00039520469327891856,
+      "loss": 0.5521,
+      "step": 48110
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00039511967011010503,
+      "loss": 0.4859,
+      "step": 48120
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0003950346469412915,
+      "loss": 0.4896,
+      "step": 48130
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.000394949623772478,
+      "loss": 0.5192,
+      "step": 48140
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00039486460060366455,
+      "loss": 0.487,
+      "step": 48150
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.000394779577434851,
+      "loss": 0.5343,
+      "step": 48160
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0003946945542660375,
+      "loss": 0.5117,
+      "step": 48170
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.000394609531097224,
+      "loss": 0.5235,
+      "step": 48180
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039452450792841053,
+      "loss": 0.4943,
+      "step": 48190
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.000394439484759597,
+      "loss": 0.556,
+      "step": 48200
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039435446159078347,
+      "loss": 0.5142,
+      "step": 48210
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039426943842197,
+      "loss": 0.5347,
+      "step": 48220
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0003941844152531565,
+      "loss": 0.5359,
+      "step": 48230
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.000394099392084343,
+      "loss": 0.5059,
+      "step": 48240
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0003940143689155295,
+      "loss": 0.5378,
+      "step": 48250
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.000393929345746716,
+      "loss": 0.5729,
+      "step": 48260
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0003938443225779025,
+      "loss": 0.4962,
+      "step": 48270
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.000393759299409089,
+      "loss": 0.5132,
+      "step": 48280
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0003936742762402755,
+      "loss": 0.467,
+      "step": 48290
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00039358925307146197,
+      "loss": 0.5328,
+      "step": 48300
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0003935042299026485,
+      "loss": 0.5395,
+      "step": 48310
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.000393419206733835,
+      "loss": 0.4733,
+      "step": 48320
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0003933341835650215,
+      "loss": 0.4943,
+      "step": 48330
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.000393249160396208,
+      "loss": 0.5664,
+      "step": 48340
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0003931641372273945,
+      "loss": 0.5031,
+      "step": 48350
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.000393079114058581,
+      "loss": 0.5804,
+      "step": 48360
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00039299409088976747,
+      "loss": 0.5255,
+      "step": 48370
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.000392909067720954,
+      "loss": 0.5227,
+      "step": 48380
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00039282404455214047,
+      "loss": 0.4866,
+      "step": 48390
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.000392739021383327,
+      "loss": 0.4712,
+      "step": 48400
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00039265399821451346,
+      "loss": 0.5055,
+      "step": 48410
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0003925689750457,
+      "loss": 0.4995,
+      "step": 48420
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0003924839518768865,
+      "loss": 0.5401,
+      "step": 48430
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.000392398928708073,
+      "loss": 0.4945,
+      "step": 48440
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039231390553925944,
+      "loss": 0.477,
+      "step": 48450
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039222888237044597,
+      "loss": 0.5595,
+      "step": 48460
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0003921438592016325,
+      "loss": 0.5466,
+      "step": 48470
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039205883603281896,
+      "loss": 0.4675,
+      "step": 48480
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039197381286400543,
+      "loss": 0.534,
+      "step": 48490
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039188878969519196,
+      "loss": 0.4957,
+      "step": 48500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0003918037665263785,
+      "loss": 0.5814,
+      "step": 48510
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00039171874335756495,
+      "loss": 0.5162,
+      "step": 48520
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0003916337201887514,
+      "loss": 0.5295,
+      "step": 48530
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.000391548697019938,
+      "loss": 0.4603,
+      "step": 48540
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00039146367385112447,
+      "loss": 0.5067,
+      "step": 48550
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00039137865068231093,
+      "loss": 0.5341,
+      "step": 48560
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0003912936275134974,
+      "loss": 0.5091,
+      "step": 48570
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.000391208604344684,
+      "loss": 0.5114,
+      "step": 48580
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00039112358117587045,
+      "loss": 0.6221,
+      "step": 48590
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0003910385580070569,
+      "loss": 0.5616,
+      "step": 48600
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00039095353483824345,
+      "loss": 0.4702,
+      "step": 48610
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00039086851166942997,
+      "loss": 0.4831,
+      "step": 48620
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00039078348850061644,
+      "loss": 0.4837,
+      "step": 48630
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0003906984653318029,
+      "loss": 0.5806,
+      "step": 48640
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00039061344216298943,
+      "loss": 0.4626,
+      "step": 48650
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00039052841899417596,
+      "loss": 0.4812,
+      "step": 48660
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0003904433958253624,
+      "loss": 0.5074,
+      "step": 48670
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0003903583726565489,
+      "loss": 0.5122,
+      "step": 48680
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0003902733494877354,
+      "loss": 0.4967,
+      "step": 48690
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00039018832631892194,
+      "loss": 0.557,
+      "step": 48700
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0003901033031501084,
+      "loss": 0.5896,
+      "step": 48710
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00039001827998129494,
+      "loss": 0.5244,
+      "step": 48720
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0003899332568124814,
+      "loss": 0.4933,
+      "step": 48730
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00038984823364366793,
+      "loss": 0.5041,
+      "step": 48740
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0003897632104748544,
+      "loss": 0.5095,
+      "step": 48750
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0003896781873060409,
+      "loss": 0.4585,
+      "step": 48760
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0003895931641372274,
+      "loss": 0.5329,
+      "step": 48770
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003895081409684139,
+      "loss": 0.5089,
+      "step": 48780
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003894231177996004,
+      "loss": 0.5491,
+      "step": 48790
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003893380946307869,
+      "loss": 0.5183,
+      "step": 48800
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00038925307146197343,
+      "loss": 0.5114,
+      "step": 48810
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003891680482931599,
+      "loss": 0.508,
+      "step": 48820
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003890830251243464,
+      "loss": 0.5253,
+      "step": 48830
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003889980019555329,
+      "loss": 0.468,
+      "step": 48840
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003889129787867194,
+      "loss": 0.458,
+      "step": 48850
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003888279556179059,
+      "loss": 0.5555,
+      "step": 48860
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003887429324490924,
+      "loss": 0.522,
+      "step": 48870
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003886579092802789,
+      "loss": 0.4606,
+      "step": 48880
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0003885728861114654,
+      "loss": 0.5438,
+      "step": 48890
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003884878629426519,
+      "loss": 0.493,
+      "step": 48900
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003884028397738384,
+      "loss": 0.5221,
+      "step": 48910
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00038831781660502487,
+      "loss": 0.5099,
+      "step": 48920
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003882327934362114,
+      "loss": 0.5203,
+      "step": 48930
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003881477702673979,
+      "loss": 0.5618,
+      "step": 48940
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003880627470985844,
+      "loss": 0.503,
+      "step": 48950
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00038797772392977085,
+      "loss": 0.4795,
+      "step": 48960
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003878927007609574,
+      "loss": 0.5749,
+      "step": 48970
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003878076775921439,
+      "loss": 0.5158,
+      "step": 48980
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00038772265442333037,
+      "loss": 0.573,
+      "step": 48990
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00038763763125451684,
+      "loss": 0.4988,
+      "step": 49000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0003875526080857034,
+      "loss": 0.5515,
+      "step": 49010
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0003874675849168899,
+      "loss": 0.5138,
+      "step": 49020
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038738256174807636,
+      "loss": 0.4846,
+      "step": 49030
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038729753857926283,
+      "loss": 0.4985,
+      "step": 49040
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0003872125154104494,
+      "loss": 0.5018,
+      "step": 49050
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0003871274922416359,
+      "loss": 0.5082,
+      "step": 49060
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038704246907282234,
+      "loss": 0.4648,
+      "step": 49070
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038695744590400887,
+      "loss": 0.5279,
+      "step": 49080
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0003868724227351954,
+      "loss": 0.4841,
+      "step": 49090
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038678739956638186,
+      "loss": 0.5621,
+      "step": 49100
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038670237639756833,
+      "loss": 0.5244,
+      "step": 49110
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00038661735322875485,
+      "loss": 0.5275,
+      "step": 49120
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0003865323300599414,
+      "loss": 0.4733,
+      "step": 49130
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00038644730689112785,
+      "loss": 0.5448,
+      "step": 49140
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0003863622837223143,
+      "loss": 0.5019,
+      "step": 49150
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00038627726055350084,
+      "loss": 0.5729,
+      "step": 49160
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00038619223738468737,
+      "loss": 0.4862,
+      "step": 49170
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00038610721421587383,
+      "loss": 0.5593,
+      "step": 49180
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0003860221910470603,
+      "loss": 0.447,
+      "step": 49190
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00038593716787824683,
+      "loss": 0.4886,
+      "step": 49200
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00038585214470943335,
+      "loss": 0.531,
+      "step": 49210
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0003857671215406198,
+      "loss": 0.4613,
+      "step": 49220
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00038568209837180634,
+      "loss": 0.51,
+      "step": 49230
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0003855970752029928,
+      "loss": 0.5681,
+      "step": 49240
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00038551205203417934,
+      "loss": 0.4468,
+      "step": 49250
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003854270288653658,
+      "loss": 0.4846,
+      "step": 49260
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00038534200569655233,
+      "loss": 0.4849,
+      "step": 49270
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00038525698252773886,
+      "loss": 0.5406,
+      "step": 49280
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003851719593589253,
+      "loss": 0.5076,
+      "step": 49290
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003850869361901118,
+      "loss": 0.5053,
+      "step": 49300
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003850019130212983,
+      "loss": 0.5081,
+      "step": 49310
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00038491688985248484,
+      "loss": 0.5248,
+      "step": 49320
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003848318666836713,
+      "loss": 0.5326,
+      "step": 49330
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00038474684351485784,
+      "loss": 0.6058,
+      "step": 49340
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0003846618203460443,
+      "loss": 0.5645,
+      "step": 49350
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00038457679717723083,
+      "loss": 0.5645,
+      "step": 49360
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003844917740084173,
+      "loss": 0.5805,
+      "step": 49370
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003844067508396038,
+      "loss": 0.4683,
+      "step": 49380
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003843217276707903,
+      "loss": 0.514,
+      "step": 49390
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003842367045019768,
+      "loss": 0.5271,
+      "step": 49400
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00038415168133316334,
+      "loss": 0.5792,
+      "step": 49410
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003840666581643498,
+      "loss": 0.5498,
+      "step": 49420
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003839816349955363,
+      "loss": 0.5119,
+      "step": 49430
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003838966118267228,
+      "loss": 0.4433,
+      "step": 49440
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003838115886579093,
+      "loss": 0.5217,
+      "step": 49450
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003837265654890958,
+      "loss": 0.5517,
+      "step": 49460
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00038364154232028226,
+      "loss": 0.4892,
+      "step": 49470
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0003835565191514688,
+      "loss": 0.5177,
+      "step": 49480
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0003834714959826553,
+      "loss": 0.5095,
+      "step": 49490
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0003833864728138418,
+      "loss": 0.4929,
+      "step": 49500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038330144964502825,
+      "loss": 0.5059,
+      "step": 49510
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038321642647621483,
+      "loss": 0.5945,
+      "step": 49520
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0003831314033074013,
+      "loss": 0.5121,
+      "step": 49530
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038304638013858777,
+      "loss": 0.5618,
+      "step": 49540
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0003829613569697743,
+      "loss": 0.4923,
+      "step": 49550
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0003828763338009608,
+      "loss": 0.4494,
+      "step": 49560
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0003827913106321473,
+      "loss": 0.5222,
+      "step": 49570
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00038270628746333375,
+      "loss": 0.4434,
+      "step": 49580
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0003826212642945203,
+      "loss": 0.5156,
+      "step": 49590
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0003825362411257068,
+      "loss": 0.5501,
+      "step": 49600
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038245121795689327,
+      "loss": 0.5116,
+      "step": 49610
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038236619478807974,
+      "loss": 0.5711,
+      "step": 49620
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038228117161926626,
+      "loss": 0.4923,
+      "step": 49630
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0003821961484504528,
+      "loss": 0.4807,
+      "step": 49640
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038211112528163926,
+      "loss": 0.5865,
+      "step": 49650
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038202610211282573,
+      "loss": 0.5381,
+      "step": 49660
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038194107894401225,
+      "loss": 0.5412,
+      "step": 49670
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0003818560557751988,
+      "loss": 0.5531,
+      "step": 49680
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038177103260638524,
+      "loss": 0.5077,
+      "step": 49690
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00038168600943757177,
+      "loss": 0.5457,
+      "step": 49700
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0003816009862687583,
+      "loss": 0.5395,
+      "step": 49710
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00038151596309994476,
+      "loss": 0.4857,
+      "step": 49720
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00038143093993113123,
+      "loss": 0.5506,
+      "step": 49730
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00038134591676231775,
+      "loss": 0.5028,
+      "step": 49740
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0003812608935935043,
+      "loss": 0.5279,
+      "step": 49750
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00038117587042469075,
+      "loss": 0.531,
+      "step": 49760
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0003810908472558772,
+      "loss": 0.5649,
+      "step": 49770
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00038100582408706374,
+      "loss": 0.5874,
+      "step": 49780
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00038092080091825026,
+      "loss": 0.4958,
+      "step": 49790
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00038083577774943673,
+      "loss": 0.5369,
+      "step": 49800
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00038075075458062326,
+      "loss": 0.5437,
+      "step": 49810
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00038066573141180973,
+      "loss": 0.5719,
+      "step": 49820
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00038058070824299625,
+      "loss": 0.5036,
+      "step": 49830
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003804956850741827,
+      "loss": 0.5478,
+      "step": 49840
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00038041066190536924,
+      "loss": 0.5353,
+      "step": 49850
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003803256387365557,
+      "loss": 0.5371,
+      "step": 49860
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00038024061556774224,
+      "loss": 0.5648,
+      "step": 49870
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003801555923989287,
+      "loss": 0.4697,
+      "step": 49880
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00038007056923011523,
+      "loss": 0.4662,
+      "step": 49890
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003799855460613017,
+      "loss": 0.5538,
+      "step": 49900
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003799005228924882,
+      "loss": 0.4544,
+      "step": 49910
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00037981549972367475,
+      "loss": 0.5072,
+      "step": 49920
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003797304765548612,
+      "loss": 0.5096,
+      "step": 49930
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003796454533860477,
+      "loss": 0.513,
+      "step": 49940
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0003795604302172342,
+      "loss": 0.5213,
+      "step": 49950
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00037947540704842073,
+      "loss": 0.5228,
+      "step": 49960
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003793903838796072,
+      "loss": 0.5387,
+      "step": 49970
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003793053607107937,
+      "loss": 0.5422,
+      "step": 49980
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003792203375419802,
+      "loss": 0.5211,
+      "step": 49990
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003791353143731667,
+      "loss": 0.5563,
+      "step": 50000
+    },
+    {
+      "epoch": 4.23,
+      "eval_loss": 0.5834364295005798,
+      "eval_runtime": 312.7191,
+      "eval_samples_per_second": 16.804,
+      "eval_steps_per_second": 2.101,
+      "step": 50000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003790502912043532,
+      "loss": 0.5687,
+      "step": 50010
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003789652680355397,
+      "loss": 0.5102,
+      "step": 50020
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00037888024486672624,
+      "loss": 0.5121,
+      "step": 50030
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003787952216979127,
+      "loss": 0.497,
+      "step": 50040
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003787101985290992,
+      "loss": 0.5719,
+      "step": 50050
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003786251753602857,
+      "loss": 0.4977,
+      "step": 50060
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0003785401521914722,
+      "loss": 0.5558,
+      "step": 50070
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003784551290226587,
+      "loss": 0.5525,
+      "step": 50080
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00037837010585384516,
+      "loss": 0.5078,
+      "step": 50090
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003782850826850317,
+      "loss": 0.4713,
+      "step": 50100
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003782000595162182,
+      "loss": 0.5118,
+      "step": 50110
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003781150363474047,
+      "loss": 0.5293,
+      "step": 50120
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00037803001317859115,
+      "loss": 0.5232,
+      "step": 50130
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003779449900097777,
+      "loss": 0.4851,
+      "step": 50140
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003778599668409642,
+      "loss": 0.5069,
+      "step": 50150
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00037777494367215067,
+      "loss": 0.5427,
+      "step": 50160
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00037768992050333714,
+      "loss": 0.5017,
+      "step": 50170
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003776048973345237,
+      "loss": 0.5185,
+      "step": 50180
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0003775198741657102,
+      "loss": 0.5139,
+      "step": 50190
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00037743485099689665,
+      "loss": 0.4598,
+      "step": 50200
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003773498278280832,
+      "loss": 0.5009,
+      "step": 50210
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003772648046592697,
+      "loss": 0.5383,
+      "step": 50220
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00037717978149045617,
+      "loss": 0.5955,
+      "step": 50230
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00037709475832164264,
+      "loss": 0.4999,
+      "step": 50240
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00037700973515282916,
+      "loss": 0.5299,
+      "step": 50250
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003769247119840157,
+      "loss": 0.4909,
+      "step": 50260
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00037683968881520216,
+      "loss": 0.552,
+      "step": 50270
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003767546656463886,
+      "loss": 0.513,
+      "step": 50280
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00037666964247757515,
+      "loss": 0.5527,
+      "step": 50290
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0003765846193087617,
+      "loss": 0.4537,
+      "step": 50300
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00037649959613994814,
+      "loss": 0.5227,
+      "step": 50310
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037641457297113467,
+      "loss": 0.5339,
+      "step": 50320
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037632954980232114,
+      "loss": 0.5364,
+      "step": 50330
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037624452663350766,
+      "loss": 0.5362,
+      "step": 50340
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037615950346469413,
+      "loss": 0.4797,
+      "step": 50350
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037607448029588065,
+      "loss": 0.5956,
+      "step": 50360
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0003759894571270671,
+      "loss": 0.5496,
+      "step": 50370
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037590443395825365,
+      "loss": 0.5021,
+      "step": 50380
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0003758194107894401,
+      "loss": 0.4639,
+      "step": 50390
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037573438762062664,
+      "loss": 0.5933,
+      "step": 50400
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0003756493644518131,
+      "loss": 0.5584,
+      "step": 50410
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00037556434128299963,
+      "loss": 0.5152,
+      "step": 50420
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037547931811418616,
+      "loss": 0.5776,
+      "step": 50430
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037539429494537263,
+      "loss": 0.4903,
+      "step": 50440
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0003753092717765591,
+      "loss": 0.5462,
+      "step": 50450
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0003752242486077456,
+      "loss": 0.5005,
+      "step": 50460
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037513922543893214,
+      "loss": 0.4852,
+      "step": 50470
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0003750542022701186,
+      "loss": 0.5259,
+      "step": 50480
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037496917910130514,
+      "loss": 0.4937,
+      "step": 50490
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037488415593249166,
+      "loss": 0.5692,
+      "step": 50500
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037479913276367813,
+      "loss": 0.5924,
+      "step": 50510
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0003747141095948646,
+      "loss": 0.5176,
+      "step": 50520
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0003746290864260511,
+      "loss": 0.5091,
+      "step": 50530
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00037454406325723765,
+      "loss": 0.4914,
+      "step": 50540
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003744590400884241,
+      "loss": 0.474,
+      "step": 50550
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003743740169196106,
+      "loss": 0.5483,
+      "step": 50560
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003742889937507971,
+      "loss": 0.5008,
+      "step": 50570
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00037420397058198363,
+      "loss": 0.4729,
+      "step": 50580
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003741189474131701,
+      "loss": 0.452,
+      "step": 50590
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003740339242443566,
+      "loss": 0.4449,
+      "step": 50600
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003739489010755431,
+      "loss": 0.5247,
+      "step": 50610
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003738638779067296,
+      "loss": 0.5265,
+      "step": 50620
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003737788547379161,
+      "loss": 0.5782,
+      "step": 50630
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00037369383156910256,
+      "loss": 0.5323,
+      "step": 50640
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00037360880840028914,
+      "loss": 0.4915,
+      "step": 50650
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0003735237852314756,
+      "loss": 0.5224,
+      "step": 50660
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003734387620626621,
+      "loss": 0.4651,
+      "step": 50670
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00037335373889384855,
+      "loss": 0.5505,
+      "step": 50680
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003732687157250351,
+      "loss": 0.5573,
+      "step": 50690
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003731836925562216,
+      "loss": 0.5441,
+      "step": 50700
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00037309866938740806,
+      "loss": 0.5432,
+      "step": 50710
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003730136462185946,
+      "loss": 0.5289,
+      "step": 50720
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003729286230497811,
+      "loss": 0.4923,
+      "step": 50730
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003728435998809676,
+      "loss": 0.4856,
+      "step": 50740
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00037275857671215405,
+      "loss": 0.5142,
+      "step": 50750
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003726735535433406,
+      "loss": 0.4878,
+      "step": 50760
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0003725885303745271,
+      "loss": 0.525,
+      "step": 50770
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00037250350720571357,
+      "loss": 0.5305,
+      "step": 50780
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0003724184840369001,
+      "loss": 0.488,
+      "step": 50790
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037233346086808656,
+      "loss": 0.6538,
+      "step": 50800
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0003722484376992731,
+      "loss": 0.5237,
+      "step": 50810
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037216341453045955,
+      "loss": 0.5391,
+      "step": 50820
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0003720783913616461,
+      "loss": 0.4358,
+      "step": 50830
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037199336819283255,
+      "loss": 0.571,
+      "step": 50840
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037190834502401907,
+      "loss": 0.4881,
+      "step": 50850
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037182332185520554,
+      "loss": 0.5177,
+      "step": 50860
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037173829868639206,
+      "loss": 0.5264,
+      "step": 50870
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037165327551757853,
+      "loss": 0.5432,
+      "step": 50880
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00037156825234876506,
+      "loss": 0.5462,
+      "step": 50890
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0003714832291799516,
+      "loss": 0.533,
+      "step": 50900
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00037139820601113805,
+      "loss": 0.526,
+      "step": 50910
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0003713131828423246,
+      "loss": 0.5291,
+      "step": 50920
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00037122815967351104,
+      "loss": 0.5307,
+      "step": 50930
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00037114313650469757,
+      "loss": 0.5015,
+      "step": 50940
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00037105811333588404,
+      "loss": 0.5431,
+      "step": 50950
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00037097309016707056,
+      "loss": 0.4637,
+      "step": 50960
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00037088806699825703,
+      "loss": 0.5173,
+      "step": 50970
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00037080304382944355,
+      "loss": 0.5399,
+      "step": 50980
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00037071802066063,
+      "loss": 0.6044,
+      "step": 50990
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00037063299749181655,
+      "loss": 0.5563,
+      "step": 51000
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00037054797432300307,
+      "loss": 0.538,
+      "step": 51010
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00037046295115418954,
+      "loss": 0.5444,
+      "step": 51020
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.000370377927985376,
+      "loss": 0.5144,
+      "step": 51030
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00037029290481656253,
+      "loss": 0.5121,
+      "step": 51040
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00037020788164774906,
+      "loss": 0.5384,
+      "step": 51050
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0003701228584789355,
+      "loss": 0.6402,
+      "step": 51060
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.000370037835310122,
+      "loss": 0.5078,
+      "step": 51070
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0003699528121413085,
+      "loss": 0.6061,
+      "step": 51080
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00036986778897249504,
+      "loss": 0.431,
+      "step": 51090
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0003697827658036815,
+      "loss": 0.5704,
+      "step": 51100
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.000369697742634868,
+      "loss": 0.4852,
+      "step": 51110
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00036961271946605456,
+      "loss": 0.529,
+      "step": 51120
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00036952769629724103,
+      "loss": 0.5295,
+      "step": 51130
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0003694426731284275,
+      "loss": 0.5303,
+      "step": 51140
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00036935764995961397,
+      "loss": 0.5609,
+      "step": 51150
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00036927262679080055,
+      "loss": 0.5066,
+      "step": 51160
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.000369187603621987,
+      "loss": 0.526,
+      "step": 51170
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0003691025804531735,
+      "loss": 0.4802,
+      "step": 51180
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00036901755728436,
+      "loss": 0.5158,
+      "step": 51190
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00036893253411554653,
+      "loss": 0.5489,
+      "step": 51200
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.000368847510946733,
+      "loss": 0.5542,
+      "step": 51210
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0003687624877779195,
+      "loss": 0.5804,
+      "step": 51220
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.000368677464609106,
+      "loss": 0.57,
+      "step": 51230
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0003685924414402925,
+      "loss": 0.4562,
+      "step": 51240
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.000368507418271479,
+      "loss": 0.5182,
+      "step": 51250
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036842239510266546,
+      "loss": 0.5406,
+      "step": 51260
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.000368337371933852,
+      "loss": 0.5537,
+      "step": 51270
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0003682523487650385,
+      "loss": 0.5224,
+      "step": 51280
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.000368167325596225,
+      "loss": 0.5091,
+      "step": 51290
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0003680823024274115,
+      "loss": 0.4801,
+      "step": 51300
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036799727925859797,
+      "loss": 0.503,
+      "step": 51310
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0003679122560897845,
+      "loss": 0.4799,
+      "step": 51320
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036782723292097096,
+      "loss": 0.5528,
+      "step": 51330
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0003677422097521575,
+      "loss": 0.5688,
+      "step": 51340
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036765718658334396,
+      "loss": 0.5049,
+      "step": 51350
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0003675721634145305,
+      "loss": 0.5491,
+      "step": 51360
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00036748714024571695,
+      "loss": 0.5207,
+      "step": 51370
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0003674021170769035,
+      "loss": 0.6112,
+      "step": 51380
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036731709390809,
+      "loss": 0.5293,
+      "step": 51390
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036723207073927647,
+      "loss": 0.4868,
+      "step": 51400
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.000367147047570463,
+      "loss": 0.5126,
+      "step": 51410
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036706202440164946,
+      "loss": 0.5294,
+      "step": 51420
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.000366977001232836,
+      "loss": 0.5325,
+      "step": 51430
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036689197806402245,
+      "loss": 0.5318,
+      "step": 51440
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.000366806954895209,
+      "loss": 0.5304,
+      "step": 51450
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036672193172639545,
+      "loss": 0.5428,
+      "step": 51460
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036663690855758197,
+      "loss": 0.5077,
+      "step": 51470
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0003665518853887685,
+      "loss": 0.567,
+      "step": 51480
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00036646686221995496,
+      "loss": 0.5031,
+      "step": 51490
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036638183905114143,
+      "loss": 0.5152,
+      "step": 51500
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036629681588232796,
+      "loss": 0.5426,
+      "step": 51510
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0003662117927135145,
+      "loss": 0.5805,
+      "step": 51520
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036612676954470095,
+      "loss": 0.5298,
+      "step": 51530
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0003660417463758874,
+      "loss": 0.5487,
+      "step": 51540
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036595672320707394,
+      "loss": 0.5477,
+      "step": 51550
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036587170003826047,
+      "loss": 0.5387,
+      "step": 51560
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036578667686944694,
+      "loss": 0.5254,
+      "step": 51570
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0003657016537006334,
+      "loss": 0.6019,
+      "step": 51580
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036561663053182,
+      "loss": 0.5348,
+      "step": 51590
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00036553160736300645,
+      "loss": 0.5811,
+      "step": 51600
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0003654465841941929,
+      "loss": 0.4956,
+      "step": 51610
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003653615610253794,
+      "loss": 0.5659,
+      "step": 51620
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00036527653785656597,
+      "loss": 0.5134,
+      "step": 51630
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00036519151468775244,
+      "loss": 0.4982,
+      "step": 51640
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003651064915189389,
+      "loss": 0.4945,
+      "step": 51650
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003650214683501254,
+      "loss": 0.5583,
+      "step": 51660
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00036493644518131196,
+      "loss": 0.4926,
+      "step": 51670
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003648514220124984,
+      "loss": 0.5194,
+      "step": 51680
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003647663988436849,
+      "loss": 0.5568,
+      "step": 51690
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003646813756748714,
+      "loss": 0.4859,
+      "step": 51700
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00036459635250605794,
+      "loss": 0.5114,
+      "step": 51710
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0003645113293372444,
+      "loss": 0.5623,
+      "step": 51720
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003644263061684309,
+      "loss": 0.528,
+      "step": 51730
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003643412829996174,
+      "loss": 0.5455,
+      "step": 51740
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00036425625983080393,
+      "loss": 0.5019,
+      "step": 51750
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003641712366619904,
+      "loss": 0.494,
+      "step": 51760
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00036408621349317687,
+      "loss": 0.585,
+      "step": 51770
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003640011903243634,
+      "loss": 0.5432,
+      "step": 51780
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003639161671555499,
+      "loss": 0.5089,
+      "step": 51790
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003638311439867364,
+      "loss": 0.5533,
+      "step": 51800
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003637461208179229,
+      "loss": 0.5265,
+      "step": 51810
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003636610976491094,
+      "loss": 0.5315,
+      "step": 51820
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0003635760744802959,
+      "loss": 0.498,
+      "step": 51830
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00036349105131148237,
+      "loss": 0.5511,
+      "step": 51840
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003634060281426689,
+      "loss": 0.5552,
+      "step": 51850
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003633210049738554,
+      "loss": 0.5385,
+      "step": 51860
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003632359818050419,
+      "loss": 0.567,
+      "step": 51870
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003631509586362284,
+      "loss": 0.5423,
+      "step": 51880
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003630659354674149,
+      "loss": 0.5293,
+      "step": 51890
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003629809122986014,
+      "loss": 0.5287,
+      "step": 51900
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003628958891297879,
+      "loss": 0.4898,
+      "step": 51910
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003628108659609744,
+      "loss": 0.6007,
+      "step": 51920
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00036272584279216087,
+      "loss": 0.5187,
+      "step": 51930
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003626408196233474,
+      "loss": 0.5074,
+      "step": 51940
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00036255579645453386,
+      "loss": 0.5473,
+      "step": 51950
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0003624707732857204,
+      "loss": 0.4997,
+      "step": 51960
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00036238575011690686,
+      "loss": 0.5459,
+      "step": 51970
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003623007269480934,
+      "loss": 0.5423,
+      "step": 51980
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003622157037792799,
+      "loss": 0.6045,
+      "step": 51990
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003621306806104664,
+      "loss": 0.5263,
+      "step": 52000
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00036204565744165284,
+      "loss": 0.5696,
+      "step": 52010
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00036196063427283937,
+      "loss": 0.6039,
+      "step": 52020
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003618756111040259,
+      "loss": 0.4552,
+      "step": 52030
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00036179058793521236,
+      "loss": 0.5881,
+      "step": 52040
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00036170556476639883,
+      "loss": 0.5197,
+      "step": 52050
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00036162054159758535,
+      "loss": 0.5224,
+      "step": 52060
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0003615355184287719,
+      "loss": 0.4592,
+      "step": 52070
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00036145049525995835,
+      "loss": 0.5202,
+      "step": 52080
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0003613654720911448,
+      "loss": 0.5266,
+      "step": 52090
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0003612804489223314,
+      "loss": 0.5546,
+      "step": 52100
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00036119542575351786,
+      "loss": 0.5248,
+      "step": 52110
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00036111040258470433,
+      "loss": 0.4987,
+      "step": 52120
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00036102537941589086,
+      "loss": 0.5327,
+      "step": 52130
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0003609403562470774,
+      "loss": 0.5562,
+      "step": 52140
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00036085533307826385,
+      "loss": 0.5486,
+      "step": 52150
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0003607703099094503,
+      "loss": 0.4791,
+      "step": 52160
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00036068528674063684,
+      "loss": 0.6114,
+      "step": 52170
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00036060026357182337,
+      "loss": 0.5258,
+      "step": 52180
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00036051524040300984,
+      "loss": 0.5347,
+      "step": 52190
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0003604302172341963,
+      "loss": 0.52,
+      "step": 52200
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00036034519406538283,
+      "loss": 0.4997,
+      "step": 52210
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00036026017089656935,
+      "loss": 0.5252,
+      "step": 52220
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003601751477277558,
+      "loss": 0.5358,
+      "step": 52230
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003600901245589423,
+      "loss": 0.5201,
+      "step": 52240
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003600051013901288,
+      "loss": 0.4966,
+      "step": 52250
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00035992007822131534,
+      "loss": 0.5067,
+      "step": 52260
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003598350550525018,
+      "loss": 0.5807,
+      "step": 52270
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00035975003188368833,
+      "loss": 0.5056,
+      "step": 52280
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003596650087148748,
+      "loss": 0.529,
+      "step": 52290
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003595799855460613,
+      "loss": 0.6229,
+      "step": 52300
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003594949623772478,
+      "loss": 0.5198,
+      "step": 52310
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0003594099392084343,
+      "loss": 0.5583,
+      "step": 52320
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00035932491603962084,
+      "loss": 0.5041,
+      "step": 52330
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003592398928708073,
+      "loss": 0.5034,
+      "step": 52340
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003591548697019938,
+      "loss": 0.4879,
+      "step": 52350
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003590698465331803,
+      "loss": 0.5724,
+      "step": 52360
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00035898482336436683,
+      "loss": 0.5619,
+      "step": 52370
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003588998001955533,
+      "loss": 0.5465,
+      "step": 52380
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003588147770267398,
+      "loss": 0.5457,
+      "step": 52390
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003587297538579263,
+      "loss": 0.5222,
+      "step": 52400
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003586447306891128,
+      "loss": 0.4642,
+      "step": 52410
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003585597075202993,
+      "loss": 0.5212,
+      "step": 52420
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0003584746843514858,
+      "loss": 0.5268,
+      "step": 52430
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003583896611826723,
+      "loss": 0.5516,
+      "step": 52440
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003583046380138588,
+      "loss": 0.5617,
+      "step": 52450
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00035821961484504527,
+      "loss": 0.5369,
+      "step": 52460
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003581345916762318,
+      "loss": 0.5104,
+      "step": 52470
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00035804956850741827,
+      "loss": 0.5822,
+      "step": 52480
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003579645453386048,
+      "loss": 0.5468,
+      "step": 52490
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003578795221697913,
+      "loss": 0.4879,
+      "step": 52500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003577944990009778,
+      "loss": 0.5181,
+      "step": 52510
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00035770947583216425,
+      "loss": 0.5358,
+      "step": 52520
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003576244526633508,
+      "loss": 0.6339,
+      "step": 52530
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0003575394294945373,
+      "loss": 0.5518,
+      "step": 52540
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00035745440632572377,
+      "loss": 0.5346,
+      "step": 52550
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00035736938315691024,
+      "loss": 0.5082,
+      "step": 52560
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003572843599880968,
+      "loss": 0.5657,
+      "step": 52570
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003571993368192833,
+      "loss": 0.5227,
+      "step": 52580
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00035711431365046976,
+      "loss": 0.4873,
+      "step": 52590
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003570292904816563,
+      "loss": 0.532,
+      "step": 52600
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003569442673128428,
+      "loss": 0.5114,
+      "step": 52610
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00035685924414402927,
+      "loss": 0.5215,
+      "step": 52620
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00035677422097521574,
+      "loss": 0.5277,
+      "step": 52630
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00035668919780640227,
+      "loss": 0.5205,
+      "step": 52640
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0003566041746375888,
+      "loss": 0.4925,
+      "step": 52650
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00035651915146877526,
+      "loss": 0.5181,
+      "step": 52660
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00035643412829996173,
+      "loss": 0.5461,
+      "step": 52670
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00035634910513114825,
+      "loss": 0.4751,
+      "step": 52680
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003562640819623348,
+      "loss": 0.6134,
+      "step": 52690
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00035617905879352125,
+      "loss": 0.5348,
+      "step": 52700
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003560940356247077,
+      "loss": 0.5254,
+      "step": 52710
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00035600901245589424,
+      "loss": 0.5177,
+      "step": 52720
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00035592398928708076,
+      "loss": 0.4219,
+      "step": 52730
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00035583896611826723,
+      "loss": 0.5054,
+      "step": 52740
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003557539429494537,
+      "loss": 0.5495,
+      "step": 52750
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003556689197806403,
+      "loss": 0.5965,
+      "step": 52760
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00035558389661182675,
+      "loss": 0.5304,
+      "step": 52770
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0003554988734430132,
+      "loss": 0.5487,
+      "step": 52780
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00035541385027419974,
+      "loss": 0.5275,
+      "step": 52790
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00035532882710538627,
+      "loss": 0.5099,
+      "step": 52800
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00035524380393657274,
+      "loss": 0.5577,
+      "step": 52810
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0003551587807677592,
+      "loss": 0.5208,
+      "step": 52820
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00035507375759894573,
+      "loss": 0.5194,
+      "step": 52830
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00035498873443013225,
+      "loss": 0.5175,
+      "step": 52840
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0003549037112613187,
+      "loss": 0.5414,
+      "step": 52850
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0003548186880925052,
+      "loss": 0.5556,
+      "step": 52860
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0003547336649236917,
+      "loss": 0.5086,
+      "step": 52870
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00035464864175487824,
+      "loss": 0.5022,
+      "step": 52880
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0003545636185860647,
+      "loss": 0.4798,
+      "step": 52890
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00035447859541725123,
+      "loss": 0.5508,
+      "step": 52900
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0003543935722484377,
+      "loss": 0.5184,
+      "step": 52910
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003543085490796242,
+      "loss": 0.5153,
+      "step": 52920
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003542235259108107,
+      "loss": 0.5514,
+      "step": 52930
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003541385027419972,
+      "loss": 0.5536,
+      "step": 52940
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003540534795731837,
+      "loss": 0.5693,
+      "step": 52950
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003539684564043702,
+      "loss": 0.5182,
+      "step": 52960
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00035388343323555674,
+      "loss": 0.5571,
+      "step": 52970
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003537984100667432,
+      "loss": 0.6009,
+      "step": 52980
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003537133868979297,
+      "loss": 0.485,
+      "step": 52990
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003536283637291162,
+      "loss": 0.5534,
+      "step": 53000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003535433405603027,
+      "loss": 0.5904,
+      "step": 53010
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0003534583173914892,
+      "loss": 0.5559,
+      "step": 53020
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00035337329422267566,
+      "loss": 0.4807,
+      "step": 53030
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003532882710538622,
+      "loss": 0.5426,
+      "step": 53040
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003532032478850487,
+      "loss": 0.5515,
+      "step": 53050
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003531182247162352,
+      "loss": 0.4995,
+      "step": 53060
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003530332015474217,
+      "loss": 0.5422,
+      "step": 53070
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003529481783786082,
+      "loss": 0.5094,
+      "step": 53080
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003528631552097947,
+      "loss": 0.6359,
+      "step": 53090
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00035277813204098117,
+      "loss": 0.4792,
+      "step": 53100
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003526931088721677,
+      "loss": 0.5006,
+      "step": 53110
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003526080857033542,
+      "loss": 0.4652,
+      "step": 53120
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0003525230625345407,
+      "loss": 0.5049,
+      "step": 53130
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00035243803936572715,
+      "loss": 0.5296,
+      "step": 53140
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0003523530161969137,
+      "loss": 0.4684,
+      "step": 53150
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0003522679930281002,
+      "loss": 0.5366,
+      "step": 53160
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035218296985928667,
+      "loss": 0.5854,
+      "step": 53170
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035209794669047314,
+      "loss": 0.4877,
+      "step": 53180
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035201292352165966,
+      "loss": 0.5521,
+      "step": 53190
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0003519279003528462,
+      "loss": 0.5228,
+      "step": 53200
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035184287718403266,
+      "loss": 0.4671,
+      "step": 53210
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0003517578540152191,
+      "loss": 0.5192,
+      "step": 53220
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0003516728308464057,
+      "loss": 0.4735,
+      "step": 53230
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035158780767759217,
+      "loss": 0.6152,
+      "step": 53240
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035150278450877864,
+      "loss": 0.5592,
+      "step": 53250
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00035141776133996517,
+      "loss": 0.5672,
+      "step": 53260
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0003513327381711517,
+      "loss": 0.5304,
+      "step": 53270
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00035124771500233816,
+      "loss": 0.5772,
+      "step": 53280
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00035116269183352463,
+      "loss": 0.4803,
+      "step": 53290
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00035107766866471115,
+      "loss": 0.5129,
+      "step": 53300
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0003509926454958977,
+      "loss": 0.5552,
+      "step": 53310
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00035090762232708415,
+      "loss": 0.4866,
+      "step": 53320
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0003508225991582706,
+      "loss": 0.56,
+      "step": 53330
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00035073757598945714,
+      "loss": 0.5519,
+      "step": 53340
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00035065255282064366,
+      "loss": 0.5335,
+      "step": 53350
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00035056752965183013,
+      "loss": 0.4854,
+      "step": 53360
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00035048250648301666,
+      "loss": 0.5457,
+      "step": 53370
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0003503974833142031,
+      "loss": 0.6045,
+      "step": 53380
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00035031246014538965,
+      "loss": 0.4912,
+      "step": 53390
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003502274369765761,
+      "loss": 0.5425,
+      "step": 53400
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00035014241380776264,
+      "loss": 0.5366,
+      "step": 53410
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003500573906389491,
+      "loss": 0.4924,
+      "step": 53420
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00034997236747013564,
+      "loss": 0.5292,
+      "step": 53430
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003498873443013221,
+      "loss": 0.5533,
+      "step": 53440
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00034980232113250863,
+      "loss": 0.5504,
+      "step": 53450
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003497172979636951,
+      "loss": 0.5547,
+      "step": 53460
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003496322747948816,
+      "loss": 0.5531,
+      "step": 53470
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00034954725162606815,
+      "loss": 0.559,
+      "step": 53480
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003494622284572546,
+      "loss": 0.6018,
+      "step": 53490
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0003493772052884411,
+      "loss": 0.534,
+      "step": 53500
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003492921821196276,
+      "loss": 0.5721,
+      "step": 53510
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00034920715895081413,
+      "loss": 0.4901,
+      "step": 53520
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003491221357820006,
+      "loss": 0.5106,
+      "step": 53530
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003490371126131871,
+      "loss": 0.4982,
+      "step": 53540
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003489520894443736,
+      "loss": 0.5618,
+      "step": 53550
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003488670662755601,
+      "loss": 0.513,
+      "step": 53560
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003487820431067466,
+      "loss": 0.5784,
+      "step": 53570
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003486970199379331,
+      "loss": 0.4934,
+      "step": 53580
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00034861199676911964,
+      "loss": 0.5077,
+      "step": 53590
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003485269736003061,
+      "loss": 0.4458,
+      "step": 53600
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003484419504314926,
+      "loss": 0.4881,
+      "step": 53610
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0003483569272626791,
+      "loss": 0.5406,
+      "step": 53620
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003482719040938656,
+      "loss": 0.511,
+      "step": 53630
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003481868809250521,
+      "loss": 0.4923,
+      "step": 53640
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00034810185775623856,
+      "loss": 0.5762,
+      "step": 53650
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003480168345874251,
+      "loss": 0.4975,
+      "step": 53660
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003479318114186116,
+      "loss": 0.5798,
+      "step": 53670
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003478467882497981,
+      "loss": 0.5345,
+      "step": 53680
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00034776176508098455,
+      "loss": 0.4943,
+      "step": 53690
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003476767419121711,
+      "loss": 0.5686,
+      "step": 53700
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003475917187433576,
+      "loss": 0.5681,
+      "step": 53710
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00034750669557454407,
+      "loss": 0.5093,
+      "step": 53720
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00034742167240573053,
+      "loss": 0.5854,
+      "step": 53730
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0003473366492369171,
+      "loss": 0.4829,
+      "step": 53740
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0003472516260681036,
+      "loss": 0.5466,
+      "step": 53750
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034716660289929005,
+      "loss": 0.5679,
+      "step": 53760
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0003470815797304766,
+      "loss": 0.4964,
+      "step": 53770
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0003469965565616631,
+      "loss": 0.5164,
+      "step": 53780
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034691153339284957,
+      "loss": 0.4946,
+      "step": 53790
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034682651022403604,
+      "loss": 0.5566,
+      "step": 53800
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034674148705522256,
+      "loss": 0.6041,
+      "step": 53810
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0003466564638864091,
+      "loss": 0.491,
+      "step": 53820
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034657144071759556,
+      "loss": 0.535,
+      "step": 53830
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.000346486417548782,
+      "loss": 0.51,
+      "step": 53840
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00034640139437996855,
+      "loss": 0.4818,
+      "step": 53850
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034631637121115507,
+      "loss": 0.5437,
+      "step": 53860
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034623134804234154,
+      "loss": 0.5361,
+      "step": 53870
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034614632487352807,
+      "loss": 0.5654,
+      "step": 53880
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034606130170471453,
+      "loss": 0.4582,
+      "step": 53890
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034597627853590106,
+      "loss": 0.508,
+      "step": 53900
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034589125536708753,
+      "loss": 0.4988,
+      "step": 53910
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034580623219827405,
+      "loss": 0.5169,
+      "step": 53920
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0003457212090294605,
+      "loss": 0.5467,
+      "step": 53930
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034563618586064705,
+      "loss": 0.5322,
+      "step": 53940
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034555116269183357,
+      "loss": 0.5424,
+      "step": 53950
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034546613952302004,
+      "loss": 0.5432,
+      "step": 53960
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00034538111635420656,
+      "loss": 0.5763,
+      "step": 53970
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034529609318539303,
+      "loss": 0.58,
+      "step": 53980
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034521107001657956,
+      "loss": 0.5219,
+      "step": 53990
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.000345126046847766,
+      "loss": 0.5916,
+      "step": 54000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034504102367895255,
+      "loss": 0.5452,
+      "step": 54010
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.000344956000510139,
+      "loss": 0.5699,
+      "step": 54020
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034487097734132554,
+      "loss": 0.5658,
+      "step": 54030
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.000344785954172512,
+      "loss": 0.5703,
+      "step": 54040
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034470093100369854,
+      "loss": 0.5849,
+      "step": 54050
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034461590783488506,
+      "loss": 0.5191,
+      "step": 54060
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00034453088466607153,
+      "loss": 0.5594,
+      "step": 54070
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.000344445861497258,
+      "loss": 0.5664,
+      "step": 54080
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0003443608383284445,
+      "loss": 0.5977,
+      "step": 54090
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034427581515963105,
+      "loss": 0.4835,
+      "step": 54100
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0003441907919908175,
+      "loss": 0.5253,
+      "step": 54110
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.000344105768822004,
+      "loss": 0.508,
+      "step": 54120
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0003440207456531905,
+      "loss": 0.5232,
+      "step": 54130
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034393572248437703,
+      "loss": 0.5265,
+      "step": 54140
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0003438506993155635,
+      "loss": 0.5305,
+      "step": 54150
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034376567614674997,
+      "loss": 0.546,
+      "step": 54160
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034368065297793655,
+      "loss": 0.5714,
+      "step": 54170
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.000343595629809123,
+      "loss": 0.5538,
+      "step": 54180
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0003435106066403095,
+      "loss": 0.5851,
+      "step": 54190
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034342558347149596,
+      "loss": 0.5566,
+      "step": 54200
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00034334056030268254,
+      "loss": 0.5263,
+      "step": 54210
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.000343255537133869,
+      "loss": 0.5325,
+      "step": 54220
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0003431705139650555,
+      "loss": 0.501,
+      "step": 54230
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00034308549079624194,
+      "loss": 0.5809,
+      "step": 54240
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0003430004676274285,
+      "loss": 0.4688,
+      "step": 54250
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.000342915444458615,
+      "loss": 0.4949,
+      "step": 54260
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00034283042128980146,
+      "loss": 0.4998,
+      "step": 54270
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.000342745398120988,
+      "loss": 0.5167,
+      "step": 54280
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0003426603749521745,
+      "loss": 0.5064,
+      "step": 54290
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.000342575351783361,
+      "loss": 0.5914,
+      "step": 54300
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00034249032861454745,
+      "loss": 0.5229,
+      "step": 54310
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00034240530544573397,
+      "loss": 0.548,
+      "step": 54320
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0003423202822769205,
+      "loss": 0.4738,
+      "step": 54330
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034223525910810696,
+      "loss": 0.5442,
+      "step": 54340
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0003421502359392935,
+      "loss": 0.5297,
+      "step": 54350
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034206521277047996,
+      "loss": 0.5299,
+      "step": 54360
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0003419801896016665,
+      "loss": 0.5138,
+      "step": 54370
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034189516643285295,
+      "loss": 0.4688,
+      "step": 54380
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0003418101432640395,
+      "loss": 0.5419,
+      "step": 54390
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034172512009522594,
+      "loss": 0.4745,
+      "step": 54400
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034164009692641247,
+      "loss": 0.4292,
+      "step": 54410
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034155507375759894,
+      "loss": 0.5799,
+      "step": 54420
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00034147005058878546,
+      "loss": 0.5379,
+      "step": 54430
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.000341385027419972,
+      "loss": 0.492,
+      "step": 54440
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00034130000425115845,
+      "loss": 0.5417,
+      "step": 54450
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.000341214981082345,
+      "loss": 0.5952,
+      "step": 54460
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00034112995791353145,
+      "loss": 0.4757,
+      "step": 54470
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00034104493474471797,
+      "loss": 0.5145,
+      "step": 54480
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00034095991157590444,
+      "loss": 0.5053,
+      "step": 54490
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00034087488840709097,
+      "loss": 0.5495,
+      "step": 54500
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00034078986523827743,
+      "loss": 0.5282,
+      "step": 54510
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00034070484206946396,
+      "loss": 0.5227,
+      "step": 54520
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00034061981890065043,
+      "loss": 0.5,
+      "step": 54530
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00034053479573183695,
+      "loss": 0.5328,
+      "step": 54540
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0003404497725630234,
+      "loss": 0.497,
+      "step": 54550
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00034036474939420994,
+      "loss": 0.4767,
+      "step": 54560
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00034027972622539647,
+      "loss": 0.5449,
+      "step": 54570
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00034019470305658294,
+      "loss": 0.5527,
+      "step": 54580
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0003401096798877694,
+      "loss": 0.5351,
+      "step": 54590
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00034002465671895593,
+      "loss": 0.5159,
+      "step": 54600
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00033993963355014246,
+      "loss": 0.5485,
+      "step": 54610
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0003398546103813289,
+      "loss": 0.54,
+      "step": 54620
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0003397695872125154,
+      "loss": 0.5656,
+      "step": 54630
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00033968456404370197,
+      "loss": 0.5294,
+      "step": 54640
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00033959954087488844,
+      "loss": 0.4543,
+      "step": 54650
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0003395145177060749,
+      "loss": 0.5703,
+      "step": 54660
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0003394294945372614,
+      "loss": 0.4716,
+      "step": 54670
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00033934447136844796,
+      "loss": 0.5387,
+      "step": 54680
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00033925944819963443,
+      "loss": 0.5284,
+      "step": 54690
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003391744250308209,
+      "loss": 0.5282,
+      "step": 54700
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00033908940186200737,
+      "loss": 0.6201,
+      "step": 54710
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00033900437869319395,
+      "loss": 0.5378,
+      "step": 54720
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003389193555243804,
+      "loss": 0.5604,
+      "step": 54730
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003388343323555669,
+      "loss": 0.5148,
+      "step": 54740
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003387493091867534,
+      "loss": 0.5401,
+      "step": 54750
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00033866428601793993,
+      "loss": 0.5264,
+      "step": 54760
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003385792628491264,
+      "loss": 0.4812,
+      "step": 54770
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00033849423968031287,
+      "loss": 0.5141,
+      "step": 54780
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003384092165114994,
+      "loss": 0.4828,
+      "step": 54790
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0003383241933426859,
+      "loss": 0.5165,
+      "step": 54800
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003382391701738724,
+      "loss": 0.4496,
+      "step": 54810
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00033815414700505886,
+      "loss": 0.5691,
+      "step": 54820
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003380691238362454,
+      "loss": 0.5547,
+      "step": 54830
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003379841006674319,
+      "loss": 0.5776,
+      "step": 54840
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003378990774986184,
+      "loss": 0.5237,
+      "step": 54850
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003378140543298049,
+      "loss": 0.5142,
+      "step": 54860
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00033772903116099137,
+      "loss": 0.5839,
+      "step": 54870
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003376440079921779,
+      "loss": 0.6003,
+      "step": 54880
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00033755898482336436,
+      "loss": 0.5322,
+      "step": 54890
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003374739616545509,
+      "loss": 0.495,
+      "step": 54900
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003373889384857374,
+      "loss": 0.4607,
+      "step": 54910
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0003373039153169239,
+      "loss": 0.5607,
+      "step": 54920
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033721889214811035,
+      "loss": 0.5813,
+      "step": 54930
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033713386897929687,
+      "loss": 0.5338,
+      "step": 54940
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0003370488458104834,
+      "loss": 0.5536,
+      "step": 54950
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033696382264166986,
+      "loss": 0.5658,
+      "step": 54960
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0003368787994728564,
+      "loss": 0.5016,
+      "step": 54970
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033679377630404286,
+      "loss": 0.5205,
+      "step": 54980
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0003367087531352294,
+      "loss": 0.5173,
+      "step": 54990
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033662372996641585,
+      "loss": 0.4935,
+      "step": 55000
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0003365387067976024,
+      "loss": 0.4568,
+      "step": 55010
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033645368362878884,
+      "loss": 0.4718,
+      "step": 55020
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00033636866045997537,
+      "loss": 0.5502,
+      "step": 55030
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0003362836372911619,
+      "loss": 0.5353,
+      "step": 55040
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033619861412234836,
+      "loss": 0.5819,
+      "step": 55050
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033611359095353483,
+      "loss": 0.5979,
+      "step": 55060
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033602856778472135,
+      "loss": 0.5481,
+      "step": 55070
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0003359435446159079,
+      "loss": 0.4943,
+      "step": 55080
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033585852144709435,
+      "loss": 0.4748,
+      "step": 55090
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0003357734982782808,
+      "loss": 0.5229,
+      "step": 55100
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033568847510946734,
+      "loss": 0.5143,
+      "step": 55110
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033560345194065386,
+      "loss": 0.5525,
+      "step": 55120
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00033551842877184033,
+      "loss": 0.5466,
+      "step": 55130
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0003354334056030268,
+      "loss": 0.6043,
+      "step": 55140
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0003353483824342134,
+      "loss": 0.5832,
+      "step": 55150
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00033526335926539985,
+      "loss": 0.5019,
+      "step": 55160
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003351783360965863,
+      "loss": 0.4971,
+      "step": 55170
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00033509331292777284,
+      "loss": 0.4779,
+      "step": 55180
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00033500828975895937,
+      "loss": 0.5577,
+      "step": 55190
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00033492326659014584,
+      "loss": 0.5562,
+      "step": 55200
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003348382434213323,
+      "loss": 0.5149,
+      "step": 55210
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00033475322025251883,
+      "loss": 0.612,
+      "step": 55220
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00033466819708370535,
+      "loss": 0.5185,
+      "step": 55230
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003345831739148918,
+      "loss": 0.5557,
+      "step": 55240
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003344981507460783,
+      "loss": 0.4686,
+      "step": 55250
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0003344131275772648,
+      "loss": 0.5315,
+      "step": 55260
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00033432810440845134,
+      "loss": 0.4944,
+      "step": 55270
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003342430812396378,
+      "loss": 0.5356,
+      "step": 55280
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003341580580708243,
+      "loss": 0.4866,
+      "step": 55290
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003340730349020108,
+      "loss": 0.4988,
+      "step": 55300
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00033398801173319733,
+      "loss": 0.4776,
+      "step": 55310
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003339029885643838,
+      "loss": 0.5058,
+      "step": 55320
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00033381796539557027,
+      "loss": 0.4927,
+      "step": 55330
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003337329422267568,
+      "loss": 0.5238,
+      "step": 55340
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003336479190579433,
+      "loss": 0.5038,
+      "step": 55350
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003335628958891298,
+      "loss": 0.5061,
+      "step": 55360
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003334778727203163,
+      "loss": 0.5731,
+      "step": 55370
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00033339284955150283,
+      "loss": 0.5511,
+      "step": 55380
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0003333078263826893,
+      "loss": 0.5388,
+      "step": 55390
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00033322280321387577,
+      "loss": 0.5455,
+      "step": 55400
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003331377800450623,
+      "loss": 0.4859,
+      "step": 55410
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003330527568762488,
+      "loss": 0.486,
+      "step": 55420
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003329677337074353,
+      "loss": 0.4877,
+      "step": 55430
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003328827105386218,
+      "loss": 0.5588,
+      "step": 55440
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003327976873698083,
+      "loss": 0.4944,
+      "step": 55450
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003327126642009948,
+      "loss": 0.4935,
+      "step": 55460
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003326276410321813,
+      "loss": 0.5299,
+      "step": 55470
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003325426178633678,
+      "loss": 0.437,
+      "step": 55480
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00033245759469455427,
+      "loss": 0.5257,
+      "step": 55490
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0003323725715257408,
+      "loss": 0.53,
+      "step": 55500
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00033228754835692726,
+      "loss": 0.4825,
+      "step": 55510
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003322025251881138,
+      "loss": 0.5723,
+      "step": 55520
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00033211750201930025,
+      "loss": 0.5301,
+      "step": 55530
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003320324788504868,
+      "loss": 0.539,
+      "step": 55540
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003319474556816733,
+      "loss": 0.4349,
+      "step": 55550
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00033186243251285977,
+      "loss": 0.5032,
+      "step": 55560
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00033177740934404624,
+      "loss": 0.5146,
+      "step": 55570
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00033169238617523276,
+      "loss": 0.5263,
+      "step": 55580
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003316073630064193,
+      "loss": 0.5339,
+      "step": 55590
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00033152233983760576,
+      "loss": 0.5059,
+      "step": 55600
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003314373166687922,
+      "loss": 0.5353,
+      "step": 55610
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00033135229349997875,
+      "loss": 0.5415,
+      "step": 55620
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0003312672703311653,
+      "loss": 0.5245,
+      "step": 55630
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00033118224716235174,
+      "loss": 0.5084,
+      "step": 55640
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00033109722399353827,
+      "loss": 0.5699,
+      "step": 55650
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0003310122008247248,
+      "loss": 0.517,
+      "step": 55660
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00033092717765591126,
+      "loss": 0.47,
+      "step": 55670
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00033084215448709773,
+      "loss": 0.5049,
+      "step": 55680
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00033075713131828425,
+      "loss": 0.546,
+      "step": 55690
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0003306721081494708,
+      "loss": 0.5614,
+      "step": 55700
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00033058708498065725,
+      "loss": 0.499,
+      "step": 55710
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0003305020618118437,
+      "loss": 0.5601,
+      "step": 55720
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00033041703864303024,
+      "loss": 0.4638,
+      "step": 55730
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00033033201547421676,
+      "loss": 0.4581,
+      "step": 55740
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00033024699230540323,
+      "loss": 0.4873,
+      "step": 55750
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0003301619691365897,
+      "loss": 0.5886,
+      "step": 55760
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00033007694596777623,
+      "loss": 0.5316,
+      "step": 55770
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00032999192279896275,
+      "loss": 0.5498,
+      "step": 55780
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0003299068996301492,
+      "loss": 0.5555,
+      "step": 55790
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0003298218764613357,
+      "loss": 0.519,
+      "step": 55800
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0003297368532925222,
+      "loss": 0.4784,
+      "step": 55810
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00032965183012370874,
+      "loss": 0.5657,
+      "step": 55820
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0003295668069548952,
+      "loss": 0.5009,
+      "step": 55830
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00032948178378608173,
+      "loss": 0.5344,
+      "step": 55840
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00032939676061726825,
+      "loss": 0.6168,
+      "step": 55850
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0003293117374484547,
+      "loss": 0.5541,
+      "step": 55860
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003292267142796412,
+      "loss": 0.513,
+      "step": 55870
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003291416911108277,
+      "loss": 0.5439,
+      "step": 55880
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00032905666794201424,
+      "loss": 0.4807,
+      "step": 55890
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003289716447732007,
+      "loss": 0.5351,
+      "step": 55900
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003288866216043872,
+      "loss": 0.5837,
+      "step": 55910
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003288015984355737,
+      "loss": 0.5553,
+      "step": 55920
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00032871657526676023,
+      "loss": 0.4608,
+      "step": 55930
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003286315520979467,
+      "loss": 0.4627,
+      "step": 55940
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003285465289291332,
+      "loss": 0.5128,
+      "step": 55950
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003284615057603197,
+      "loss": 0.5455,
+      "step": 55960
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003283764825915062,
+      "loss": 0.5312,
+      "step": 55970
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0003282914594226927,
+      "loss": 0.504,
+      "step": 55980
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003282064362538792,
+      "loss": 0.5474,
+      "step": 55990
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003281214130850657,
+      "loss": 0.4877,
+      "step": 56000
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003280363899162522,
+      "loss": 0.564,
+      "step": 56010
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00032795136674743867,
+      "loss": 0.5492,
+      "step": 56020
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003278663435786252,
+      "loss": 0.4991,
+      "step": 56030
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00032778132040981166,
+      "loss": 0.4654,
+      "step": 56040
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003276962972409982,
+      "loss": 0.5261,
+      "step": 56050
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003276112740721847,
+      "loss": 0.4939,
+      "step": 56060
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003275262509033712,
+      "loss": 0.5546,
+      "step": 56070
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00032744122773455765,
+      "loss": 0.5319,
+      "step": 56080
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003273562045657442,
+      "loss": 0.5442,
+      "step": 56090
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0003272711813969307,
+      "loss": 0.4998,
+      "step": 56100
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00032718615822811717,
+      "loss": 0.5607,
+      "step": 56110
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003271011350593037,
+      "loss": 0.5119,
+      "step": 56120
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003270161118904902,
+      "loss": 0.5624,
+      "step": 56130
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003269310887216767,
+      "loss": 0.5328,
+      "step": 56140
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00032684606555286315,
+      "loss": 0.5439,
+      "step": 56150
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003267610423840497,
+      "loss": 0.5379,
+      "step": 56160
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003266760192152362,
+      "loss": 0.5345,
+      "step": 56170
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00032659099604642267,
+      "loss": 0.5538,
+      "step": 56180
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00032650597287760914,
+      "loss": 0.5129,
+      "step": 56190
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00032642094970879566,
+      "loss": 0.4918,
+      "step": 56200
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0003263359265399822,
+      "loss": 0.5459,
+      "step": 56210
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00032625090337116866,
+      "loss": 0.5154,
+      "step": 56220
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003261658802023551,
+      "loss": 0.5636,
+      "step": 56230
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00032608085703354165,
+      "loss": 0.5074,
+      "step": 56240
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003259958338647282,
+      "loss": 0.4804,
+      "step": 56250
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00032591081069591464,
+      "loss": 0.4885,
+      "step": 56260
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003258257875271011,
+      "loss": 0.5379,
+      "step": 56270
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003257407643582877,
+      "loss": 0.5407,
+      "step": 56280
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00032565574118947416,
+      "loss": 0.5217,
+      "step": 56290
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00032557071802066063,
+      "loss": 0.5173,
+      "step": 56300
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003254856948518471,
+      "loss": 0.5045,
+      "step": 56310
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003254006716830337,
+      "loss": 0.4782,
+      "step": 56320
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00032531564851422015,
+      "loss": 0.505,
+      "step": 56330
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0003252306253454066,
+      "loss": 0.5488,
+      "step": 56340
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00032514560217659314,
+      "loss": 0.4729,
+      "step": 56350
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00032506057900777966,
+      "loss": 0.5606,
+      "step": 56360
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00032497555583896613,
+      "loss": 0.5256,
+      "step": 56370
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003248905326701526,
+      "loss": 0.4986,
+      "step": 56380
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003248055095013391,
+      "loss": 0.5004,
+      "step": 56390
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00032472048633252565,
+      "loss": 0.564,
+      "step": 56400
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003246354631637121,
+      "loss": 0.4997,
+      "step": 56410
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00032455043999489864,
+      "loss": 0.5142,
+      "step": 56420
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003244654168260851,
+      "loss": 0.4842,
+      "step": 56430
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00032438039365727164,
+      "loss": 0.527,
+      "step": 56440
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0003242953704884581,
+      "loss": 0.481,
+      "step": 56450
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00032421034731964463,
+      "loss": 0.4978,
+      "step": 56460
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003241253241508311,
+      "loss": 0.4916,
+      "step": 56470
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003240403009820176,
+      "loss": 0.5216,
+      "step": 56480
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003239552778132041,
+      "loss": 0.5445,
+      "step": 56490
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003238702546443906,
+      "loss": 0.5798,
+      "step": 56500
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003237852314755771,
+      "loss": 0.5347,
+      "step": 56510
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003237002083067636,
+      "loss": 0.5595,
+      "step": 56520
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00032361518513795013,
+      "loss": 0.5822,
+      "step": 56530
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003235301619691366,
+      "loss": 0.5599,
+      "step": 56540
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003234451388003231,
+      "loss": 0.5203,
+      "step": 56550
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003233601156315096,
+      "loss": 0.5167,
+      "step": 56560
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0003232750924626961,
+      "loss": 0.4825,
+      "step": 56570
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003231900692938826,
+      "loss": 0.5815,
+      "step": 56580
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003231050461250691,
+      "loss": 0.5645,
+      "step": 56590
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003230200229562556,
+      "loss": 0.5287,
+      "step": 56600
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003229349997874421,
+      "loss": 0.5626,
+      "step": 56610
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003228499766186286,
+      "loss": 0.5605,
+      "step": 56620
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003227649534498151,
+      "loss": 0.518,
+      "step": 56630
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003226799302810016,
+      "loss": 0.4921,
+      "step": 56640
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003225949071121881,
+      "loss": 0.5012,
+      "step": 56650
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00032250988394337456,
+      "loss": 0.4777,
+      "step": 56660
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003224248607745611,
+      "loss": 0.5087,
+      "step": 56670
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003223398376057476,
+      "loss": 0.5589,
+      "step": 56680
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0003222548144369341,
+      "loss": 0.5416,
+      "step": 56690
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032216979126812055,
+      "loss": 0.5224,
+      "step": 56700
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0003220847680993071,
+      "loss": 0.5395,
+      "step": 56710
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0003219997449304936,
+      "loss": 0.5008,
+      "step": 56720
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032191472176168007,
+      "loss": 0.5618,
+      "step": 56730
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032182969859286654,
+      "loss": 0.5602,
+      "step": 56740
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0003217446754240531,
+      "loss": 0.5726,
+      "step": 56750
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0003216596522552396,
+      "loss": 0.4994,
+      "step": 56760
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032157462908642605,
+      "loss": 0.5755,
+      "step": 56770
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0003214896059176125,
+      "loss": 0.5418,
+      "step": 56780
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0003214045827487991,
+      "loss": 0.5361,
+      "step": 56790
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032131955957998557,
+      "loss": 0.4973,
+      "step": 56800
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00032123453641117204,
+      "loss": 0.5262,
+      "step": 56810
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00032114951324235856,
+      "loss": 0.5245,
+      "step": 56820
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0003210644900735451,
+      "loss": 0.5423,
+      "step": 56830
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00032097946690473156,
+      "loss": 0.4909,
+      "step": 56840
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.000320894443735918,
+      "loss": 0.5326,
+      "step": 56850
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00032080942056710455,
+      "loss": 0.5233,
+      "step": 56860
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0003207243973982911,
+      "loss": 0.5561,
+      "step": 56870
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00032063937422947754,
+      "loss": 0.5023,
+      "step": 56880
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.000320554351060664,
+      "loss": 0.5299,
+      "step": 56890
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00032046932789185054,
+      "loss": 0.4875,
+      "step": 56900
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00032038430472303706,
+      "loss": 0.5578,
+      "step": 56910
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00032029928155422353,
+      "loss": 0.5205,
+      "step": 56920
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00032021425838541005,
+      "loss": 0.5586,
+      "step": 56930
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0003201292352165965,
+      "loss": 0.494,
+      "step": 56940
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00032004421204778305,
+      "loss": 0.52,
+      "step": 56950
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0003199591888789695,
+      "loss": 0.6038,
+      "step": 56960
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00031987416571015604,
+      "loss": 0.4576,
+      "step": 56970
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0003197891425413425,
+      "loss": 0.4616,
+      "step": 56980
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00031970411937252903,
+      "loss": 0.5574,
+      "step": 56990
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0003196190962037155,
+      "loss": 0.5788,
+      "step": 57000
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.000319534073034902,
+      "loss": 0.5055,
+      "step": 57010
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00031944904986608855,
+      "loss": 0.5901,
+      "step": 57020
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.000319364026697275,
+      "loss": 0.538,
+      "step": 57030
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00031927900352846154,
+      "loss": 0.5231,
+      "step": 57040
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.000319193980359648,
+      "loss": 0.5095,
+      "step": 57050
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031910895719083454,
+      "loss": 0.5366,
+      "step": 57060
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.000319023934022021,
+      "loss": 0.4992,
+      "step": 57070
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031893891085320753,
+      "loss": 0.4853,
+      "step": 57080
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.000318853887684394,
+      "loss": 0.5549,
+      "step": 57090
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0003187688645155805,
+      "loss": 0.5611,
+      "step": 57100
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031868384134676705,
+      "loss": 0.4936,
+      "step": 57110
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0003185988181779535,
+      "loss": 0.5338,
+      "step": 57120
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031851379500914,
+      "loss": 0.5076,
+      "step": 57130
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0003184287718403265,
+      "loss": 0.5585,
+      "step": 57140
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00031834374867151303,
+      "loss": 0.5806,
+      "step": 57150
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0003182587255026995,
+      "loss": 0.5666,
+      "step": 57160
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00031817370233388597,
+      "loss": 0.5151,
+      "step": 57170
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003180886791650725,
+      "loss": 0.5917,
+      "step": 57180
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.000318003655996259,
+      "loss": 0.5159,
+      "step": 57190
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003179186328274455,
+      "loss": 0.5347,
+      "step": 57200
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00031783360965863196,
+      "loss": 0.4872,
+      "step": 57210
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00031774858648981854,
+      "loss": 0.5065,
+      "step": 57220
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.000317663563321005,
+      "loss": 0.5445,
+      "step": 57230
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003175785401521915,
+      "loss": 0.5745,
+      "step": 57240
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00031749351698337795,
+      "loss": 0.5559,
+      "step": 57250
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0003174084938145645,
+      "loss": 0.5934,
+      "step": 57260
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.000317323470645751,
+      "loss": 0.5644,
+      "step": 57270
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00031723844747693746,
+      "loss": 0.4677,
+      "step": 57280
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00031715342430812393,
+      "loss": 0.5107,
+      "step": 57290
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0003170684011393105,
+      "loss": 0.4862,
+      "step": 57300
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.000316983377970497,
+      "loss": 0.4984,
+      "step": 57310
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00031689835480168345,
+      "loss": 0.4956,
+      "step": 57320
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00031681333163287,
+      "loss": 0.4603,
+      "step": 57330
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0003167283084640565,
+      "loss": 0.5892,
+      "step": 57340
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00031664328529524297,
+      "loss": 0.5473,
+      "step": 57350
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00031655826212642944,
+      "loss": 0.5571,
+      "step": 57360
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00031647323895761596,
+      "loss": 0.5363,
+      "step": 57370
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0003163882157888025,
+      "loss": 0.5824,
+      "step": 57380
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00031630319261998895,
+      "loss": 0.5915,
+      "step": 57390
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0003162181694511754,
+      "loss": 0.4946,
+      "step": 57400
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031613314628236195,
+      "loss": 0.5466,
+      "step": 57410
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031604812311354847,
+      "loss": 0.4947,
+      "step": 57420
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031596309994473494,
+      "loss": 0.5414,
+      "step": 57430
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031587807677592146,
+      "loss": 0.4803,
+      "step": 57440
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031579305360710793,
+      "loss": 0.5389,
+      "step": 57450
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031570803043829446,
+      "loss": 0.5793,
+      "step": 57460
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003156230072694809,
+      "loss": 0.5164,
+      "step": 57470
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031553798410066745,
+      "loss": 0.504,
+      "step": 57480
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.000315452960931854,
+      "loss": 0.5561,
+      "step": 57490
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031536793776304044,
+      "loss": 0.5412,
+      "step": 57500
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031528291459422697,
+      "loss": 0.4909,
+      "step": 57510
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00031519789142541344,
+      "loss": 0.5126,
+      "step": 57520
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00031511286825659996,
+      "loss": 0.4895,
+      "step": 57530
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00031502784508778643,
+      "loss": 0.5032,
+      "step": 57540
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00031494282191897295,
+      "loss": 0.5097,
+      "step": 57550
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0003148577987501594,
+      "loss": 0.54,
+      "step": 57560
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00031477277558134595,
+      "loss": 0.4998,
+      "step": 57570
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0003146877524125324,
+      "loss": 0.5518,
+      "step": 57580
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00031460272924371894,
+      "loss": 0.4833,
+      "step": 57590
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0003145177060749054,
+      "loss": 0.585,
+      "step": 57600
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00031443268290609193,
+      "loss": 0.5355,
+      "step": 57610
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00031434765973727846,
+      "loss": 0.4916,
+      "step": 57620
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0003142626365684649,
+      "loss": 0.5079,
+      "step": 57630
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0003141776133996514,
+      "loss": 0.5728,
+      "step": 57640
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003140925902308379,
+      "loss": 0.4919,
+      "step": 57650
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00031400756706202444,
+      "loss": 0.4768,
+      "step": 57660
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003139225438932109,
+      "loss": 0.4937,
+      "step": 57670
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003138375207243974,
+      "loss": 0.5384,
+      "step": 57680
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003137524975555839,
+      "loss": 0.5437,
+      "step": 57690
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00031366747438677043,
+      "loss": 0.5409,
+      "step": 57700
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003135824512179569,
+      "loss": 0.5324,
+      "step": 57710
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00031349742804914337,
+      "loss": 0.4764,
+      "step": 57720
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00031341240488032995,
+      "loss": 0.5457,
+      "step": 57730
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003133273817115164,
+      "loss": 0.5096,
+      "step": 57740
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0003132423585427029,
+      "loss": 0.4987,
+      "step": 57750
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00031315733537388936,
+      "loss": 0.5146,
+      "step": 57760
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00031307231220507593,
+      "loss": 0.5474,
+      "step": 57770
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003129872890362624,
+      "loss": 0.5027,
+      "step": 57780
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00031290226586744887,
+      "loss": 0.5439,
+      "step": 57790
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003128172426986354,
+      "loss": 0.5382,
+      "step": 57800
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003127322195298219,
+      "loss": 0.5684,
+      "step": 57810
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003126471963610084,
+      "loss": 0.4948,
+      "step": 57820
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00031256217319219486,
+      "loss": 0.5002,
+      "step": 57830
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003124771500233814,
+      "loss": 0.6046,
+      "step": 57840
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003123921268545679,
+      "loss": 0.4387,
+      "step": 57850
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0003123071036857544,
+      "loss": 0.5268,
+      "step": 57860
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00031222208051694085,
+      "loss": 0.5522,
+      "step": 57870
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00031213705734812737,
+      "loss": 0.5108,
+      "step": 57880
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003120520341793139,
+      "loss": 0.4794,
+      "step": 57890
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00031196701101050036,
+      "loss": 0.5477,
+      "step": 57900
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003118819878416869,
+      "loss": 0.5485,
+      "step": 57910
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00031179696467287336,
+      "loss": 0.5321,
+      "step": 57920
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003117119415040599,
+      "loss": 0.5217,
+      "step": 57930
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00031162691833524635,
+      "loss": 0.5237,
+      "step": 57940
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00031154189516643287,
+      "loss": 0.509,
+      "step": 57950
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003114568719976194,
+      "loss": 0.4937,
+      "step": 57960
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00031137184882880587,
+      "loss": 0.5392,
+      "step": 57970
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00031128682565999234,
+      "loss": 0.4879,
+      "step": 57980
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00031120180249117886,
+      "loss": 0.5305,
+      "step": 57990
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0003111167793223654,
+      "loss": 0.5403,
+      "step": 58000
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00031103175615355185,
+      "loss": 0.5572,
+      "step": 58010
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0003109467329847384,
+      "loss": 0.5686,
+      "step": 58020
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00031086170981592485,
+      "loss": 0.4544,
+      "step": 58030
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00031077668664711137,
+      "loss": 0.5139,
+      "step": 58040
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00031069166347829784,
+      "loss": 0.5262,
+      "step": 58050
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00031060664030948436,
+      "loss": 0.5269,
+      "step": 58060
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00031052161714067083,
+      "loss": 0.482,
+      "step": 58070
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00031043659397185736,
+      "loss": 0.539,
+      "step": 58080
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0003103515708030438,
+      "loss": 0.5664,
+      "step": 58090
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00031026654763423035,
+      "loss": 0.577,
+      "step": 58100
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0003101815244654168,
+      "loss": 0.5434,
+      "step": 58110
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00031009650129660334,
+      "loss": 0.5675,
+      "step": 58120
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00031001147812778987,
+      "loss": 0.597,
+      "step": 58130
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00030992645495897634,
+      "loss": 0.5312,
+      "step": 58140
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0003098414317901628,
+      "loss": 0.5379,
+      "step": 58150
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00030975640862134933,
+      "loss": 0.5944,
+      "step": 58160
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00030967138545253585,
+      "loss": 0.5055,
+      "step": 58170
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0003095863622837223,
+      "loss": 0.505,
+      "step": 58180
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0003095013391149088,
+      "loss": 0.5231,
+      "step": 58190
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00030941631594609537,
+      "loss": 0.5308,
+      "step": 58200
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00030933129277728184,
+      "loss": 0.5144,
+      "step": 58210
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0003092462696084683,
+      "loss": 0.5298,
+      "step": 58220
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00030916124643965483,
+      "loss": 0.5496,
+      "step": 58230
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00030907622327084136,
+      "loss": 0.5023,
+      "step": 58240
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003089912001020278,
+      "loss": 0.5705,
+      "step": 58250
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003089061769332143,
+      "loss": 0.4355,
+      "step": 58260
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003088211537644008,
+      "loss": 0.5096,
+      "step": 58270
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00030873613059558734,
+      "loss": 0.5527,
+      "step": 58280
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003086511074267738,
+      "loss": 0.5116,
+      "step": 58290
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003085660842579603,
+      "loss": 0.5181,
+      "step": 58300
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003084810610891468,
+      "loss": 0.5601,
+      "step": 58310
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00030839603792033333,
+      "loss": 0.5675,
+      "step": 58320
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003083110147515198,
+      "loss": 0.4918,
+      "step": 58330
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00030822599158270627,
+      "loss": 0.4852,
+      "step": 58340
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0003081409684138928,
+      "loss": 0.4957,
+      "step": 58350
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003080559452450793,
+      "loss": 0.5137,
+      "step": 58360
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003079709220762658,
+      "loss": 0.5427,
+      "step": 58370
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00030788589890745226,
+      "loss": 0.5604,
+      "step": 58380
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003078008757386388,
+      "loss": 0.5209,
+      "step": 58390
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003077158525698253,
+      "loss": 0.5046,
+      "step": 58400
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00030763082940101177,
+      "loss": 0.4942,
+      "step": 58410
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003075458062321983,
+      "loss": 0.507,
+      "step": 58420
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003074607830633848,
+      "loss": 0.4856,
+      "step": 58430
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003073757598945713,
+      "loss": 0.5666,
+      "step": 58440
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00030729073672575776,
+      "loss": 0.4983,
+      "step": 58450
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0003072057135569443,
+      "loss": 0.5355,
+      "step": 58460
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003071206903881308,
+      "loss": 0.4778,
+      "step": 58470
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003070356672193173,
+      "loss": 0.5177,
+      "step": 58480
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00030695064405050375,
+      "loss": 0.533,
+      "step": 58490
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00030686562088169027,
+      "loss": 0.5327,
+      "step": 58500
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003067805977128768,
+      "loss": 0.5309,
+      "step": 58510
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00030669557454406326,
+      "loss": 0.5089,
+      "step": 58520
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003066105513752498,
+      "loss": 0.5152,
+      "step": 58530
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00030652552820643626,
+      "loss": 0.5713,
+      "step": 58540
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003064405050376228,
+      "loss": 0.5523,
+      "step": 58550
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00030635548186880925,
+      "loss": 0.5696,
+      "step": 58560
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00030627045869999577,
+      "loss": 0.4973,
+      "step": 58570
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00030618543553118224,
+      "loss": 0.5192,
+      "step": 58580
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030610041236236877,
+      "loss": 0.5372,
+      "step": 58590
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0003060153891935553,
+      "loss": 0.5961,
+      "step": 58600
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030593036602474176,
+      "loss": 0.595,
+      "step": 58610
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030584534285592823,
+      "loss": 0.5328,
+      "step": 58620
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030576031968711475,
+      "loss": 0.483,
+      "step": 58630
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0003056752965183013,
+      "loss": 0.5064,
+      "step": 58640
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030559027334948775,
+      "loss": 0.4972,
+      "step": 58650
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0003055052501806742,
+      "loss": 0.4494,
+      "step": 58660
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030542022701186074,
+      "loss": 0.4723,
+      "step": 58670
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030533520384304726,
+      "loss": 0.537,
+      "step": 58680
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030525018067423373,
+      "loss": 0.5296,
+      "step": 58690
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00030516515750542026,
+      "loss": 0.6283,
+      "step": 58700
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003050801343366068,
+      "loss": 0.4928,
+      "step": 58710
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00030499511116779325,
+      "loss": 0.5408,
+      "step": 58720
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003049100879989797,
+      "loss": 0.5639,
+      "step": 58730
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00030482506483016624,
+      "loss": 0.5415,
+      "step": 58740
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00030474004166135277,
+      "loss": 0.4787,
+      "step": 58750
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00030465501849253924,
+      "loss": 0.5484,
+      "step": 58760
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003045699953237257,
+      "loss": 0.5325,
+      "step": 58770
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00030448497215491223,
+      "loss": 0.5291,
+      "step": 58780
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00030439994898609875,
+      "loss": 0.4968,
+      "step": 58790
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003043149258172852,
+      "loss": 0.5538,
+      "step": 58800
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003042299026484717,
+      "loss": 0.4972,
+      "step": 58810
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003041448794796582,
+      "loss": 0.5821,
+      "step": 58820
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00030405985631084474,
+      "loss": 0.5511,
+      "step": 58830
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003039748331420312,
+      "loss": 0.4831,
+      "step": 58840
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003038898099732177,
+      "loss": 0.4967,
+      "step": 58850
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003038047868044042,
+      "loss": 0.6207,
+      "step": 58860
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003037197636355907,
+      "loss": 0.5342,
+      "step": 58870
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003036347404667772,
+      "loss": 0.5155,
+      "step": 58880
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003035497172979637,
+      "loss": 0.543,
+      "step": 58890
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00030346469412915024,
+      "loss": 0.5411,
+      "step": 58900
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003033796709603367,
+      "loss": 0.5228,
+      "step": 58910
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003032946477915232,
+      "loss": 0.5722,
+      "step": 58920
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0003032096246227097,
+      "loss": 0.5117,
+      "step": 58930
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00030312460145389623,
+      "loss": 0.4745,
+      "step": 58940
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003030395782850827,
+      "loss": 0.5012,
+      "step": 58950
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00030295455511626917,
+      "loss": 0.5253,
+      "step": 58960
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003028695319474557,
+      "loss": 0.5496,
+      "step": 58970
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003027845087786422,
+      "loss": 0.5283,
+      "step": 58980
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003026994856098287,
+      "loss": 0.4905,
+      "step": 58990
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003026144624410152,
+      "loss": 0.4854,
+      "step": 59000
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003025294392722017,
+      "loss": 0.4524,
+      "step": 59010
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003024444161033882,
+      "loss": 0.56,
+      "step": 59020
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00030235939293457467,
+      "loss": 0.5173,
+      "step": 59030
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003022743697657612,
+      "loss": 0.5534,
+      "step": 59040
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00030218934659694767,
+      "loss": 0.5556,
+      "step": 59050
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0003021043234281342,
+      "loss": 0.4632,
+      "step": 59060
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00030201930025932066,
+      "loss": 0.5404,
+      "step": 59070
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003019342770905072,
+      "loss": 0.5028,
+      "step": 59080
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00030184925392169365,
+      "loss": 0.5763,
+      "step": 59090
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003017642307528802,
+      "loss": 0.524,
+      "step": 59100
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003016792075840667,
+      "loss": 0.5775,
+      "step": 59110
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00030159418441525317,
+      "loss": 0.5299,
+      "step": 59120
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00030150916124643964,
+      "loss": 0.5217,
+      "step": 59130
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00030142413807762616,
+      "loss": 0.4312,
+      "step": 59140
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003013391149088127,
+      "loss": 0.4941,
+      "step": 59150
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00030125409173999916,
+      "loss": 0.5204,
+      "step": 59160
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0003011690685711857,
+      "loss": 0.4746,
+      "step": 59170
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00030108404540237215,
+      "loss": 0.5039,
+      "step": 59180
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00030099902223355867,
+      "loss": 0.5247,
+      "step": 59190
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00030091399906474514,
+      "loss": 0.4868,
+      "step": 59200
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00030082897589593167,
+      "loss": 0.5113,
+      "step": 59210
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0003007439527271182,
+      "loss": 0.4589,
+      "step": 59220
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00030065892955830466,
+      "loss": 0.484,
+      "step": 59230
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00030057390638949113,
+      "loss": 0.4786,
+      "step": 59240
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00030048888322067765,
+      "loss": 0.4822,
+      "step": 59250
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0003004038600518642,
+      "loss": 0.4964,
+      "step": 59260
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00030031883688305065,
+      "loss": 0.4496,
+      "step": 59270
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0003002338137142371,
+      "loss": 0.4667,
+      "step": 59280
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00030014879054542364,
+      "loss": 0.528,
+      "step": 59290
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00030006376737661016,
+      "loss": 0.4652,
+      "step": 59300
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00029997874420779663,
+      "loss": 0.5118,
+      "step": 59310
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002998937210389831,
+      "loss": 0.4831,
+      "step": 59320
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002998086978701697,
+      "loss": 0.4578,
+      "step": 59330
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00029972367470135615,
+      "loss": 0.5165,
+      "step": 59340
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002996386515325426,
+      "loss": 0.4585,
+      "step": 59350
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002995536283637291,
+      "loss": 0.4564,
+      "step": 59360
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00029946860519491567,
+      "loss": 0.457,
+      "step": 59370
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00029938358202610214,
+      "loss": 0.4563,
+      "step": 59380
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002992985588572886,
+      "loss": 0.4958,
+      "step": 59390
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00029921353568847513,
+      "loss": 0.4798,
+      "step": 59400
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00029912851251966165,
+      "loss": 0.5039,
+      "step": 59410
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002990434893508481,
+      "loss": 0.5891,
+      "step": 59420
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002989584661820346,
+      "loss": 0.5387,
+      "step": 59430
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002988734430132211,
+      "loss": 0.4958,
+      "step": 59440
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00029878841984440764,
+      "loss": 0.506,
+      "step": 59450
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002987033966755941,
+      "loss": 0.4646,
+      "step": 59460
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002986183735067806,
+      "loss": 0.5048,
+      "step": 59470
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002985333503379671,
+      "loss": 0.5092,
+      "step": 59480
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002984483271691536,
+      "loss": 0.4912,
+      "step": 59490
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002983633040003401,
+      "loss": 0.4462,
+      "step": 59500
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002982782808315266,
+      "loss": 0.5097,
+      "step": 59510
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002981932576627131,
+      "loss": 0.5083,
+      "step": 59520
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002981082344938996,
+      "loss": 0.483,
+      "step": 59530
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002980232113250861,
+      "loss": 0.3952,
+      "step": 59540
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002979381881562726,
+      "loss": 0.494,
+      "step": 59550
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002978531649874591,
+      "loss": 0.5519,
+      "step": 59560
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002977681418186456,
+      "loss": 0.5297,
+      "step": 59570
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002976831186498321,
+      "loss": 0.5135,
+      "step": 59580
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002975980954810186,
+      "loss": 0.5639,
+      "step": 59590
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00029751307231220506,
+      "loss": 0.4695,
+      "step": 59600
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002974280491433916,
+      "loss": 0.5043,
+      "step": 59610
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002973430259745781,
+      "loss": 0.4862,
+      "step": 59620
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002972580028057646,
+      "loss": 0.5222,
+      "step": 59630
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002971729796369511,
+      "loss": 0.4905,
+      "step": 59640
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00029708795646813757,
+      "loss": 0.4885,
+      "step": 59650
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002970029332993241,
+      "loss": 0.4826,
+      "step": 59660
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029691791013051056,
+      "loss": 0.4914,
+      "step": 59670
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002968328869616971,
+      "loss": 0.5018,
+      "step": 59680
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002967478637928836,
+      "loss": 0.4935,
+      "step": 59690
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002966628406240701,
+      "loss": 0.4744,
+      "step": 59700
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029657781745525655,
+      "loss": 0.4844,
+      "step": 59710
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002964927942864431,
+      "loss": 0.4797,
+      "step": 59720
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002964077711176296,
+      "loss": 0.4615,
+      "step": 59730
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029632274794881607,
+      "loss": 0.549,
+      "step": 59740
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029623772478000254,
+      "loss": 0.4684,
+      "step": 59750
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00029615270161118906,
+      "loss": 0.4843,
+      "step": 59760
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0002960676784423756,
+      "loss": 0.4949,
+      "step": 59770
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00029598265527356205,
+      "loss": 0.4953,
+      "step": 59780
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002958976321047485,
+      "loss": 0.4941,
+      "step": 59790
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002958126089359351,
+      "loss": 0.4859,
+      "step": 59800
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00029572758576712157,
+      "loss": 0.4794,
+      "step": 59810
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00029564256259830804,
+      "loss": 0.4994,
+      "step": 59820
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002955575394294945,
+      "loss": 0.4913,
+      "step": 59830
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002954725162606811,
+      "loss": 0.4798,
+      "step": 59840
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00029538749309186756,
+      "loss": 0.4818,
+      "step": 59850
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00029530246992305403,
+      "loss": 0.4723,
+      "step": 59860
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002952174467542405,
+      "loss": 0.5,
+      "step": 59870
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002951324235854271,
+      "loss": 0.4664,
+      "step": 59880
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029504740041661354,
+      "loss": 0.4955,
+      "step": 59890
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002949623772478,
+      "loss": 0.4889,
+      "step": 59900
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029487735407898654,
+      "loss": 0.4914,
+      "step": 59910
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029479233091017306,
+      "loss": 0.5391,
+      "step": 59920
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029470730774135953,
+      "loss": 0.4874,
+      "step": 59930
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.000294622284572546,
+      "loss": 0.4584,
+      "step": 59940
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002945372614037325,
+      "loss": 0.5094,
+      "step": 59950
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029445223823491905,
+      "loss": 0.4869,
+      "step": 59960
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002943672150661055,
+      "loss": 0.4499,
+      "step": 59970
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029428219189729204,
+      "loss": 0.5522,
+      "step": 59980
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002941971687284785,
+      "loss": 0.4489,
+      "step": 59990
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00029411214555966503,
+      "loss": 0.4855,
+      "step": 60000
+    },
+    {
+      "epoch": 5.07,
+      "eval_loss": 0.5830700993537903,
+      "eval_runtime": 312.2014,
+      "eval_samples_per_second": 16.832,
+      "eval_steps_per_second": 2.104,
+      "step": 60000
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002940271223908515,
+      "loss": 0.4732,
+      "step": 60010
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00029394209922203803,
+      "loss": 0.4598,
+      "step": 60020
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002938570760532245,
+      "loss": 0.5199,
+      "step": 60030
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.000293772052884411,
+      "loss": 0.4938,
+      "step": 60040
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002936870297155975,
+      "loss": 0.5678,
+      "step": 60050
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.000293602006546784,
+      "loss": 0.4848,
+      "step": 60060
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002935169833779705,
+      "loss": 0.4972,
+      "step": 60070
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.000293431960209157,
+      "loss": 0.5744,
+      "step": 60080
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00029334693704034353,
+      "loss": 0.4444,
+      "step": 60090
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00029326191387153,
+      "loss": 0.4849,
+      "step": 60100
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002931768907027165,
+      "loss": 0.4313,
+      "step": 60110
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.000293091867533903,
+      "loss": 0.4932,
+      "step": 60120
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002930068443650895,
+      "loss": 0.484,
+      "step": 60130
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.000292921821196276,
+      "loss": 0.54,
+      "step": 60140
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002928367980274625,
+      "loss": 0.4837,
+      "step": 60150
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.000292751774858649,
+      "loss": 0.4349,
+      "step": 60160
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002926667516898355,
+      "loss": 0.4789,
+      "step": 60170
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.000292581728521022,
+      "loss": 0.4566,
+      "step": 60180
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002924967053522085,
+      "loss": 0.4837,
+      "step": 60190
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.000292411682183395,
+      "loss": 0.581,
+      "step": 60200
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002923266590145815,
+      "loss": 0.5337,
+      "step": 60210
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00029224163584576796,
+      "loss": 0.4478,
+      "step": 60220
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002921566126769545,
+      "loss": 0.4631,
+      "step": 60230
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.000292071589508141,
+      "loss": 0.5467,
+      "step": 60240
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002919865663393275,
+      "loss": 0.4949,
+      "step": 60250
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029190154317051395,
+      "loss": 0.5019,
+      "step": 60260
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029181652000170047,
+      "loss": 0.4497,
+      "step": 60270
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.000291731496832887,
+      "loss": 0.594,
+      "step": 60280
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029164647366407346,
+      "loss": 0.4935,
+      "step": 60290
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029156145049525993,
+      "loss": 0.4719,
+      "step": 60300
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002914764273264465,
+      "loss": 0.5156,
+      "step": 60310
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.000291391404157633,
+      "loss": 0.503,
+      "step": 60320
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029130638098881945,
+      "loss": 0.5805,
+      "step": 60330
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002912213578200059,
+      "loss": 0.4737,
+      "step": 60340
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002911363346511925,
+      "loss": 0.517,
+      "step": 60350
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00029105131148237897,
+      "loss": 0.4966,
+      "step": 60360
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00029096628831356544,
+      "loss": 0.5178,
+      "step": 60370
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00029088126514475196,
+      "loss": 0.5573,
+      "step": 60380
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002907962419759385,
+      "loss": 0.4765,
+      "step": 60390
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00029071121880712495,
+      "loss": 0.528,
+      "step": 60400
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002906261956383114,
+      "loss": 0.4935,
+      "step": 60410
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00029054117246949795,
+      "loss": 0.5286,
+      "step": 60420
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00029045614930068447,
+      "loss": 0.4454,
+      "step": 60430
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00029037112613187094,
+      "loss": 0.5088,
+      "step": 60440
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002902861029630574,
+      "loss": 0.5247,
+      "step": 60450
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00029020107979424393,
+      "loss": 0.5154,
+      "step": 60460
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00029011605662543046,
+      "loss": 0.471,
+      "step": 60470
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00029003103345661693,
+      "loss": 0.4573,
+      "step": 60480
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028994601028780345,
+      "loss": 0.4916,
+      "step": 60490
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002898609871189899,
+      "loss": 0.5033,
+      "step": 60500
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028977596395017644,
+      "loss": 0.4837,
+      "step": 60510
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002896909407813629,
+      "loss": 0.5612,
+      "step": 60520
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028960591761254944,
+      "loss": 0.586,
+      "step": 60530
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028952089444373596,
+      "loss": 0.525,
+      "step": 60540
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028943587127492243,
+      "loss": 0.5425,
+      "step": 60550
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002893508481061089,
+      "loss": 0.5136,
+      "step": 60560
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002892658249372954,
+      "loss": 0.4688,
+      "step": 60570
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00028918080176848195,
+      "loss": 0.427,
+      "step": 60580
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002890957785996684,
+      "loss": 0.544,
+      "step": 60590
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00028901075543085494,
+      "loss": 0.5098,
+      "step": 60600
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002889257322620414,
+      "loss": 0.5192,
+      "step": 60610
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00028884070909322793,
+      "loss": 0.5352,
+      "step": 60620
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002887556859244144,
+      "loss": 0.451,
+      "step": 60630
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00028867066275560093,
+      "loss": 0.5175,
+      "step": 60640
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002885856395867874,
+      "loss": 0.4722,
+      "step": 60650
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002885006164179739,
+      "loss": 0.4553,
+      "step": 60660
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00028841559324916044,
+      "loss": 0.4798,
+      "step": 60670
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002883305700803469,
+      "loss": 0.5005,
+      "step": 60680
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002882455469115334,
+      "loss": 0.5077,
+      "step": 60690
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002881605237427199,
+      "loss": 0.5487,
+      "step": 60700
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00028807550057390643,
+      "loss": 0.5321,
+      "step": 60710
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002879904774050929,
+      "loss": 0.472,
+      "step": 60720
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00028790545423627937,
+      "loss": 0.4895,
+      "step": 60730
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002878204310674659,
+      "loss": 0.5164,
+      "step": 60740
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002877354078986524,
+      "loss": 0.5644,
+      "step": 60750
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002876503847298389,
+      "loss": 0.4857,
+      "step": 60760
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00028756536156102536,
+      "loss": 0.4808,
+      "step": 60770
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00028748033839221194,
+      "loss": 0.5009,
+      "step": 60780
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002873953152233984,
+      "loss": 0.4585,
+      "step": 60790
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002873102920545849,
+      "loss": 0.5052,
+      "step": 60800
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00028722526888577134,
+      "loss": 0.5446,
+      "step": 60810
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002871402457169579,
+      "loss": 0.443,
+      "step": 60820
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002870552225481444,
+      "loss": 0.5022,
+      "step": 60830
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028697019937933086,
+      "loss": 0.4994,
+      "step": 60840
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002868851762105174,
+      "loss": 0.511,
+      "step": 60850
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002868001530417039,
+      "loss": 0.498,
+      "step": 60860
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002867151298728904,
+      "loss": 0.511,
+      "step": 60870
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028663010670407685,
+      "loss": 0.5127,
+      "step": 60880
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028654508353526337,
+      "loss": 0.4578,
+      "step": 60890
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002864600603664499,
+      "loss": 0.4588,
+      "step": 60900
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028637503719763636,
+      "loss": 0.5224,
+      "step": 60910
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028629001402882283,
+      "loss": 0.4749,
+      "step": 60920
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028620499086000936,
+      "loss": 0.5084,
+      "step": 60930
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002861199676911959,
+      "loss": 0.452,
+      "step": 60940
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00028603494452238235,
+      "loss": 0.477,
+      "step": 60950
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0002859499213535688,
+      "loss": 0.4962,
+      "step": 60960
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028586489818475534,
+      "loss": 0.4993,
+      "step": 60970
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028577987501594187,
+      "loss": 0.467,
+      "step": 60980
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028569485184712834,
+      "loss": 0.4094,
+      "step": 60990
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028560982867831486,
+      "loss": 0.4936,
+      "step": 61000
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0002855248055095014,
+      "loss": 0.4705,
+      "step": 61010
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028543978234068785,
+      "loss": 0.5212,
+      "step": 61020
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0002853547591718743,
+      "loss": 0.5002,
+      "step": 61030
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028526973600306085,
+      "loss": 0.4566,
+      "step": 61040
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028518471283424737,
+      "loss": 0.5313,
+      "step": 61050
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028509968966543384,
+      "loss": 0.4671,
+      "step": 61060
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00028501466649662036,
+      "loss": 0.5086,
+      "step": 61070
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028492964332780683,
+      "loss": 0.4169,
+      "step": 61080
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028484462015899336,
+      "loss": 0.519,
+      "step": 61090
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028475959699017983,
+      "loss": 0.4739,
+      "step": 61100
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028467457382136635,
+      "loss": 0.5243,
+      "step": 61110
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002845895506525528,
+      "loss": 0.4905,
+      "step": 61120
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028450452748373934,
+      "loss": 0.4475,
+      "step": 61130
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002844195043149258,
+      "loss": 0.4885,
+      "step": 61140
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028433448114611234,
+      "loss": 0.5293,
+      "step": 61150
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0002842494579772988,
+      "loss": 0.5089,
+      "step": 61160
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028416443480848533,
+      "loss": 0.4541,
+      "step": 61170
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00028407941163967185,
+      "loss": 0.5152,
+      "step": 61180
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002839943884708583,
+      "loss": 0.5318,
+      "step": 61190
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002839093653020448,
+      "loss": 0.4856,
+      "step": 61200
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002838243421332313,
+      "loss": 0.5223,
+      "step": 61210
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00028373931896441784,
+      "loss": 0.5204,
+      "step": 61220
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002836542957956043,
+      "loss": 0.5042,
+      "step": 61230
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002835692726267908,
+      "loss": 0.4602,
+      "step": 61240
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002834842494579773,
+      "loss": 0.4747,
+      "step": 61250
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00028339922628916383,
+      "loss": 0.5109,
+      "step": 61260
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002833142031203503,
+      "loss": 0.5415,
+      "step": 61270
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002832291799515368,
+      "loss": 0.4777,
+      "step": 61280
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00028314415678272334,
+      "loss": 0.417,
+      "step": 61290
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002830591336139098,
+      "loss": 0.4823,
+      "step": 61300
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002829741104450963,
+      "loss": 0.4398,
+      "step": 61310
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002828890872762828,
+      "loss": 0.4398,
+      "step": 61320
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00028280406410746933,
+      "loss": 0.5315,
+      "step": 61330
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002827190409386558,
+      "loss": 0.469,
+      "step": 61340
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00028263401776984227,
+      "loss": 0.5164,
+      "step": 61350
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002825489946010288,
+      "loss": 0.5108,
+      "step": 61360
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002824639714322153,
+      "loss": 0.5116,
+      "step": 61370
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002823789482634018,
+      "loss": 0.4315,
+      "step": 61380
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00028229392509458826,
+      "loss": 0.4879,
+      "step": 61390
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002822089019257748,
+      "loss": 0.4802,
+      "step": 61400
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002821238787569613,
+      "loss": 0.5041,
+      "step": 61410
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002820388555881478,
+      "loss": 0.586,
+      "step": 61420
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028195383241933424,
+      "loss": 0.5399,
+      "step": 61430
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028186880925052077,
+      "loss": 0.5129,
+      "step": 61440
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002817837860817073,
+      "loss": 0.5028,
+      "step": 61450
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028169876291289376,
+      "loss": 0.4577,
+      "step": 61460
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002816137397440803,
+      "loss": 0.4712,
+      "step": 61470
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002815287165752668,
+      "loss": 0.4314,
+      "step": 61480
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002814436934064533,
+      "loss": 0.4683,
+      "step": 61490
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028135867023763975,
+      "loss": 0.4938,
+      "step": 61500
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028127364706882627,
+      "loss": 0.5012,
+      "step": 61510
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0002811886239000128,
+      "loss": 0.4715,
+      "step": 61520
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028110360073119926,
+      "loss": 0.5236,
+      "step": 61530
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00028101857756238573,
+      "loss": 0.4423,
+      "step": 61540
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00028093355439357226,
+      "loss": 0.4878,
+      "step": 61550
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002808485312247588,
+      "loss": 0.4645,
+      "step": 61560
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00028076350805594525,
+      "loss": 0.4496,
+      "step": 61570
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002806784848871318,
+      "loss": 0.5082,
+      "step": 61580
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00028059346171831824,
+      "loss": 0.5212,
+      "step": 61590
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00028050843854950477,
+      "loss": 0.4978,
+      "step": 61600
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00028042341538069124,
+      "loss": 0.4577,
+      "step": 61610
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00028033839221187776,
+      "loss": 0.4749,
+      "step": 61620
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00028025336904306423,
+      "loss": 0.5389,
+      "step": 61630
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00028016834587425075,
+      "loss": 0.4869,
+      "step": 61640
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002800833227054372,
+      "loss": 0.5632,
+      "step": 61650
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00027999829953662375,
+      "loss": 0.407,
+      "step": 61660
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002799132763678102,
+      "loss": 0.5088,
+      "step": 61670
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027982825319899674,
+      "loss": 0.4881,
+      "step": 61680
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027974323003018326,
+      "loss": 0.5457,
+      "step": 61690
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027965820686136973,
+      "loss": 0.4706,
+      "step": 61700
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002795731836925562,
+      "loss": 0.4542,
+      "step": 61710
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002794881605237427,
+      "loss": 0.4607,
+      "step": 61720
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027940313735492925,
+      "loss": 0.4778,
+      "step": 61730
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002793181141861157,
+      "loss": 0.4923,
+      "step": 61740
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027923309101730224,
+      "loss": 0.4759,
+      "step": 61750
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027914806784848877,
+      "loss": 0.4884,
+      "step": 61760
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00027906304467967524,
+      "loss": 0.5366,
+      "step": 61770
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002789780215108617,
+      "loss": 0.6017,
+      "step": 61780
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00027889299834204823,
+      "loss": 0.4631,
+      "step": 61790
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00027880797517323475,
+      "loss": 0.4678,
+      "step": 61800
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002787229520044212,
+      "loss": 0.4947,
+      "step": 61810
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002786379288356077,
+      "loss": 0.4594,
+      "step": 61820
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002785529056667942,
+      "loss": 0.4817,
+      "step": 61830
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00027846788249798074,
+      "loss": 0.4594,
+      "step": 61840
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002783828593291672,
+      "loss": 0.5366,
+      "step": 61850
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002782978361603537,
+      "loss": 0.4909,
+      "step": 61860
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002782128129915402,
+      "loss": 0.4632,
+      "step": 61870
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00027812778982272673,
+      "loss": 0.5032,
+      "step": 61880
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0002780427666539132,
+      "loss": 0.5455,
+      "step": 61890
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00027795774348509967,
+      "loss": 0.5087,
+      "step": 61900
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002778727203162862,
+      "loss": 0.4551,
+      "step": 61910
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002777876971474727,
+      "loss": 0.4822,
+      "step": 61920
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002777026739786592,
+      "loss": 0.453,
+      "step": 61930
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00027761765080984565,
+      "loss": 0.5146,
+      "step": 61940
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00027753262764103223,
+      "loss": 0.5082,
+      "step": 61950
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002774476044722187,
+      "loss": 0.5312,
+      "step": 61960
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00027736258130340517,
+      "loss": 0.4911,
+      "step": 61970
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002772775581345917,
+      "loss": 0.5251,
+      "step": 61980
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002771925349657782,
+      "loss": 0.4874,
+      "step": 61990
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002771075117969647,
+      "loss": 0.5299,
+      "step": 62000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00027702248862815116,
+      "loss": 0.4895,
+      "step": 62010
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002769374654593377,
+      "loss": 0.4099,
+      "step": 62020
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002768524422905242,
+      "loss": 0.5758,
+      "step": 62030
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002767674191217107,
+      "loss": 0.5456,
+      "step": 62040
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002766823959528972,
+      "loss": 0.4449,
+      "step": 62050
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00027659737278408367,
+      "loss": 0.4921,
+      "step": 62060
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002765123496152702,
+      "loss": 0.4438,
+      "step": 62070
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00027642732644645666,
+      "loss": 0.4808,
+      "step": 62080
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002763423032776432,
+      "loss": 0.4931,
+      "step": 62090
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00027625728010882965,
+      "loss": 0.5016,
+      "step": 62100
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002761722569400162,
+      "loss": 0.4671,
+      "step": 62110
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00027608723377120265,
+      "loss": 0.4885,
+      "step": 62120
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00027600221060238917,
+      "loss": 0.4838,
+      "step": 62130
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027591718743357564,
+      "loss": 0.4774,
+      "step": 62140
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027583216426476216,
+      "loss": 0.4775,
+      "step": 62150
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002757471410959487,
+      "loss": 0.5085,
+      "step": 62160
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027566211792713516,
+      "loss": 0.4876,
+      "step": 62170
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002755770947583216,
+      "loss": 0.4915,
+      "step": 62180
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027549207158950815,
+      "loss": 0.5303,
+      "step": 62190
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002754070484206947,
+      "loss": 0.4947,
+      "step": 62200
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027532202525188114,
+      "loss": 0.4948,
+      "step": 62210
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027523700208306767,
+      "loss": 0.4573,
+      "step": 62220
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027515197891425414,
+      "loss": 0.5334,
+      "step": 62230
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027506695574544066,
+      "loss": 0.5138,
+      "step": 62240
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00027498193257662713,
+      "loss": 0.4696,
+      "step": 62250
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00027489690940781365,
+      "loss": 0.4879,
+      "step": 62260
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002748118862390002,
+      "loss": 0.5139,
+      "step": 62270
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00027472686307018665,
+      "loss": 0.4967,
+      "step": 62280
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002746418399013731,
+      "loss": 0.4883,
+      "step": 62290
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00027455681673255964,
+      "loss": 0.488,
+      "step": 62300
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00027447179356374616,
+      "loss": 0.5302,
+      "step": 62310
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00027438677039493263,
+      "loss": 0.5345,
+      "step": 62320
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002743017472261191,
+      "loss": 0.5443,
+      "step": 62330
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002742167240573056,
+      "loss": 0.5295,
+      "step": 62340
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00027413170088849215,
+      "loss": 0.5006,
+      "step": 62350
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002740466777196786,
+      "loss": 0.511,
+      "step": 62360
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0002739616545508651,
+      "loss": 0.4991,
+      "step": 62370
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00027387663138205167,
+      "loss": 0.4954,
+      "step": 62380
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00027379160821323814,
+      "loss": 0.5022,
+      "step": 62390
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002737065850444246,
+      "loss": 0.5386,
+      "step": 62400
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002736215618756111,
+      "loss": 0.4926,
+      "step": 62410
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00027353653870679765,
+      "loss": 0.5196,
+      "step": 62420
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002734515155379841,
+      "loss": 0.5148,
+      "step": 62430
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002733664923691706,
+      "loss": 0.4762,
+      "step": 62440
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002732814692003571,
+      "loss": 0.4194,
+      "step": 62450
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00027319644603154364,
+      "loss": 0.5438,
+      "step": 62460
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002731114228627301,
+      "loss": 0.5267,
+      "step": 62470
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002730263996939166,
+      "loss": 0.5327,
+      "step": 62480
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002729413765251031,
+      "loss": 0.5196,
+      "step": 62490
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002728563533562896,
+      "loss": 0.495,
+      "step": 62500
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002727713301874761,
+      "loss": 0.4588,
+      "step": 62510
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00027268630701866257,
+      "loss": 0.4764,
+      "step": 62520
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002726012838498491,
+      "loss": 0.5219,
+      "step": 62530
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002725162606810356,
+      "loss": 0.5123,
+      "step": 62540
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002724312375122221,
+      "loss": 0.4583,
+      "step": 62550
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002723462143434086,
+      "loss": 0.5119,
+      "step": 62560
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002722611911745951,
+      "loss": 0.4995,
+      "step": 62570
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002721761680057816,
+      "loss": 0.5086,
+      "step": 62580
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00027209114483696807,
+      "loss": 0.4589,
+      "step": 62590
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002720061216681546,
+      "loss": 0.6352,
+      "step": 62600
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00027192109849934106,
+      "loss": 0.5318,
+      "step": 62610
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002718360753305276,
+      "loss": 0.5226,
+      "step": 62620
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00027175105216171406,
+      "loss": 0.4469,
+      "step": 62630
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002716660289929006,
+      "loss": 0.5633,
+      "step": 62640
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00027158100582408705,
+      "loss": 0.532,
+      "step": 62650
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002714959826552736,
+      "loss": 0.4709,
+      "step": 62660
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002714109594864601,
+      "loss": 0.4939,
+      "step": 62670
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00027132593631764657,
+      "loss": 0.4912,
+      "step": 62680
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002712409131488331,
+      "loss": 0.4757,
+      "step": 62690
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00027115588998001956,
+      "loss": 0.5815,
+      "step": 62700
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002710708668112061,
+      "loss": 0.4965,
+      "step": 62710
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00027098584364239255,
+      "loss": 0.5284,
+      "step": 62720
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002709008204735791,
+      "loss": 0.5086,
+      "step": 62730
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00027081579730476555,
+      "loss": 0.5181,
+      "step": 62740
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00027073077413595207,
+      "loss": 0.5603,
+      "step": 62750
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00027064575096713854,
+      "loss": 0.4905,
+      "step": 62760
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00027056072779832506,
+      "loss": 0.5715,
+      "step": 62770
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002704757046295116,
+      "loss": 0.5053,
+      "step": 62780
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00027039068146069806,
+      "loss": 0.5105,
+      "step": 62790
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002703056582918845,
+      "loss": 0.5027,
+      "step": 62800
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00027022063512307105,
+      "loss": 0.4806,
+      "step": 62810
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002701356119542576,
+      "loss": 0.4768,
+      "step": 62820
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00027005058878544404,
+      "loss": 0.492,
+      "step": 62830
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002699655656166305,
+      "loss": 0.5433,
+      "step": 62840
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002698805424478171,
+      "loss": 0.4537,
+      "step": 62850
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00026979551927900356,
+      "loss": 0.4979,
+      "step": 62860
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00026971049611019003,
+      "loss": 0.5699,
+      "step": 62870
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002696254729413765,
+      "loss": 0.5584,
+      "step": 62880
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002695404497725631,
+      "loss": 0.4335,
+      "step": 62890
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00026945542660374955,
+      "loss": 0.4377,
+      "step": 62900
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.000269370403434936,
+      "loss": 0.4557,
+      "step": 62910
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002692853802661225,
+      "loss": 0.5267,
+      "step": 62920
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00026920035709730906,
+      "loss": 0.5481,
+      "step": 62930
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00026911533392849553,
+      "loss": 0.4623,
+      "step": 62940
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.000269030310759682,
+      "loss": 0.4862,
+      "step": 62950
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002689452875908685,
+      "loss": 0.4699,
+      "step": 62960
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00026886026442205505,
+      "loss": 0.5548,
+      "step": 62970
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002687752412532415,
+      "loss": 0.5104,
+      "step": 62980
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.000268690218084428,
+      "loss": 0.5026,
+      "step": 62990
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002686051949156145,
+      "loss": 0.5121,
+      "step": 63000
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00026852017174680104,
+      "loss": 0.4723,
+      "step": 63010
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002684351485779875,
+      "loss": 0.4977,
+      "step": 63020
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.000268350125409174,
+      "loss": 0.5113,
+      "step": 63030
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002682651022403605,
+      "loss": 0.5194,
+      "step": 63040
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.000268180079071547,
+      "loss": 0.5291,
+      "step": 63050
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002680950559027335,
+      "loss": 0.4758,
+      "step": 63060
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00026801003273392,
+      "loss": 0.4793,
+      "step": 63070
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002679250095651065,
+      "loss": 0.491,
+      "step": 63080
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.000267839986396293,
+      "loss": 0.4609,
+      "step": 63090
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002677549632274795,
+      "loss": 0.5283,
+      "step": 63100
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.000267669940058666,
+      "loss": 0.5649,
+      "step": 63110
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00026758491688985247,
+      "loss": 0.5057,
+      "step": 63120
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.000267499893721039,
+      "loss": 0.4514,
+      "step": 63130
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002674148705522255,
+      "loss": 0.5294,
+      "step": 63140
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.000267329847383412,
+      "loss": 0.4733,
+      "step": 63150
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002672448242145985,
+      "loss": 0.4744,
+      "step": 63160
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.000267159801045785,
+      "loss": 0.5444,
+      "step": 63170
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002670747778769715,
+      "loss": 0.5348,
+      "step": 63180
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.000266989754708158,
+      "loss": 0.5499,
+      "step": 63190
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002669047315393445,
+      "loss": 0.4164,
+      "step": 63200
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00026681970837053097,
+      "loss": 0.5048,
+      "step": 63210
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002667346852017175,
+      "loss": 0.4803,
+      "step": 63220
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00026664966203290396,
+      "loss": 0.5175,
+      "step": 63230
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002665646388640905,
+      "loss": 0.4157,
+      "step": 63240
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.000266479615695277,
+      "loss": 0.5015,
+      "step": 63250
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002663945925264635,
+      "loss": 0.5014,
+      "step": 63260
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00026630956935764995,
+      "loss": 0.494,
+      "step": 63270
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00026622454618883647,
+      "loss": 0.561,
+      "step": 63280
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.000266139523020023,
+      "loss": 0.5476,
+      "step": 63290
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00026605449985120947,
+      "loss": 0.4539,
+      "step": 63300
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00026596947668239594,
+      "loss": 0.4961,
+      "step": 63310
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00026588445351358246,
+      "loss": 0.3977,
+      "step": 63320
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.000265799430344769,
+      "loss": 0.534,
+      "step": 63330
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00026571440717595545,
+      "loss": 0.458,
+      "step": 63340
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002656293840071419,
+      "loss": 0.4859,
+      "step": 63350
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002655443608383285,
+      "loss": 0.466,
+      "step": 63360
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00026545933766951497,
+      "loss": 0.4685,
+      "step": 63370
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00026537431450070144,
+      "loss": 0.4963,
+      "step": 63380
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002652892913318879,
+      "loss": 0.5287,
+      "step": 63390
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002652042681630745,
+      "loss": 0.4446,
+      "step": 63400
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00026511924499426096,
+      "loss": 0.5793,
+      "step": 63410
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002650342218254474,
+      "loss": 0.5257,
+      "step": 63420
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00026494919865663395,
+      "loss": 0.5221,
+      "step": 63430
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002648641754878205,
+      "loss": 0.5265,
+      "step": 63440
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026477915231900694,
+      "loss": 0.5081,
+      "step": 63450
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002646941291501934,
+      "loss": 0.5061,
+      "step": 63460
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026460910598137994,
+      "loss": 0.5214,
+      "step": 63470
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026452408281256646,
+      "loss": 0.5103,
+      "step": 63480
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026443905964375293,
+      "loss": 0.5231,
+      "step": 63490
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002643540364749394,
+      "loss": 0.5245,
+      "step": 63500
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002642690133061259,
+      "loss": 0.5119,
+      "step": 63510
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026418399013731245,
+      "loss": 0.4617,
+      "step": 63520
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002640989669684989,
+      "loss": 0.4862,
+      "step": 63530
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00026401394379968544,
+      "loss": 0.5001,
+      "step": 63540
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002639289206308719,
+      "loss": 0.5686,
+      "step": 63550
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00026384389746205843,
+      "loss": 0.5253,
+      "step": 63560
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002637588742932449,
+      "loss": 0.4234,
+      "step": 63570
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002636738511244314,
+      "loss": 0.4575,
+      "step": 63580
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00026358882795561795,
+      "loss": 0.5158,
+      "step": 63590
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002635038047868044,
+      "loss": 0.4906,
+      "step": 63600
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002634187816179909,
+      "loss": 0.5376,
+      "step": 63610
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002633337584491774,
+      "loss": 0.4519,
+      "step": 63620
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00026324873528036394,
+      "loss": 0.5592,
+      "step": 63630
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002631637121115504,
+      "loss": 0.4816,
+      "step": 63640
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00026307868894273693,
+      "loss": 0.472,
+      "step": 63650
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002629936657739234,
+      "loss": 0.5327,
+      "step": 63660
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0002629086426051099,
+      "loss": 0.5455,
+      "step": 63670
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002628236194362964,
+      "loss": 0.5052,
+      "step": 63680
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002627385962674829,
+      "loss": 0.5199,
+      "step": 63690
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002626535730986694,
+      "loss": 0.4731,
+      "step": 63700
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002625685499298559,
+      "loss": 0.4779,
+      "step": 63710
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002624835267610424,
+      "loss": 0.5842,
+      "step": 63720
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002623985035922289,
+      "loss": 0.4997,
+      "step": 63730
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00026231348042341537,
+      "loss": 0.4947,
+      "step": 63740
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002622284572546019,
+      "loss": 0.4913,
+      "step": 63750
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002621434340857884,
+      "loss": 0.509,
+      "step": 63760
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002620584109169749,
+      "loss": 0.4881,
+      "step": 63770
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00026197338774816136,
+      "loss": 0.4749,
+      "step": 63780
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002618883645793479,
+      "loss": 0.4954,
+      "step": 63790
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002618033414105344,
+      "loss": 0.444,
+      "step": 63800
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002617183182417209,
+      "loss": 0.4688,
+      "step": 63810
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00026163329507290735,
+      "loss": 0.4814,
+      "step": 63820
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002615482719040939,
+      "loss": 0.4592,
+      "step": 63830
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002614632487352804,
+      "loss": 0.5477,
+      "step": 63840
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00026137822556646686,
+      "loss": 0.453,
+      "step": 63850
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00026129320239765333,
+      "loss": 0.559,
+      "step": 63860
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002612081792288399,
+      "loss": 0.4829,
+      "step": 63870
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002611231560600264,
+      "loss": 0.5065,
+      "step": 63880
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00026103813289121285,
+      "loss": 0.5106,
+      "step": 63890
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00026095310972239937,
+      "loss": 0.5586,
+      "step": 63900
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002608680865535859,
+      "loss": 0.5246,
+      "step": 63910
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00026078306338477237,
+      "loss": 0.4749,
+      "step": 63920
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00026069804021595884,
+      "loss": 0.5467,
+      "step": 63930
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00026061301704714536,
+      "loss": 0.4928,
+      "step": 63940
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002605279938783319,
+      "loss": 0.4605,
+      "step": 63950
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00026044297070951835,
+      "loss": 0.5073,
+      "step": 63960
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002603579475407048,
+      "loss": 0.4326,
+      "step": 63970
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00026027292437189135,
+      "loss": 0.4814,
+      "step": 63980
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00026018790120307787,
+      "loss": 0.5122,
+      "step": 63990
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00026010287803426434,
+      "loss": 0.495,
+      "step": 64000
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0002600178548654508,
+      "loss": 0.5053,
+      "step": 64010
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00025993283169663733,
+      "loss": 0.5171,
+      "step": 64020
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025984780852782386,
+      "loss": 0.4534,
+      "step": 64030
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002597627853590103,
+      "loss": 0.508,
+      "step": 64040
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025967776219019685,
+      "loss": 0.4611,
+      "step": 64050
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025959273902138337,
+      "loss": 0.4593,
+      "step": 64060
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025950771585256984,
+      "loss": 0.5679,
+      "step": 64070
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002594226926837563,
+      "loss": 0.4625,
+      "step": 64080
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025933766951494284,
+      "loss": 0.4373,
+      "step": 64090
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025925264634612936,
+      "loss": 0.471,
+      "step": 64100
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025916762317731583,
+      "loss": 0.5274,
+      "step": 64110
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002590826000085023,
+      "loss": 0.5154,
+      "step": 64120
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002589975768396888,
+      "loss": 0.5409,
+      "step": 64130
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00025891255367087535,
+      "loss": 0.5542,
+      "step": 64140
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002588275305020618,
+      "loss": 0.5048,
+      "step": 64150
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00025874250733324834,
+      "loss": 0.4404,
+      "step": 64160
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002586574841644348,
+      "loss": 0.5011,
+      "step": 64170
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00025857246099562133,
+      "loss": 0.463,
+      "step": 64180
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002584874378268078,
+      "loss": 0.4316,
+      "step": 64190
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002584024146579943,
+      "loss": 0.5242,
+      "step": 64200
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002583173914891808,
+      "loss": 0.4937,
+      "step": 64210
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002582323683203673,
+      "loss": 0.4359,
+      "step": 64220
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00025814734515155384,
+      "loss": 0.4418,
+      "step": 64230
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002580623219827403,
+      "loss": 0.4974,
+      "step": 64240
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002579772988139268,
+      "loss": 0.5156,
+      "step": 64250
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002578922756451133,
+      "loss": 0.5266,
+      "step": 64260
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00025780725247629983,
+      "loss": 0.5032,
+      "step": 64270
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002577222293074863,
+      "loss": 0.4588,
+      "step": 64280
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00025763720613867277,
+      "loss": 0.5343,
+      "step": 64290
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002575521829698593,
+      "loss": 0.4636,
+      "step": 64300
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002574671598010458,
+      "loss": 0.4727,
+      "step": 64310
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002573821366322323,
+      "loss": 0.4843,
+      "step": 64320
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00025729711346341875,
+      "loss": 0.5181,
+      "step": 64330
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00025721209029460533,
+      "loss": 0.4919,
+      "step": 64340
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002571270671257918,
+      "loss": 0.5088,
+      "step": 64350
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00025704204395697827,
+      "loss": 0.5001,
+      "step": 64360
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002569570207881648,
+      "loss": 0.5001,
+      "step": 64370
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002568719976193513,
+      "loss": 0.5417,
+      "step": 64380
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002567869744505378,
+      "loss": 0.5234,
+      "step": 64390
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00025670195128172426,
+      "loss": 0.5243,
+      "step": 64400
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002566169281129108,
+      "loss": 0.4647,
+      "step": 64410
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002565319049440973,
+      "loss": 0.4911,
+      "step": 64420
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002564468817752838,
+      "loss": 0.4955,
+      "step": 64430
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00025636185860647024,
+      "loss": 0.491,
+      "step": 64440
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00025627683543765677,
+      "loss": 0.5125,
+      "step": 64450
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002561918122688433,
+      "loss": 0.5088,
+      "step": 64460
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00025610678910002976,
+      "loss": 0.5279,
+      "step": 64470
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00025602176593121623,
+      "loss": 0.5191,
+      "step": 64480
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00025593674276240276,
+      "loss": 0.4659,
+      "step": 64490
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002558517195935893,
+      "loss": 0.4861,
+      "step": 64500
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025576669642477575,
+      "loss": 0.4896,
+      "step": 64510
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025568167325596227,
+      "loss": 0.5648,
+      "step": 64520
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002555966500871488,
+      "loss": 0.518,
+      "step": 64530
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025551162691833527,
+      "loss": 0.4247,
+      "step": 64540
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025542660374952173,
+      "loss": 0.4377,
+      "step": 64550
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025534158058070826,
+      "loss": 0.5103,
+      "step": 64560
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002552565574118948,
+      "loss": 0.4895,
+      "step": 64570
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025517153424308125,
+      "loss": 0.4934,
+      "step": 64580
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002550865110742677,
+      "loss": 0.503,
+      "step": 64590
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025500148790545425,
+      "loss": 0.5095,
+      "step": 64600
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00025491646473664077,
+      "loss": 0.5076,
+      "step": 64610
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00025483144156782724,
+      "loss": 0.4776,
+      "step": 64620
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00025474641839901376,
+      "loss": 0.5188,
+      "step": 64630
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00025466139523020023,
+      "loss": 0.4765,
+      "step": 64640
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00025457637206138676,
+      "loss": 0.4448,
+      "step": 64650
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002544913488925732,
+      "loss": 0.4631,
+      "step": 64660
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00025440632572375975,
+      "loss": 0.4652,
+      "step": 64670
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002543213025549462,
+      "loss": 0.4559,
+      "step": 64680
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00025423627938613274,
+      "loss": 0.4588,
+      "step": 64690
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002541512562173192,
+      "loss": 0.4967,
+      "step": 64700
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00025406623304850574,
+      "loss": 0.5217,
+      "step": 64710
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002539812098796922,
+      "loss": 0.5116,
+      "step": 64720
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00025389618671087873,
+      "loss": 0.5638,
+      "step": 64730
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00025381116354206525,
+      "loss": 0.4975,
+      "step": 64740
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002537261403732517,
+      "loss": 0.4686,
+      "step": 64750
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002536411172044382,
+      "loss": 0.4902,
+      "step": 64760
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002535560940356247,
+      "loss": 0.4873,
+      "step": 64770
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00025347107086681124,
+      "loss": 0.4976,
+      "step": 64780
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002533860476979977,
+      "loss": 0.4917,
+      "step": 64790
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00025330102452918423,
+      "loss": 0.4937,
+      "step": 64800
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002532160013603707,
+      "loss": 0.4543,
+      "step": 64810
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002531309781915572,
+      "loss": 0.4909,
+      "step": 64820
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002530459550227437,
+      "loss": 0.4925,
+      "step": 64830
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002529609318539302,
+      "loss": 0.5464,
+      "step": 64840
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00025287590868511674,
+      "loss": 0.418,
+      "step": 64850
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002527908855163032,
+      "loss": 0.526,
+      "step": 64860
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002527058623474897,
+      "loss": 0.545,
+      "step": 64870
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002526208391786762,
+      "loss": 0.5109,
+      "step": 64880
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00025253581600986273,
+      "loss": 0.5044,
+      "step": 64890
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002524507928410492,
+      "loss": 0.5584,
+      "step": 64900
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00025236576967223567,
+      "loss": 0.5195,
+      "step": 64910
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002522807465034222,
+      "loss": 0.4935,
+      "step": 64920
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002521957233346087,
+      "loss": 0.4502,
+      "step": 64930
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002521107001657952,
+      "loss": 0.452,
+      "step": 64940
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00025202567699698165,
+      "loss": 0.4932,
+      "step": 64950
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002519406538281682,
+      "loss": 0.5457,
+      "step": 64960
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002518556306593547,
+      "loss": 0.4951,
+      "step": 64970
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00025177060749054117,
+      "loss": 0.494,
+      "step": 64980
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00025168558432172764,
+      "loss": 0.4465,
+      "step": 64990
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002516005611529142,
+      "loss": 0.5003,
+      "step": 65000
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002515155379841007,
+      "loss": 0.4845,
+      "step": 65010
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00025143051481528716,
+      "loss": 0.4399,
+      "step": 65020
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002513454916464737,
+      "loss": 0.4593,
+      "step": 65030
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002512604684776602,
+      "loss": 0.5178,
+      "step": 65040
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002511754453088467,
+      "loss": 0.4543,
+      "step": 65050
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00025109042214003314,
+      "loss": 0.5415,
+      "step": 65060
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00025100539897121967,
+      "loss": 0.4492,
+      "step": 65070
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0002509203758024062,
+      "loss": 0.5056,
+      "step": 65080
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00025083535263359266,
+      "loss": 0.5397,
+      "step": 65090
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00025075032946477913,
+      "loss": 0.4909,
+      "step": 65100
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00025066530629596565,
+      "loss": 0.5021,
+      "step": 65110
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0002505802831271522,
+      "loss": 0.4794,
+      "step": 65120
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00025049525995833865,
+      "loss": 0.5084,
+      "step": 65130
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00025041023678952517,
+      "loss": 0.5127,
+      "step": 65140
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00025032521362071164,
+      "loss": 0.4475,
+      "step": 65150
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00025024019045189817,
+      "loss": 0.4711,
+      "step": 65160
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00025015516728308463,
+      "loss": 0.5201,
+      "step": 65170
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00025007014411427116,
+      "loss": 0.4907,
+      "step": 65180
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00024998512094545763,
+      "loss": 0.5262,
+      "step": 65190
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00024990009777664415,
+      "loss": 0.5324,
+      "step": 65200
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002498150746078306,
+      "loss": 0.5344,
+      "step": 65210
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024973005143901714,
+      "loss": 0.4628,
+      "step": 65220
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002496450282702036,
+      "loss": 0.4633,
+      "step": 65230
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024956000510139014,
+      "loss": 0.5252,
+      "step": 65240
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024947498193257666,
+      "loss": 0.4985,
+      "step": 65250
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024938995876376313,
+      "loss": 0.4964,
+      "step": 65260
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024930493559494966,
+      "loss": 0.4631,
+      "step": 65270
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002492199124261361,
+      "loss": 0.4409,
+      "step": 65280
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024913488925732265,
+      "loss": 0.5184,
+      "step": 65290
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002490498660885091,
+      "loss": 0.583,
+      "step": 65300
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024896484291969564,
+      "loss": 0.5584,
+      "step": 65310
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024887981975088217,
+      "loss": 0.4933,
+      "step": 65320
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00024879479658206863,
+      "loss": 0.524,
+      "step": 65330
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002487097734132551,
+      "loss": 0.4502,
+      "step": 65340
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00024862475024444163,
+      "loss": 0.5523,
+      "step": 65350
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00024853972707562815,
+      "loss": 0.5353,
+      "step": 65360
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002484547039068146,
+      "loss": 0.4905,
+      "step": 65370
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002483696807380011,
+      "loss": 0.5547,
+      "step": 65380
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002482846575691876,
+      "loss": 0.488,
+      "step": 65390
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00024819963440037414,
+      "loss": 0.4889,
+      "step": 65400
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002481146112315606,
+      "loss": 0.5284,
+      "step": 65410
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002480295880627471,
+      "loss": 0.461,
+      "step": 65420
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00024794456489393366,
+      "loss": 0.4877,
+      "step": 65430
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002478595417251201,
+      "loss": 0.4861,
+      "step": 65440
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002477745185563066,
+      "loss": 0.4886,
+      "step": 65450
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024768949538749306,
+      "loss": 0.5427,
+      "step": 65460
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024760447221867964,
+      "loss": 0.4391,
+      "step": 65470
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002475194490498661,
+      "loss": 0.5782,
+      "step": 65480
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002474344258810526,
+      "loss": 0.4733,
+      "step": 65490
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024734940271223905,
+      "loss": 0.4951,
+      "step": 65500
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024726437954342563,
+      "loss": 0.4933,
+      "step": 65510
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002471793563746121,
+      "loss": 0.5042,
+      "step": 65520
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024709433320579857,
+      "loss": 0.5014,
+      "step": 65530
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002470093100369851,
+      "loss": 0.4516,
+      "step": 65540
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002469242868681716,
+      "loss": 0.5002,
+      "step": 65550
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002468392636993581,
+      "loss": 0.5316,
+      "step": 65560
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024675424053054455,
+      "loss": 0.5387,
+      "step": 65570
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002466692173617311,
+      "loss": 0.534,
+      "step": 65580
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002465841941929176,
+      "loss": 0.4802,
+      "step": 65590
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024649917102410407,
+      "loss": 0.4918,
+      "step": 65600
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002464141478552906,
+      "loss": 0.4815,
+      "step": 65610
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024632912468647706,
+      "loss": 0.4854,
+      "step": 65620
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002462441015176636,
+      "loss": 0.462,
+      "step": 65630
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024615907834885006,
+      "loss": 0.5524,
+      "step": 65640
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002460740551800366,
+      "loss": 0.4758,
+      "step": 65650
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024598903201122305,
+      "loss": 0.5059,
+      "step": 65660
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002459040088424096,
+      "loss": 0.5551,
+      "step": 65670
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024581898567359604,
+      "loss": 0.4558,
+      "step": 65680
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024573396250478257,
+      "loss": 0.5018,
+      "step": 65690
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024564893933596904,
+      "loss": 0.4841,
+      "step": 65700
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024556391616715556,
+      "loss": 0.5297,
+      "step": 65710
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0002454788929983421,
+      "loss": 0.5072,
+      "step": 65720
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024539386982952855,
+      "loss": 0.5123,
+      "step": 65730
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0002453088466607151,
+      "loss": 0.4741,
+      "step": 65740
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024522382349190155,
+      "loss": 0.5241,
+      "step": 65750
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024513880032308807,
+      "loss": 0.5102,
+      "step": 65760
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024505377715427454,
+      "loss": 0.4472,
+      "step": 65770
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024496875398546106,
+      "loss": 0.5034,
+      "step": 65780
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024488373081664753,
+      "loss": 0.5,
+      "step": 65790
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024479870764783406,
+      "loss": 0.4884,
+      "step": 65800
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024471368447902053,
+      "loss": 0.4682,
+      "step": 65810
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024462866131020705,
+      "loss": 0.4991,
+      "step": 65820
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0002445436381413936,
+      "loss": 0.5185,
+      "step": 65830
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024445861497258004,
+      "loss": 0.4555,
+      "step": 65840
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0002443735918037665,
+      "loss": 0.4666,
+      "step": 65850
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024428856863495304,
+      "loss": 0.5934,
+      "step": 65860
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024420354546613956,
+      "loss": 0.4634,
+      "step": 65870
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024411852229732603,
+      "loss": 0.5163,
+      "step": 65880
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024403349912851253,
+      "loss": 0.4715,
+      "step": 65890
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.000243948475959699,
+      "loss": 0.4839,
+      "step": 65900
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024386345279088555,
+      "loss": 0.4669,
+      "step": 65910
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024377842962207202,
+      "loss": 0.4939,
+      "step": 65920
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024369340645325851,
+      "loss": 0.4954,
+      "step": 65930
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024360838328444504,
+      "loss": 0.4609,
+      "step": 65940
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024352336011563153,
+      "loss": 0.5059,
+      "step": 65950
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.000243438336946818,
+      "loss": 0.5064,
+      "step": 65960
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0002433533137780045,
+      "loss": 0.4681,
+      "step": 65970
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024326829060919102,
+      "loss": 0.4922,
+      "step": 65980
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024318326744037752,
+      "loss": 0.4726,
+      "step": 65990
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.000243098244271564,
+      "loss": 0.519,
+      "step": 66000
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024301322110275054,
+      "loss": 0.5152,
+      "step": 66010
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.000242928197933937,
+      "loss": 0.5483,
+      "step": 66020
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0002428431747651235,
+      "loss": 0.4946,
+      "step": 66030
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024275815159630998,
+      "loss": 0.4476,
+      "step": 66040
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024267312842749653,
+      "loss": 0.465,
+      "step": 66050
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.000242588105258683,
+      "loss": 0.4854,
+      "step": 66060
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0002425030820898695,
+      "loss": 0.4853,
+      "step": 66070
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.000242418058921056,
+      "loss": 0.5722,
+      "step": 66080
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024233303575224251,
+      "loss": 0.5613,
+      "step": 66090
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024224801258342898,
+      "loss": 0.5547,
+      "step": 66100
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024216298941461548,
+      "loss": 0.5137,
+      "step": 66110
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.000242077966245802,
+      "loss": 0.5223,
+      "step": 66120
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0002419929430769885,
+      "loss": 0.4998,
+      "step": 66130
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.000241907919908175,
+      "loss": 0.4969,
+      "step": 66140
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00024182289673936147,
+      "loss": 0.4996,
+      "step": 66150
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.000241737873570548,
+      "loss": 0.458,
+      "step": 66160
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0002416528504017345,
+      "loss": 0.5181,
+      "step": 66170
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024156782723292098,
+      "loss": 0.4881,
+      "step": 66180
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024148280406410745,
+      "loss": 0.4543,
+      "step": 66190
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024139778089529398,
+      "loss": 0.5016,
+      "step": 66200
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024131275772648047,
+      "loss": 0.4884,
+      "step": 66210
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024122773455766697,
+      "loss": 0.5226,
+      "step": 66220
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0002411427113888535,
+      "loss": 0.5098,
+      "step": 66230
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024105768822004,
+      "loss": 0.4806,
+      "step": 66240
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024097266505122646,
+      "loss": 0.4856,
+      "step": 66250
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024088764188241296,
+      "loss": 0.5064,
+      "step": 66260
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024080261871359948,
+      "loss": 0.4996,
+      "step": 66270
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00024071759554478598,
+      "loss": 0.5138,
+      "step": 66280
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00024063257237597245,
+      "loss": 0.5208,
+      "step": 66290
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00024054754920715897,
+      "loss": 0.4795,
+      "step": 66300
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00024046252603834547,
+      "loss": 0.5132,
+      "step": 66310
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00024037750286953196,
+      "loss": 0.5615,
+      "step": 66320
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00024029247970071843,
+      "loss": 0.4841,
+      "step": 66330
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00024020745653190498,
+      "loss": 0.4631,
+      "step": 66340
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00024012243336309145,
+      "loss": 0.5145,
+      "step": 66350
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00024003741019427795,
+      "loss": 0.4689,
+      "step": 66360
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00023995238702546442,
+      "loss": 0.5149,
+      "step": 66370
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00023986736385665097,
+      "loss": 0.502,
+      "step": 66380
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00023978234068783744,
+      "loss": 0.4705,
+      "step": 66390
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023969731751902394,
+      "loss": 0.4362,
+      "step": 66400
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023961229435021046,
+      "loss": 0.5054,
+      "step": 66410
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023952727118139696,
+      "loss": 0.5053,
+      "step": 66420
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023944224801258343,
+      "loss": 0.532,
+      "step": 66430
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023935722484376992,
+      "loss": 0.4706,
+      "step": 66440
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023927220167495645,
+      "loss": 0.4583,
+      "step": 66450
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023918717850614294,
+      "loss": 0.5088,
+      "step": 66460
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023910215533732941,
+      "loss": 0.4333,
+      "step": 66470
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0002390171321685159,
+      "loss": 0.4896,
+      "step": 66480
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023893210899970243,
+      "loss": 0.5034,
+      "step": 66490
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023884708583088893,
+      "loss": 0.4634,
+      "step": 66500
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00023876206266207543,
+      "loss": 0.5315,
+      "step": 66510
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023867703949326195,
+      "loss": 0.493,
+      "step": 66520
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023859201632444842,
+      "loss": 0.5373,
+      "step": 66530
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023850699315563492,
+      "loss": 0.5499,
+      "step": 66540
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023842196998682141,
+      "loss": 0.4827,
+      "step": 66550
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023833694681800794,
+      "loss": 0.5159,
+      "step": 66560
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0002382519236491944,
+      "loss": 0.5312,
+      "step": 66570
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0002381669004803809,
+      "loss": 0.4977,
+      "step": 66580
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0002380818773115674,
+      "loss": 0.463,
+      "step": 66590
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023799685414275392,
+      "loss": 0.4712,
+      "step": 66600
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00023791183097394042,
+      "loss": 0.4308,
+      "step": 66610
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0002378268078051269,
+      "loss": 0.5077,
+      "step": 66620
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023774178463631341,
+      "loss": 0.5291,
+      "step": 66630
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0002376567614674999,
+      "loss": 0.4702,
+      "step": 66640
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0002375717382986864,
+      "loss": 0.5,
+      "step": 66650
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023748671512987288,
+      "loss": 0.5367,
+      "step": 66660
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0002374016919610594,
+      "loss": 0.5265,
+      "step": 66670
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0002373166687922459,
+      "loss": 0.4804,
+      "step": 66680
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0002372316456234324,
+      "loss": 0.5425,
+      "step": 66690
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023714662245461892,
+      "loss": 0.5249,
+      "step": 66700
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023706159928580541,
+      "loss": 0.4704,
+      "step": 66710
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023697657611699188,
+      "loss": 0.497,
+      "step": 66720
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00023689155294817838,
+      "loss": 0.4643,
+      "step": 66730
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0002368065297793649,
+      "loss": 0.4906,
+      "step": 66740
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0002367215066105514,
+      "loss": 0.4484,
+      "step": 66750
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023663648344173787,
+      "loss": 0.5098,
+      "step": 66760
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023655146027292437,
+      "loss": 0.5233,
+      "step": 66770
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0002364664371041109,
+      "loss": 0.4473,
+      "step": 66780
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0002363814139352974,
+      "loss": 0.4832,
+      "step": 66790
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023629639076648386,
+      "loss": 0.4932,
+      "step": 66800
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0002362113675976704,
+      "loss": 0.4521,
+      "step": 66810
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023612634442885688,
+      "loss": 0.517,
+      "step": 66820
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023604132126004337,
+      "loss": 0.5447,
+      "step": 66830
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023595629809122984,
+      "loss": 0.5494,
+      "step": 66840
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0002358712749224164,
+      "loss": 0.4416,
+      "step": 66850
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00023578625175360286,
+      "loss": 0.5008,
+      "step": 66860
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023570122858478936,
+      "loss": 0.4452,
+      "step": 66870
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023561620541597586,
+      "loss": 0.576,
+      "step": 66880
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023553118224716238,
+      "loss": 0.5484,
+      "step": 66890
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023544615907834885,
+      "loss": 0.4873,
+      "step": 66900
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023536113590953535,
+      "loss": 0.5159,
+      "step": 66910
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023527611274072187,
+      "loss": 0.4707,
+      "step": 66920
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023519108957190837,
+      "loss": 0.4807,
+      "step": 66930
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023510606640309484,
+      "loss": 0.5514,
+      "step": 66940
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023502104323428133,
+      "loss": 0.4946,
+      "step": 66950
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023493602006546786,
+      "loss": 0.483,
+      "step": 66960
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023485099689665435,
+      "loss": 0.5472,
+      "step": 66970
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00023476597372784085,
+      "loss": 0.589,
+      "step": 66980
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023468095055902737,
+      "loss": 0.5411,
+      "step": 66990
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023459592739021384,
+      "loss": 0.4706,
+      "step": 67000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023451090422140034,
+      "loss": 0.503,
+      "step": 67010
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023442588105258684,
+      "loss": 0.4536,
+      "step": 67020
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023434085788377336,
+      "loss": 0.5147,
+      "step": 67030
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023425583471495983,
+      "loss": 0.5483,
+      "step": 67040
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023417081154614633,
+      "loss": 0.4603,
+      "step": 67050
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023408578837733282,
+      "loss": 0.5401,
+      "step": 67060
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023400076520851935,
+      "loss": 0.556,
+      "step": 67070
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023391574203970584,
+      "loss": 0.4994,
+      "step": 67080
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002338307188708923,
+      "loss": 0.4806,
+      "step": 67090
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00023374569570207884,
+      "loss": 0.4918,
+      "step": 67100
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023366067253326533,
+      "loss": 0.4867,
+      "step": 67110
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023357564936445183,
+      "loss": 0.4844,
+      "step": 67120
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0002334906261956383,
+      "loss": 0.4709,
+      "step": 67130
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023340560302682482,
+      "loss": 0.5099,
+      "step": 67140
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023332057985801132,
+      "loss": 0.4491,
+      "step": 67150
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023323555668919782,
+      "loss": 0.4662,
+      "step": 67160
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0002331505335203843,
+      "loss": 0.5242,
+      "step": 67170
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023306551035157084,
+      "loss": 0.4533,
+      "step": 67180
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0002329804871827573,
+      "loss": 0.5171,
+      "step": 67190
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0002328954640139438,
+      "loss": 0.5546,
+      "step": 67200
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00023281044084513033,
+      "loss": 0.4825,
+      "step": 67210
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023272541767631682,
+      "loss": 0.4873,
+      "step": 67220
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0002326403945075033,
+      "loss": 0.5272,
+      "step": 67230
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0002325553713386898,
+      "loss": 0.4977,
+      "step": 67240
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023247034816987631,
+      "loss": 0.4384,
+      "step": 67250
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0002323853250010628,
+      "loss": 0.5476,
+      "step": 67260
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023230030183224928,
+      "loss": 0.4728,
+      "step": 67270
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023221527866343578,
+      "loss": 0.5038,
+      "step": 67280
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0002321302554946223,
+      "loss": 0.5414,
+      "step": 67290
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0002320452323258088,
+      "loss": 0.5904,
+      "step": 67300
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023196020915699527,
+      "loss": 0.4927,
+      "step": 67310
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00023187518598818182,
+      "loss": 0.4811,
+      "step": 67320
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0002317901628193683,
+      "loss": 0.4938,
+      "step": 67330
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023170513965055478,
+      "loss": 0.4909,
+      "step": 67340
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023162011648174128,
+      "loss": 0.4951,
+      "step": 67350
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0002315350933129278,
+      "loss": 0.5547,
+      "step": 67360
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023145007014411427,
+      "loss": 0.4642,
+      "step": 67370
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023136504697530077,
+      "loss": 0.4952,
+      "step": 67380
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0002312800238064873,
+      "loss": 0.5505,
+      "step": 67390
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0002311950006376738,
+      "loss": 0.5593,
+      "step": 67400
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023110997746886026,
+      "loss": 0.5378,
+      "step": 67410
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023102495430004676,
+      "loss": 0.6072,
+      "step": 67420
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023093993113123328,
+      "loss": 0.523,
+      "step": 67430
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023085490796241978,
+      "loss": 0.4392,
+      "step": 67440
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00023076988479360627,
+      "loss": 0.5049,
+      "step": 67450
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00023068486162479274,
+      "loss": 0.4832,
+      "step": 67460
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00023059983845597927,
+      "loss": 0.5115,
+      "step": 67470
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00023051481528716576,
+      "loss": 0.4671,
+      "step": 67480
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00023042979211835226,
+      "loss": 0.5864,
+      "step": 67490
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00023034476894953878,
+      "loss": 0.4894,
+      "step": 67500
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00023025974578072525,
+      "loss": 0.5073,
+      "step": 67510
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00023017472261191175,
+      "loss": 0.4954,
+      "step": 67520
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00023008969944309825,
+      "loss": 0.499,
+      "step": 67530
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00023000467627428477,
+      "loss": 0.4751,
+      "step": 67540
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022991965310547127,
+      "loss": 0.4918,
+      "step": 67550
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022983462993665774,
+      "loss": 0.5108,
+      "step": 67560
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00022974960676784423,
+      "loss": 0.4567,
+      "step": 67570
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022966458359903076,
+      "loss": 0.4489,
+      "step": 67580
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022957956043021725,
+      "loss": 0.4846,
+      "step": 67590
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022949453726140372,
+      "loss": 0.5705,
+      "step": 67600
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022940951409259027,
+      "loss": 0.473,
+      "step": 67610
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022932449092377674,
+      "loss": 0.5662,
+      "step": 67620
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022923946775496324,
+      "loss": 0.4877,
+      "step": 67630
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0002291544445861497,
+      "loss": 0.4295,
+      "step": 67640
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022906942141733626,
+      "loss": 0.5027,
+      "step": 67650
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022898439824852273,
+      "loss": 0.5282,
+      "step": 67660
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022889937507970923,
+      "loss": 0.5606,
+      "step": 67670
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0002288143519108957,
+      "loss": 0.4778,
+      "step": 67680
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00022872932874208225,
+      "loss": 0.5406,
+      "step": 67690
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022864430557326872,
+      "loss": 0.539,
+      "step": 67700
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002285592824044552,
+      "loss": 0.5421,
+      "step": 67710
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022847425923564174,
+      "loss": 0.4957,
+      "step": 67720
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022838923606682823,
+      "loss": 0.4655,
+      "step": 67730
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002283042128980147,
+      "loss": 0.5095,
+      "step": 67740
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002282191897292012,
+      "loss": 0.5076,
+      "step": 67750
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022813416656038772,
+      "loss": 0.4901,
+      "step": 67760
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022804914339157422,
+      "loss": 0.5344,
+      "step": 67770
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002279641202227607,
+      "loss": 0.527,
+      "step": 67780
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00022787909705394724,
+      "loss": 0.5324,
+      "step": 67790
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002277940738851337,
+      "loss": 0.452,
+      "step": 67800
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0002277090507163202,
+      "loss": 0.4944,
+      "step": 67810
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0002276240275475067,
+      "loss": 0.5522,
+      "step": 67820
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022753900437869323,
+      "loss": 0.5108,
+      "step": 67830
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0002274539812098797,
+      "loss": 0.4766,
+      "step": 67840
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0002273689580410662,
+      "loss": 0.5232,
+      "step": 67850
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0002272839348722527,
+      "loss": 0.484,
+      "step": 67860
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0002271989117034392,
+      "loss": 0.4696,
+      "step": 67870
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022711388853462568,
+      "loss": 0.4947,
+      "step": 67880
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00022702886536581218,
+      "loss": 0.5048,
+      "step": 67890
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0002269438421969987,
+      "loss": 0.5066,
+      "step": 67900
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0002268588190281852,
+      "loss": 0.5053,
+      "step": 67910
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0002267737958593717,
+      "loss": 0.4428,
+      "step": 67920
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022668877269055817,
+      "loss": 0.564,
+      "step": 67930
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0002266037495217447,
+      "loss": 0.4271,
+      "step": 67940
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0002265187263529312,
+      "loss": 0.486,
+      "step": 67950
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022643370318411768,
+      "loss": 0.5116,
+      "step": 67960
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022634868001530415,
+      "loss": 0.4969,
+      "step": 67970
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0002262636568464907,
+      "loss": 0.4605,
+      "step": 67980
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022617863367767717,
+      "loss": 0.5006,
+      "step": 67990
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022609361050886367,
+      "loss": 0.4834,
+      "step": 68000
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0002260085873400502,
+      "loss": 0.5112,
+      "step": 68010
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0002259235641712367,
+      "loss": 0.5074,
+      "step": 68020
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022583854100242316,
+      "loss": 0.4693,
+      "step": 68030
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00022575351783360966,
+      "loss": 0.5264,
+      "step": 68040
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022566849466479618,
+      "loss": 0.569,
+      "step": 68050
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022558347149598268,
+      "loss": 0.4827,
+      "step": 68060
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022549844832716915,
+      "loss": 0.4796,
+      "step": 68070
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0002254134251583557,
+      "loss": 0.5039,
+      "step": 68080
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022532840198954217,
+      "loss": 0.4714,
+      "step": 68090
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022524337882072866,
+      "loss": 0.5486,
+      "step": 68100
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022515835565191513,
+      "loss": 0.5213,
+      "step": 68110
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022507333248310168,
+      "loss": 0.4844,
+      "step": 68120
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022498830931428815,
+      "loss": 0.4658,
+      "step": 68130
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022490328614547465,
+      "loss": 0.4964,
+      "step": 68140
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022481826297666112,
+      "loss": 0.4696,
+      "step": 68150
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00022473323980784767,
+      "loss": 0.4957,
+      "step": 68160
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022464821663903414,
+      "loss": 0.5638,
+      "step": 68170
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022456319347022064,
+      "loss": 0.5189,
+      "step": 68180
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022447817030140716,
+      "loss": 0.466,
+      "step": 68190
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022439314713259366,
+      "loss": 0.5068,
+      "step": 68200
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022430812396378013,
+      "loss": 0.4977,
+      "step": 68210
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022422310079496662,
+      "loss": 0.5116,
+      "step": 68220
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022413807762615315,
+      "loss": 0.4889,
+      "step": 68230
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022405305445733964,
+      "loss": 0.5189,
+      "step": 68240
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0002239680312885261,
+      "loss": 0.4695,
+      "step": 68250
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0002238830081197126,
+      "loss": 0.512,
+      "step": 68260
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022379798495089913,
+      "loss": 0.4699,
+      "step": 68270
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00022371296178208563,
+      "loss": 0.4525,
+      "step": 68280
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022362793861327213,
+      "loss": 0.481,
+      "step": 68290
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022354291544445865,
+      "loss": 0.4706,
+      "step": 68300
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022345789227564512,
+      "loss": 0.4885,
+      "step": 68310
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022337286910683162,
+      "loss": 0.4439,
+      "step": 68320
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0002232878459380181,
+      "loss": 0.4938,
+      "step": 68330
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022320282276920464,
+      "loss": 0.5115,
+      "step": 68340
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022311779960039113,
+      "loss": 0.5212,
+      "step": 68350
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0002230327764315776,
+      "loss": 0.4932,
+      "step": 68360
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0002229477532627641,
+      "loss": 0.4766,
+      "step": 68370
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022286273009395062,
+      "loss": 0.4609,
+      "step": 68380
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00022277770692513712,
+      "loss": 0.504,
+      "step": 68390
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0002226926837563236,
+      "loss": 0.4921,
+      "step": 68400
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002226076605875101,
+      "loss": 0.503,
+      "step": 68410
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002225226374186966,
+      "loss": 0.519,
+      "step": 68420
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002224376142498831,
+      "loss": 0.5592,
+      "step": 68430
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022235259108106958,
+      "loss": 0.5327,
+      "step": 68440
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022226756791225613,
+      "loss": 0.4669,
+      "step": 68450
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002221825447434426,
+      "loss": 0.4908,
+      "step": 68460
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002220975215746291,
+      "loss": 0.46,
+      "step": 68470
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022201249840581562,
+      "loss": 0.5839,
+      "step": 68480
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002219274752370021,
+      "loss": 0.4951,
+      "step": 68490
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022184245206818858,
+      "loss": 0.4829,
+      "step": 68500
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00022175742889937508,
+      "loss": 0.4767,
+      "step": 68510
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002216724057305616,
+      "loss": 0.5384,
+      "step": 68520
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0002215873825617481,
+      "loss": 0.5603,
+      "step": 68530
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022150235939293457,
+      "loss": 0.5256,
+      "step": 68540
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022141733622412107,
+      "loss": 0.5052,
+      "step": 68550
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0002213323130553076,
+      "loss": 0.5246,
+      "step": 68560
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022124728988649409,
+      "loss": 0.4723,
+      "step": 68570
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022116226671768056,
+      "loss": 0.5158,
+      "step": 68580
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0002210772435488671,
+      "loss": 0.4849,
+      "step": 68590
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022099222038005358,
+      "loss": 0.4762,
+      "step": 68600
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022090719721124007,
+      "loss": 0.4835,
+      "step": 68610
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00022082217404242654,
+      "loss": 0.4758,
+      "step": 68620
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0002207371508736131,
+      "loss": 0.4826,
+      "step": 68630
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00022065212770479956,
+      "loss": 0.502,
+      "step": 68640
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00022056710453598606,
+      "loss": 0.4862,
+      "step": 68650
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00022048208136717256,
+      "loss": 0.4718,
+      "step": 68660
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00022039705819835908,
+      "loss": 0.5067,
+      "step": 68670
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00022031203502954555,
+      "loss": 0.5198,
+      "step": 68680
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00022022701186073205,
+      "loss": 0.4655,
+      "step": 68690
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00022014198869191857,
+      "loss": 0.4484,
+      "step": 68700
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00022005696552310507,
+      "loss": 0.4646,
+      "step": 68710
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00021997194235429154,
+      "loss": 0.5123,
+      "step": 68720
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00021988691918547803,
+      "loss": 0.4566,
+      "step": 68730
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00021980189601666456,
+      "loss": 0.5271,
+      "step": 68740
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00021971687284785105,
+      "loss": 0.5242,
+      "step": 68750
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021963184967903755,
+      "loss": 0.5578,
+      "step": 68760
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021954682651022407,
+      "loss": 0.4612,
+      "step": 68770
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021946180334141054,
+      "loss": 0.47,
+      "step": 68780
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021937678017259704,
+      "loss": 0.4715,
+      "step": 68790
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021929175700378354,
+      "loss": 0.4957,
+      "step": 68800
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021920673383497006,
+      "loss": 0.512,
+      "step": 68810
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021912171066615656,
+      "loss": 0.53,
+      "step": 68820
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021903668749734303,
+      "loss": 0.5107,
+      "step": 68830
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021895166432852952,
+      "loss": 0.5032,
+      "step": 68840
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021886664115971605,
+      "loss": 0.4836,
+      "step": 68850
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00021878161799090254,
+      "loss": 0.4818,
+      "step": 68860
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.000218696594822089,
+      "loss": 0.4755,
+      "step": 68870
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021861157165327554,
+      "loss": 0.4907,
+      "step": 68880
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021852654848446203,
+      "loss": 0.499,
+      "step": 68890
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021844152531564853,
+      "loss": 0.5002,
+      "step": 68900
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.000218356502146835,
+      "loss": 0.4514,
+      "step": 68910
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021827147897802155,
+      "loss": 0.5123,
+      "step": 68920
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021818645580920802,
+      "loss": 0.5252,
+      "step": 68930
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021810143264039452,
+      "loss": 0.4696,
+      "step": 68940
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021801640947158099,
+      "loss": 0.4982,
+      "step": 68950
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021793138630276754,
+      "loss": 0.5168,
+      "step": 68960
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.000217846363133954,
+      "loss": 0.484,
+      "step": 68970
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0002177613399651405,
+      "loss": 0.4902,
+      "step": 68980
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00021767631679632703,
+      "loss": 0.4887,
+      "step": 68990
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021759129362751352,
+      "loss": 0.4749,
+      "step": 69000
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0002175062704587,
+      "loss": 0.4861,
+      "step": 69010
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0002174212472898865,
+      "loss": 0.516,
+      "step": 69020
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.000217336224121073,
+      "loss": 0.4919,
+      "step": 69030
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0002172512009522595,
+      "loss": 0.4599,
+      "step": 69040
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021716617778344598,
+      "loss": 0.4618,
+      "step": 69050
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021708115461463248,
+      "loss": 0.4845,
+      "step": 69060
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.000216996131445819,
+      "loss": 0.4917,
+      "step": 69070
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0002169111082770055,
+      "loss": 0.4947,
+      "step": 69080
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021682608510819197,
+      "loss": 0.5072,
+      "step": 69090
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021674106193937852,
+      "loss": 0.5189,
+      "step": 69100
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00021665603877056499,
+      "loss": 0.5069,
+      "step": 69110
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021657101560175148,
+      "loss": 0.4484,
+      "step": 69120
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021648599243293798,
+      "loss": 0.5041,
+      "step": 69130
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0002164009692641245,
+      "loss": 0.4818,
+      "step": 69140
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021631594609531097,
+      "loss": 0.473,
+      "step": 69150
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021623092292649747,
+      "loss": 0.5294,
+      "step": 69160
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.000216145899757684,
+      "loss": 0.5047,
+      "step": 69170
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0002160608765888705,
+      "loss": 0.4951,
+      "step": 69180
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021597585342005699,
+      "loss": 0.5091,
+      "step": 69190
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021589083025124346,
+      "loss": 0.4512,
+      "step": 69200
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021580580708242998,
+      "loss": 0.5165,
+      "step": 69210
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00021572078391361648,
+      "loss": 0.45,
+      "step": 69220
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021563576074480297,
+      "loss": 0.5179,
+      "step": 69230
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021555073757598944,
+      "loss": 0.4926,
+      "step": 69240
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021546571440717597,
+      "loss": 0.4899,
+      "step": 69250
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021538069123836246,
+      "loss": 0.4764,
+      "step": 69260
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021529566806954896,
+      "loss": 0.5138,
+      "step": 69270
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021521064490073548,
+      "loss": 0.5609,
+      "step": 69280
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021512562173192198,
+      "loss": 0.5074,
+      "step": 69290
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021504059856310845,
+      "loss": 0.4491,
+      "step": 69300
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021495557539429495,
+      "loss": 0.5198,
+      "step": 69310
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021487055222548147,
+      "loss": 0.4553,
+      "step": 69320
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021478552905666797,
+      "loss": 0.5249,
+      "step": 69330
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00021470050588785444,
+      "loss": 0.4832,
+      "step": 69340
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021461548271904093,
+      "loss": 0.5208,
+      "step": 69350
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021453045955022746,
+      "loss": 0.4945,
+      "step": 69360
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021444543638141395,
+      "loss": 0.5095,
+      "step": 69370
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021436041321260042,
+      "loss": 0.5179,
+      "step": 69380
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021427539004378697,
+      "loss": 0.4585,
+      "step": 69390
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021419036687497344,
+      "loss": 0.478,
+      "step": 69400
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021410534370615994,
+      "loss": 0.5262,
+      "step": 69410
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0002140203205373464,
+      "loss": 0.5257,
+      "step": 69420
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021393529736853296,
+      "loss": 0.5667,
+      "step": 69430
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021385027419971943,
+      "loss": 0.4868,
+      "step": 69440
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021376525103090593,
+      "loss": 0.5199,
+      "step": 69450
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00021368022786209245,
+      "loss": 0.4437,
+      "step": 69460
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021359520469327895,
+      "loss": 0.4914,
+      "step": 69470
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021351018152446542,
+      "loss": 0.4681,
+      "step": 69480
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0002134251583556519,
+      "loss": 0.4675,
+      "step": 69490
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021334013518683844,
+      "loss": 0.4769,
+      "step": 69500
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021325511201802493,
+      "loss": 0.4649,
+      "step": 69510
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0002131700888492114,
+      "loss": 0.5038,
+      "step": 69520
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0002130850656803979,
+      "loss": 0.4581,
+      "step": 69530
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021300004251158442,
+      "loss": 0.5106,
+      "step": 69540
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021291501934277092,
+      "loss": 0.519,
+      "step": 69550
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021282999617395742,
+      "loss": 0.5135,
+      "step": 69560
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00021274497300514394,
+      "loss": 0.4923,
+      "step": 69570
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0002126599498363304,
+      "loss": 0.4406,
+      "step": 69580
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0002125749266675169,
+      "loss": 0.4826,
+      "step": 69590
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0002124899034987034,
+      "loss": 0.5117,
+      "step": 69600
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021240488032988993,
+      "loss": 0.488,
+      "step": 69610
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0002123198571610764,
+      "loss": 0.5085,
+      "step": 69620
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0002122348339922629,
+      "loss": 0.4985,
+      "step": 69630
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0002121498108234494,
+      "loss": 0.4364,
+      "step": 69640
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0002120647876546359,
+      "loss": 0.4483,
+      "step": 69650
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0002119797644858224,
+      "loss": 0.4951,
+      "step": 69660
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00021189474131700888,
+      "loss": 0.5168,
+      "step": 69670
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0002118097181481954,
+      "loss": 0.4788,
+      "step": 69680
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0002117246949793819,
+      "loss": 0.5091,
+      "step": 69690
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0002116396718105684,
+      "loss": 0.4343,
+      "step": 69700
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021155464864175487,
+      "loss": 0.5436,
+      "step": 69710
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0002114696254729414,
+      "loss": 0.5083,
+      "step": 69720
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021138460230412789,
+      "loss": 0.5207,
+      "step": 69730
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021129957913531438,
+      "loss": 0.5056,
+      "step": 69740
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021121455596650085,
+      "loss": 0.4787,
+      "step": 69750
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0002111295327976874,
+      "loss": 0.5202,
+      "step": 69760
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021104450962887387,
+      "loss": 0.4645,
+      "step": 69770
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021095948646006037,
+      "loss": 0.4909,
+      "step": 69780
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0002108744632912469,
+      "loss": 0.5268,
+      "step": 69790
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0002107894401224334,
+      "loss": 0.4608,
+      "step": 69800
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021070441695361986,
+      "loss": 0.5658,
+      "step": 69810
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00021061939378480636,
+      "loss": 0.473,
+      "step": 69820
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00021053437061599288,
+      "loss": 0.5412,
+      "step": 69830
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00021044934744717938,
+      "loss": 0.5268,
+      "step": 69840
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00021036432427836585,
+      "loss": 0.5407,
+      "step": 69850
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0002102793011095524,
+      "loss": 0.5147,
+      "step": 69860
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00021019427794073887,
+      "loss": 0.5633,
+      "step": 69870
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00021010925477192536,
+      "loss": 0.4917,
+      "step": 69880
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00021002423160311183,
+      "loss": 0.5617,
+      "step": 69890
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020993920843429838,
+      "loss": 0.4744,
+      "step": 69900
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020985418526548485,
+      "loss": 0.483,
+      "step": 69910
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020976916209667135,
+      "loss": 0.4686,
+      "step": 69920
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00020968413892785785,
+      "loss": 0.4946,
+      "step": 69930
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020959911575904437,
+      "loss": 0.4902,
+      "step": 69940
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020951409259023084,
+      "loss": 0.4347,
+      "step": 69950
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020942906942141734,
+      "loss": 0.4987,
+      "step": 69960
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020934404625260386,
+      "loss": 0.4446,
+      "step": 69970
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020925902308379036,
+      "loss": 0.4253,
+      "step": 69980
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020917399991497682,
+      "loss": 0.6149,
+      "step": 69990
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020908897674616332,
+      "loss": 0.5065,
+      "step": 70000
+    },
+    {
+      "epoch": 5.92,
+      "eval_loss": 0.5744438767433167,
+      "eval_runtime": 312.2896,
+      "eval_samples_per_second": 16.827,
+      "eval_steps_per_second": 2.104,
+      "step": 70000
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020900395357734985,
+      "loss": 0.4845,
+      "step": 70010
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020891893040853634,
+      "loss": 0.4642,
+      "step": 70020
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020883390723972284,
+      "loss": 0.508,
+      "step": 70030
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0002087488840709093,
+      "loss": 0.5006,
+      "step": 70040
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00020866386090209583,
+      "loss": 0.4869,
+      "step": 70050
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020857883773328233,
+      "loss": 0.47,
+      "step": 70060
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020849381456446883,
+      "loss": 0.5191,
+      "step": 70070
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020840879139565535,
+      "loss": 0.5093,
+      "step": 70080
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020832376822684182,
+      "loss": 0.4492,
+      "step": 70090
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020823874505802832,
+      "loss": 0.5498,
+      "step": 70100
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0002081537218892148,
+      "loss": 0.4914,
+      "step": 70110
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020806869872040134,
+      "loss": 0.5231,
+      "step": 70120
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020798367555158783,
+      "loss": 0.5235,
+      "step": 70130
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0002078986523827743,
+      "loss": 0.5289,
+      "step": 70140
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0002078136292139608,
+      "loss": 0.4353,
+      "step": 70150
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020772860604514732,
+      "loss": 0.4553,
+      "step": 70160
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00020764358287633382,
+      "loss": 0.5332,
+      "step": 70170
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0002075585597075203,
+      "loss": 0.4946,
+      "step": 70180
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0002074735365387068,
+      "loss": 0.5213,
+      "step": 70190
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0002073885133698933,
+      "loss": 0.471,
+      "step": 70200
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0002073034902010798,
+      "loss": 0.4343,
+      "step": 70210
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020721846703226627,
+      "loss": 0.5364,
+      "step": 70220
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020713344386345283,
+      "loss": 0.4867,
+      "step": 70230
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0002070484206946393,
+      "loss": 0.5148,
+      "step": 70240
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0002069633975258258,
+      "loss": 0.5459,
+      "step": 70250
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020687837435701232,
+      "loss": 0.5157,
+      "step": 70260
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0002067933511881988,
+      "loss": 0.4779,
+      "step": 70270
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020670832801938528,
+      "loss": 0.4781,
+      "step": 70280
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00020662330485057178,
+      "loss": 0.5101,
+      "step": 70290
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0002065382816817583,
+      "loss": 0.508,
+      "step": 70300
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0002064532585129448,
+      "loss": 0.5201,
+      "step": 70310
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020636823534413127,
+      "loss": 0.4778,
+      "step": 70320
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020628321217531776,
+      "loss": 0.541,
+      "step": 70330
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0002061981890065043,
+      "loss": 0.465,
+      "step": 70340
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020611316583769079,
+      "loss": 0.5548,
+      "step": 70350
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020602814266887725,
+      "loss": 0.5485,
+      "step": 70360
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0002059431195000638,
+      "loss": 0.5923,
+      "step": 70370
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020585809633125028,
+      "loss": 0.531,
+      "step": 70380
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020577307316243677,
+      "loss": 0.4814,
+      "step": 70390
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00020568804999362327,
+      "loss": 0.4971,
+      "step": 70400
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0002056030268248098,
+      "loss": 0.5701,
+      "step": 70410
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020551800365599626,
+      "loss": 0.5178,
+      "step": 70420
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020543298048718276,
+      "loss": 0.5013,
+      "step": 70430
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020534795731836925,
+      "loss": 0.5065,
+      "step": 70440
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020526293414955578,
+      "loss": 0.4866,
+      "step": 70450
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020517791098074225,
+      "loss": 0.5043,
+      "step": 70460
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020509288781192874,
+      "loss": 0.5149,
+      "step": 70470
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020500786464311527,
+      "loss": 0.4717,
+      "step": 70480
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020492284147430177,
+      "loss": 0.4443,
+      "step": 70490
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020483781830548826,
+      "loss": 0.4592,
+      "step": 70500
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020475279513667473,
+      "loss": 0.5031,
+      "step": 70510
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020466777196786125,
+      "loss": 0.5155,
+      "step": 70520
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00020458274879904775,
+      "loss": 0.5338,
+      "step": 70530
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020449772563023425,
+      "loss": 0.4694,
+      "step": 70540
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020441270246142077,
+      "loss": 0.5478,
+      "step": 70550
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020432767929260724,
+      "loss": 0.4871,
+      "step": 70560
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020424265612379374,
+      "loss": 0.4961,
+      "step": 70570
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020415763295498023,
+      "loss": 0.5029,
+      "step": 70580
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020407260978616676,
+      "loss": 0.5037,
+      "step": 70590
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020398758661735326,
+      "loss": 0.4853,
+      "step": 70600
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020390256344853972,
+      "loss": 0.5016,
+      "step": 70610
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020381754027972622,
+      "loss": 0.4924,
+      "step": 70620
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020373251711091275,
+      "loss": 0.5003,
+      "step": 70630
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00020364749394209924,
+      "loss": 0.4872,
+      "step": 70640
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0002035624707732857,
+      "loss": 0.462,
+      "step": 70650
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020347744760447226,
+      "loss": 0.5107,
+      "step": 70660
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020339242443565873,
+      "loss": 0.5126,
+      "step": 70670
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020330740126684523,
+      "loss": 0.5498,
+      "step": 70680
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0002032223780980317,
+      "loss": 0.5295,
+      "step": 70690
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020313735492921825,
+      "loss": 0.4707,
+      "step": 70700
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020305233176040472,
+      "loss": 0.5077,
+      "step": 70710
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020296730859159121,
+      "loss": 0.5058,
+      "step": 70720
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020288228542277768,
+      "loss": 0.4736,
+      "step": 70730
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00020279726225396424,
+      "loss": 0.5032,
+      "step": 70740
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0002027122390851507,
+      "loss": 0.5606,
+      "step": 70750
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0002026272159163372,
+      "loss": 0.5037,
+      "step": 70760
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020254219274752373,
+      "loss": 0.484,
+      "step": 70770
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020245716957871022,
+      "loss": 0.4963,
+      "step": 70780
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002023721464098967,
+      "loss": 0.4892,
+      "step": 70790
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002022871232410832,
+      "loss": 0.4741,
+      "step": 70800
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002022021000722697,
+      "loss": 0.4924,
+      "step": 70810
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002021170769034562,
+      "loss": 0.4757,
+      "step": 70820
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020203205373464268,
+      "loss": 0.4558,
+      "step": 70830
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020194703056582917,
+      "loss": 0.5354,
+      "step": 70840
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002018620073970157,
+      "loss": 0.5306,
+      "step": 70850
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002017769842282022,
+      "loss": 0.5435,
+      "step": 70860
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002016919610593887,
+      "loss": 0.532,
+      "step": 70870
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00020160693789057522,
+      "loss": 0.4661,
+      "step": 70880
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020152191472176168,
+      "loss": 0.5626,
+      "step": 70890
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020143689155294818,
+      "loss": 0.5477,
+      "step": 70900
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020135186838413468,
+      "loss": 0.5323,
+      "step": 70910
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0002012668452153212,
+      "loss": 0.4939,
+      "step": 70920
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020118182204650767,
+      "loss": 0.545,
+      "step": 70930
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020109679887769417,
+      "loss": 0.5072,
+      "step": 70940
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0002010117757088807,
+      "loss": 0.4682,
+      "step": 70950
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0002009267525400672,
+      "loss": 0.4942,
+      "step": 70960
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020084172937125368,
+      "loss": 0.5199,
+      "step": 70970
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020075670620244015,
+      "loss": 0.4294,
+      "step": 70980
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020067168303362668,
+      "loss": 0.4822,
+      "step": 70990
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00020058665986481317,
+      "loss": 0.4435,
+      "step": 71000
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00020050163669599967,
+      "loss": 0.4677,
+      "step": 71010
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00020041661352718614,
+      "loss": 0.4299,
+      "step": 71020
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0002003315903583727,
+      "loss": 0.5032,
+      "step": 71030
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00020024656718955916,
+      "loss": 0.4414,
+      "step": 71040
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00020016154402074566,
+      "loss": 0.4589,
+      "step": 71050
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00020007652085193218,
+      "loss": 0.548,
+      "step": 71060
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019999149768311868,
+      "loss": 0.459,
+      "step": 71070
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019990647451430515,
+      "loss": 0.5162,
+      "step": 71080
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019982145134549167,
+      "loss": 0.4743,
+      "step": 71090
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019973642817667814,
+      "loss": 0.4794,
+      "step": 71100
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019965140500786466,
+      "loss": 0.4639,
+      "step": 71110
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00019956638183905113,
+      "loss": 0.4588,
+      "step": 71120
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019948135867023766,
+      "loss": 0.445,
+      "step": 71130
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019939633550142415,
+      "loss": 0.4658,
+      "step": 71140
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019931131233261065,
+      "loss": 0.4653,
+      "step": 71150
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019922628916379715,
+      "loss": 0.5157,
+      "step": 71160
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019914126599498364,
+      "loss": 0.4072,
+      "step": 71170
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019905624282617014,
+      "loss": 0.4781,
+      "step": 71180
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019897121965735664,
+      "loss": 0.4775,
+      "step": 71190
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019888619648854313,
+      "loss": 0.4678,
+      "step": 71200
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019880117331972963,
+      "loss": 0.4347,
+      "step": 71210
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019871615015091613,
+      "loss": 0.461,
+      "step": 71220
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00019863112698210262,
+      "loss": 0.4325,
+      "step": 71230
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019854610381328912,
+      "loss": 0.5059,
+      "step": 71240
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019846108064447564,
+      "loss": 0.5163,
+      "step": 71250
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019837605747566211,
+      "loss": 0.4126,
+      "step": 71260
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019829103430684864,
+      "loss": 0.4553,
+      "step": 71270
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0001982060111380351,
+      "loss": 0.5197,
+      "step": 71280
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019812098796922163,
+      "loss": 0.4789,
+      "step": 71290
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0001980359648004081,
+      "loss": 0.4545,
+      "step": 71300
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019795094163159462,
+      "loss": 0.4613,
+      "step": 71310
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019786591846278112,
+      "loss": 0.4944,
+      "step": 71320
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019778089529396762,
+      "loss": 0.4285,
+      "step": 71330
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00019769587212515411,
+      "loss": 0.4466,
+      "step": 71340
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0001976108489563406,
+      "loss": 0.4411,
+      "step": 71350
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001975258257875271,
+      "loss": 0.4361,
+      "step": 71360
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001974408026187136,
+      "loss": 0.48,
+      "step": 71370
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001973557794499001,
+      "loss": 0.46,
+      "step": 71380
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001972707562810866,
+      "loss": 0.4546,
+      "step": 71390
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001971857331122731,
+      "loss": 0.4402,
+      "step": 71400
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001971007099434596,
+      "loss": 0.4408,
+      "step": 71410
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019701568677464611,
+      "loss": 0.4756,
+      "step": 71420
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00019693066360583258,
+      "loss": 0.4689,
+      "step": 71430
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001968456404370191,
+      "loss": 0.5055,
+      "step": 71440
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001967606172682056,
+      "loss": 0.3772,
+      "step": 71450
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001966755940993921,
+      "loss": 0.4202,
+      "step": 71460
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001965905709305786,
+      "loss": 0.4666,
+      "step": 71470
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001965055477617651,
+      "loss": 0.4632,
+      "step": 71480
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001964205245929516,
+      "loss": 0.4363,
+      "step": 71490
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001963355014241381,
+      "loss": 0.4404,
+      "step": 71500
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019625047825532458,
+      "loss": 0.4913,
+      "step": 71510
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019616545508651108,
+      "loss": 0.4876,
+      "step": 71520
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019608043191769758,
+      "loss": 0.4749,
+      "step": 71530
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001959954087488841,
+      "loss": 0.5124,
+      "step": 71540
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019591038558007057,
+      "loss": 0.4347,
+      "step": 71550
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001958253624112571,
+      "loss": 0.4773,
+      "step": 71560
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019574033924244356,
+      "loss": 0.4328,
+      "step": 71570
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001956553160736301,
+      "loss": 0.4881,
+      "step": 71580
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00019557029290481656,
+      "loss": 0.4414,
+      "step": 71590
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019548526973600308,
+      "loss": 0.4393,
+      "step": 71600
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019540024656718955,
+      "loss": 0.4586,
+      "step": 71610
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019531522339837607,
+      "loss": 0.4488,
+      "step": 71620
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019523020022956254,
+      "loss": 0.4475,
+      "step": 71630
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019514517706074907,
+      "loss": 0.4643,
+      "step": 71640
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019506015389193556,
+      "loss": 0.4347,
+      "step": 71650
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019497513072312206,
+      "loss": 0.4114,
+      "step": 71660
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019489010755430856,
+      "loss": 0.4629,
+      "step": 71670
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019480508438549505,
+      "loss": 0.4536,
+      "step": 71680
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019472006121668155,
+      "loss": 0.4788,
+      "step": 71690
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019463503804786805,
+      "loss": 0.4563,
+      "step": 71700
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00019455001487905454,
+      "loss": 0.4691,
+      "step": 71710
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019446499171024104,
+      "loss": 0.49,
+      "step": 71720
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019437996854142754,
+      "loss": 0.4785,
+      "step": 71730
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019429494537261406,
+      "loss": 0.4759,
+      "step": 71740
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019420992220380053,
+      "loss": 0.4886,
+      "step": 71750
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019412489903498705,
+      "loss": 0.4777,
+      "step": 71760
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019403987586617352,
+      "loss": 0.4903,
+      "step": 71770
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019395485269736005,
+      "loss": 0.4373,
+      "step": 71780
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019386982952854654,
+      "loss": 0.4948,
+      "step": 71790
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019378480635973304,
+      "loss": 0.4499,
+      "step": 71800
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019369978319091954,
+      "loss": 0.5085,
+      "step": 71810
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019361476002210603,
+      "loss": 0.4503,
+      "step": 71820
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00019352973685329253,
+      "loss": 0.5059,
+      "step": 71830
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019344471368447903,
+      "loss": 0.4644,
+      "step": 71840
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019335969051566552,
+      "loss": 0.5465,
+      "step": 71850
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019327466734685202,
+      "loss": 0.4976,
+      "step": 71860
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019318964417803854,
+      "loss": 0.4749,
+      "step": 71870
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019310462100922501,
+      "loss": 0.4441,
+      "step": 71880
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019301959784041154,
+      "loss": 0.4506,
+      "step": 71890
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.000192934574671598,
+      "loss": 0.4294,
+      "step": 71900
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019284955150278453,
+      "loss": 0.4169,
+      "step": 71910
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.000192764528333971,
+      "loss": 0.5055,
+      "step": 71920
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019267950516515752,
+      "loss": 0.4272,
+      "step": 71930
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00019259448199634402,
+      "loss": 0.5319,
+      "step": 71940
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019250945882753052,
+      "loss": 0.4908,
+      "step": 71950
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019242443565871701,
+      "loss": 0.4855,
+      "step": 71960
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001923394124899035,
+      "loss": 0.512,
+      "step": 71970
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019225438932109,
+      "loss": 0.4799,
+      "step": 71980
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001921693661522765,
+      "loss": 0.4283,
+      "step": 71990
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.000192084342983463,
+      "loss": 0.452,
+      "step": 72000
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001919993198146495,
+      "loss": 0.4453,
+      "step": 72010
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.000191914296645836,
+      "loss": 0.4773,
+      "step": 72020
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019182927347702252,
+      "loss": 0.5157,
+      "step": 72030
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.000191744250308209,
+      "loss": 0.4414,
+      "step": 72040
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001916592271393955,
+      "loss": 0.4475,
+      "step": 72050
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00019157420397058198,
+      "loss": 0.4493,
+      "step": 72060
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0001914891808017685,
+      "loss": 0.4306,
+      "step": 72070
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019140415763295497,
+      "loss": 0.5277,
+      "step": 72080
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0001913191344641415,
+      "loss": 0.5,
+      "step": 72090
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019123411129532797,
+      "loss": 0.5113,
+      "step": 72100
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0001911490881265145,
+      "loss": 0.4631,
+      "step": 72110
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019106406495770096,
+      "loss": 0.502,
+      "step": 72120
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019097904178888748,
+      "loss": 0.5191,
+      "step": 72130
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019089401862007398,
+      "loss": 0.4821,
+      "step": 72140
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019080899545126048,
+      "loss": 0.4516,
+      "step": 72150
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019072397228244697,
+      "loss": 0.4116,
+      "step": 72160
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019063894911363347,
+      "loss": 0.3963,
+      "step": 72170
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00019055392594481997,
+      "loss": 0.4652,
+      "step": 72180
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00019046890277600646,
+      "loss": 0.4907,
+      "step": 72190
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00019038387960719296,
+      "loss": 0.4758,
+      "step": 72200
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00019029885643837946,
+      "loss": 0.4912,
+      "step": 72210
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00019021383326956595,
+      "loss": 0.4875,
+      "step": 72220
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00019012881010075248,
+      "loss": 0.4355,
+      "step": 72230
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00019004378693193897,
+      "loss": 0.4916,
+      "step": 72240
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018995876376312547,
+      "loss": 0.4735,
+      "step": 72250
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018987374059431197,
+      "loss": 0.4681,
+      "step": 72260
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018978871742549846,
+      "loss": 0.434,
+      "step": 72270
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018970369425668496,
+      "loss": 0.4276,
+      "step": 72280
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018961867108787146,
+      "loss": 0.4984,
+      "step": 72290
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00018953364791905795,
+      "loss": 0.437,
+      "step": 72300
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018944862475024445,
+      "loss": 0.5256,
+      "step": 72310
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018936360158143095,
+      "loss": 0.485,
+      "step": 72320
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018927857841261744,
+      "loss": 0.4231,
+      "step": 72330
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018919355524380397,
+      "loss": 0.5016,
+      "step": 72340
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018910853207499044,
+      "loss": 0.4673,
+      "step": 72350
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018902350890617696,
+      "loss": 0.4869,
+      "step": 72360
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018893848573736343,
+      "loss": 0.4497,
+      "step": 72370
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018885346256854995,
+      "loss": 0.4409,
+      "step": 72380
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018876843939973642,
+      "loss": 0.4249,
+      "step": 72390
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018868341623092295,
+      "loss": 0.4979,
+      "step": 72400
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018859839306210942,
+      "loss": 0.4334,
+      "step": 72410
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00018851336989329594,
+      "loss": 0.4872,
+      "step": 72420
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018842834672448244,
+      "loss": 0.4638,
+      "step": 72430
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018834332355566893,
+      "loss": 0.4405,
+      "step": 72440
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018825830038685543,
+      "loss": 0.4698,
+      "step": 72450
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018817327721804193,
+      "loss": 0.4789,
+      "step": 72460
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018808825404922842,
+      "loss": 0.4376,
+      "step": 72470
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018800323088041492,
+      "loss": 0.5009,
+      "step": 72480
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018791820771160142,
+      "loss": 0.5063,
+      "step": 72490
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018783318454278791,
+      "loss": 0.4495,
+      "step": 72500
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001877481613739744,
+      "loss": 0.4515,
+      "step": 72510
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001876631382051609,
+      "loss": 0.487,
+      "step": 72520
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001875781150363474,
+      "loss": 0.4353,
+      "step": 72530
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00018749309186753393,
+      "loss": 0.4795,
+      "step": 72540
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001874080686987204,
+      "loss": 0.4486,
+      "step": 72550
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018732304552990692,
+      "loss": 0.471,
+      "step": 72560
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001872380223610934,
+      "loss": 0.4146,
+      "step": 72570
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018715299919227991,
+      "loss": 0.4425,
+      "step": 72580
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00018706797602346638,
+      "loss": 0.4814,
+      "step": 72590
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001869829528546529,
+      "loss": 0.4681,
+      "step": 72600
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001868979296858394,
+      "loss": 0.459,
+      "step": 72610
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001868129065170259,
+      "loss": 0.4246,
+      "step": 72620
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001867278833482124,
+      "loss": 0.51,
+      "step": 72630
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001866428601793989,
+      "loss": 0.5499,
+      "step": 72640
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001865578370105854,
+      "loss": 0.5055,
+      "step": 72650
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001864728138417719,
+      "loss": 0.5151,
+      "step": 72660
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018638779067295838,
+      "loss": 0.4329,
+      "step": 72670
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018630276750414488,
+      "loss": 0.483,
+      "step": 72680
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018621774433533138,
+      "loss": 0.4648,
+      "step": 72690
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018613272116651787,
+      "loss": 0.4455,
+      "step": 72700
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001860476979977044,
+      "loss": 0.4877,
+      "step": 72710
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001859626748288909,
+      "loss": 0.4841,
+      "step": 72720
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001858776516600774,
+      "loss": 0.4427,
+      "step": 72730
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001857926284912639,
+      "loss": 0.4711,
+      "step": 72740
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018570760532245038,
+      "loss": 0.5124,
+      "step": 72750
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018562258215363688,
+      "loss": 0.4765,
+      "step": 72760
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00018553755898482338,
+      "loss": 0.4807,
+      "step": 72770
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018545253581600987,
+      "loss": 0.4863,
+      "step": 72780
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018536751264719637,
+      "loss": 0.4518,
+      "step": 72790
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018528248947838287,
+      "loss": 0.4516,
+      "step": 72800
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018519746630956936,
+      "loss": 0.4728,
+      "step": 72810
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018511244314075586,
+      "loss": 0.4576,
+      "step": 72820
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018502741997194238,
+      "loss": 0.5583,
+      "step": 72830
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018494239680312885,
+      "loss": 0.468,
+      "step": 72840
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018485737363431538,
+      "loss": 0.4581,
+      "step": 72850
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018477235046550185,
+      "loss": 0.4771,
+      "step": 72860
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018468732729668837,
+      "loss": 0.4801,
+      "step": 72870
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018460230412787484,
+      "loss": 0.4234,
+      "step": 72880
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00018451728095906136,
+      "loss": 0.4963,
+      "step": 72890
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018443225779024783,
+      "loss": 0.5279,
+      "step": 72900
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018434723462143436,
+      "loss": 0.5232,
+      "step": 72910
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018426221145262085,
+      "loss": 0.4797,
+      "step": 72920
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018417718828380735,
+      "loss": 0.4971,
+      "step": 72930
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018409216511499385,
+      "loss": 0.4776,
+      "step": 72940
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018400714194618034,
+      "loss": 0.4866,
+      "step": 72950
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018392211877736684,
+      "loss": 0.5103,
+      "step": 72960
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018383709560855334,
+      "loss": 0.4329,
+      "step": 72970
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018375207243973983,
+      "loss": 0.4828,
+      "step": 72980
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018366704927092633,
+      "loss": 0.4612,
+      "step": 72990
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018358202610211283,
+      "loss": 0.5377,
+      "step": 73000
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00018349700293329932,
+      "loss": 0.492,
+      "step": 73010
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018341197976448582,
+      "loss": 0.5127,
+      "step": 73020
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018332695659567234,
+      "loss": 0.4835,
+      "step": 73030
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001832419334268588,
+      "loss": 0.4829,
+      "step": 73040
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018315691025804534,
+      "loss": 0.4493,
+      "step": 73050
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001830718870892318,
+      "loss": 0.4594,
+      "step": 73060
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018298686392041833,
+      "loss": 0.5125,
+      "step": 73070
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018290184075160483,
+      "loss": 0.5225,
+      "step": 73080
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018281681758279132,
+      "loss": 0.4446,
+      "step": 73090
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018273179441397782,
+      "loss": 0.4708,
+      "step": 73100
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00018264677124516432,
+      "loss": 0.4162,
+      "step": 73110
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001825617480763508,
+      "loss": 0.4422,
+      "step": 73120
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001824767249075373,
+      "loss": 0.4489,
+      "step": 73130
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001823917017387238,
+      "loss": 0.4724,
+      "step": 73140
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001823066785699103,
+      "loss": 0.4889,
+      "step": 73150
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018222165540109683,
+      "loss": 0.4534,
+      "step": 73160
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001821366322322833,
+      "loss": 0.4717,
+      "step": 73170
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018205160906346982,
+      "loss": 0.5375,
+      "step": 73180
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001819665858946563,
+      "loss": 0.4583,
+      "step": 73190
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001818815627258428,
+      "loss": 0.4592,
+      "step": 73200
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00018179653955702928,
+      "loss": 0.5191,
+      "step": 73210
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001817115163882158,
+      "loss": 0.4724,
+      "step": 73220
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001816264932194023,
+      "loss": 0.438,
+      "step": 73230
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001815414700505888,
+      "loss": 0.4208,
+      "step": 73240
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001814564468817753,
+      "loss": 0.5255,
+      "step": 73250
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001813714237129618,
+      "loss": 0.4891,
+      "step": 73260
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001812864005441483,
+      "loss": 0.4748,
+      "step": 73270
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001812013773753348,
+      "loss": 0.5694,
+      "step": 73280
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018111635420652128,
+      "loss": 0.463,
+      "step": 73290
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018103133103770778,
+      "loss": 0.4612,
+      "step": 73300
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018094630786889428,
+      "loss": 0.526,
+      "step": 73310
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001808612847000808,
+      "loss": 0.4664,
+      "step": 73320
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018077626153126727,
+      "loss": 0.461,
+      "step": 73330
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001806912383624538,
+      "loss": 0.4639,
+      "step": 73340
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00018060621519364026,
+      "loss": 0.502,
+      "step": 73350
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001805211920248268,
+      "loss": 0.4884,
+      "step": 73360
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00018043616885601326,
+      "loss": 0.4693,
+      "step": 73370
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00018035114568719978,
+      "loss": 0.5465,
+      "step": 73380
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00018026612251838625,
+      "loss": 0.4628,
+      "step": 73390
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00018018109934957277,
+      "loss": 0.5262,
+      "step": 73400
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00018009607618075924,
+      "loss": 0.468,
+      "step": 73410
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00018001105301194577,
+      "loss": 0.4917,
+      "step": 73420
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017992602984313226,
+      "loss": 0.4782,
+      "step": 73430
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017984100667431876,
+      "loss": 0.4516,
+      "step": 73440
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017975598350550526,
+      "loss": 0.4686,
+      "step": 73450
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017967096033669175,
+      "loss": 0.4549,
+      "step": 73460
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017958593716787825,
+      "loss": 0.4531,
+      "step": 73470
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017950091399906475,
+      "loss": 0.5053,
+      "step": 73480
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017941589083025124,
+      "loss": 0.4702,
+      "step": 73490
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017933086766143774,
+      "loss": 0.525,
+      "step": 73500
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017924584449262424,
+      "loss": 0.4634,
+      "step": 73510
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017916082132381076,
+      "loss": 0.4556,
+      "step": 73520
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017907579815499723,
+      "loss": 0.439,
+      "step": 73530
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017899077498618375,
+      "loss": 0.4195,
+      "step": 73540
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017890575181737025,
+      "loss": 0.4012,
+      "step": 73550
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017882072864855675,
+      "loss": 0.5078,
+      "step": 73560
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017873570547974324,
+      "loss": 0.4911,
+      "step": 73570
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017865068231092974,
+      "loss": 0.4994,
+      "step": 73580
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017856565914211624,
+      "loss": 0.4535,
+      "step": 73590
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00017848063597330273,
+      "loss": 0.4737,
+      "step": 73600
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017839561280448923,
+      "loss": 0.4825,
+      "step": 73610
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017831058963567573,
+      "loss": 0.4504,
+      "step": 73620
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017822556646686225,
+      "loss": 0.464,
+      "step": 73630
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017814054329804872,
+      "loss": 0.4487,
+      "step": 73640
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017805552012923524,
+      "loss": 0.4461,
+      "step": 73650
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001779704969604217,
+      "loss": 0.4727,
+      "step": 73660
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017788547379160824,
+      "loss": 0.4881,
+      "step": 73670
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001778004506227947,
+      "loss": 0.5,
+      "step": 73680
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017771542745398123,
+      "loss": 0.4519,
+      "step": 73690
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001776304042851677,
+      "loss": 0.4681,
+      "step": 73700
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017754538111635422,
+      "loss": 0.4091,
+      "step": 73710
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00017746035794754072,
+      "loss": 0.4404,
+      "step": 73720
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017737533477872722,
+      "loss": 0.4959,
+      "step": 73730
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001772903116099137,
+      "loss": 0.485,
+      "step": 73740
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001772052884411002,
+      "loss": 0.4608,
+      "step": 73750
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001771202652722867,
+      "loss": 0.4978,
+      "step": 73760
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001770352421034732,
+      "loss": 0.4611,
+      "step": 73770
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001769502189346597,
+      "loss": 0.5109,
+      "step": 73780
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001768651957658462,
+      "loss": 0.5048,
+      "step": 73790
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001767801725970327,
+      "loss": 0.5111,
+      "step": 73800
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017669514942821922,
+      "loss": 0.4513,
+      "step": 73810
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017661012625940569,
+      "loss": 0.4673,
+      "step": 73820
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001765251030905922,
+      "loss": 0.5018,
+      "step": 73830
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017644007992177868,
+      "loss": 0.5103,
+      "step": 73840
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0001763550567529652,
+      "loss": 0.4839,
+      "step": 73850
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017627003358415167,
+      "loss": 0.468,
+      "step": 73860
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0001761850104153382,
+      "loss": 0.465,
+      "step": 73870
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017609998724652467,
+      "loss": 0.5015,
+      "step": 73880
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0001760149640777112,
+      "loss": 0.4143,
+      "step": 73890
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017592994090889766,
+      "loss": 0.5203,
+      "step": 73900
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017584491774008418,
+      "loss": 0.4201,
+      "step": 73910
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017575989457127068,
+      "loss": 0.4795,
+      "step": 73920
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017567487140245718,
+      "loss": 0.473,
+      "step": 73930
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017558984823364367,
+      "loss": 0.4777,
+      "step": 73940
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00017550482506483017,
+      "loss": 0.4852,
+      "step": 73950
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017541980189601667,
+      "loss": 0.4673,
+      "step": 73960
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017533477872720316,
+      "loss": 0.4637,
+      "step": 73970
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017524975555838966,
+      "loss": 0.4463,
+      "step": 73980
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017516473238957616,
+      "loss": 0.4388,
+      "step": 73990
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017507970922076268,
+      "loss": 0.4627,
+      "step": 74000
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017499468605194918,
+      "loss": 0.4787,
+      "step": 74010
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017490966288313567,
+      "loss": 0.4844,
+      "step": 74020
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017482463971432217,
+      "loss": 0.4698,
+      "step": 74030
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017473961654550867,
+      "loss": 0.4764,
+      "step": 74040
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017465459337669516,
+      "loss": 0.4141,
+      "step": 74050
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017456957020788166,
+      "loss": 0.4584,
+      "step": 74060
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017448454703906816,
+      "loss": 0.4353,
+      "step": 74070
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017439952387025465,
+      "loss": 0.5812,
+      "step": 74080
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017431450070144115,
+      "loss": 0.4741,
+      "step": 74090
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017422947753262765,
+      "loss": 0.5044,
+      "step": 74100
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017414445436381414,
+      "loss": 0.5105,
+      "step": 74110
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017405943119500067,
+      "loss": 0.5055,
+      "step": 74120
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017397440802618714,
+      "loss": 0.4492,
+      "step": 74130
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017388938485737366,
+      "loss": 0.4208,
+      "step": 74140
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017380436168856013,
+      "loss": 0.4647,
+      "step": 74150
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017371933851974665,
+      "loss": 0.502,
+      "step": 74160
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017363431535093312,
+      "loss": 0.45,
+      "step": 74170
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017354929218211965,
+      "loss": 0.4864,
+      "step": 74180
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00017346426901330612,
+      "loss": 0.5,
+      "step": 74190
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017337924584449264,
+      "loss": 0.45,
+      "step": 74200
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017329422267567914,
+      "loss": 0.5035,
+      "step": 74210
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017320919950686563,
+      "loss": 0.4165,
+      "step": 74220
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017312417633805213,
+      "loss": 0.5067,
+      "step": 74230
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017303915316923863,
+      "loss": 0.5056,
+      "step": 74240
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017295413000042512,
+      "loss": 0.4426,
+      "step": 74250
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017286910683161162,
+      "loss": 0.4563,
+      "step": 74260
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017278408366279812,
+      "loss": 0.4361,
+      "step": 74270
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001726990604939846,
+      "loss": 0.4835,
+      "step": 74280
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001726140373251711,
+      "loss": 0.4479,
+      "step": 74290
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017252901415635763,
+      "loss": 0.4736,
+      "step": 74300
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001724439909875441,
+      "loss": 0.4705,
+      "step": 74310
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017235896781873063,
+      "loss": 0.4552,
+      "step": 74320
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001722739446499171,
+      "loss": 0.4659,
+      "step": 74330
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00017218892148110362,
+      "loss": 0.5159,
+      "step": 74340
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001721038983122901,
+      "loss": 0.4712,
+      "step": 74350
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001720188751434766,
+      "loss": 0.4925,
+      "step": 74360
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001719338519746631,
+      "loss": 0.4722,
+      "step": 74370
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001718488288058496,
+      "loss": 0.5105,
+      "step": 74380
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001717638056370361,
+      "loss": 0.4707,
+      "step": 74390
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001716787824682226,
+      "loss": 0.5058,
+      "step": 74400
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001715937592994091,
+      "loss": 0.4463,
+      "step": 74410
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001715087361305956,
+      "loss": 0.432,
+      "step": 74420
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001714237129617821,
+      "loss": 0.4901,
+      "step": 74430
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017133868979296859,
+      "loss": 0.4306,
+      "step": 74440
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017125366662415508,
+      "loss": 0.5208,
+      "step": 74450
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017116864345534158,
+      "loss": 0.4704,
+      "step": 74460
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.0001710836202865281,
+      "loss": 0.4973,
+      "step": 74470
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017099859711771457,
+      "loss": 0.4858,
+      "step": 74480
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.0001709135739489011,
+      "loss": 0.4836,
+      "step": 74490
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.0001708285507800876,
+      "loss": 0.492,
+      "step": 74500
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.0001707435276112741,
+      "loss": 0.5452,
+      "step": 74510
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017065850444246059,
+      "loss": 0.48,
+      "step": 74520
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017057348127364708,
+      "loss": 0.4639,
+      "step": 74530
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017048845810483358,
+      "loss": 0.4552,
+      "step": 74540
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00017040343493602008,
+      "loss": 0.4577,
+      "step": 74550
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00017031841176720657,
+      "loss": 0.4604,
+      "step": 74560
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00017023338859839307,
+      "loss": 0.4152,
+      "step": 74570
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00017014836542957957,
+      "loss": 0.4772,
+      "step": 74580
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00017006334226076606,
+      "loss": 0.477,
+      "step": 74590
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016997831909195256,
+      "loss": 0.5361,
+      "step": 74600
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016989329592313908,
+      "loss": 0.4901,
+      "step": 74610
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016980827275432555,
+      "loss": 0.4302,
+      "step": 74620
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016972324958551208,
+      "loss": 0.4362,
+      "step": 74630
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016963822641669855,
+      "loss": 0.5165,
+      "step": 74640
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016955320324788507,
+      "loss": 0.4704,
+      "step": 74650
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00016946818007907154,
+      "loss": 0.452,
+      "step": 74660
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016938315691025806,
+      "loss": 0.5068,
+      "step": 74670
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016929813374144453,
+      "loss": 0.5321,
+      "step": 74680
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016921311057263106,
+      "loss": 0.4985,
+      "step": 74690
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016912808740381755,
+      "loss": 0.4867,
+      "step": 74700
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016904306423500405,
+      "loss": 0.5018,
+      "step": 74710
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016895804106619055,
+      "loss": 0.5076,
+      "step": 74720
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016887301789737704,
+      "loss": 0.4926,
+      "step": 74730
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016878799472856354,
+      "loss": 0.4293,
+      "step": 74740
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016870297155975004,
+      "loss": 0.4654,
+      "step": 74750
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016861794839093653,
+      "loss": 0.5081,
+      "step": 74760
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016853292522212303,
+      "loss": 0.5364,
+      "step": 74770
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016844790205330953,
+      "loss": 0.5138,
+      "step": 74780
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016836287888449602,
+      "loss": 0.4814,
+      "step": 74790
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016827785571568252,
+      "loss": 0.5201,
+      "step": 74800
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016819283254686904,
+      "loss": 0.4652,
+      "step": 74810
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0001681078093780555,
+      "loss": 0.4886,
+      "step": 74820
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016802278620924204,
+      "loss": 0.4462,
+      "step": 74830
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016793776304042853,
+      "loss": 0.4641,
+      "step": 74840
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016785273987161503,
+      "loss": 0.4561,
+      "step": 74850
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016776771670280153,
+      "loss": 0.4545,
+      "step": 74860
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016768269353398802,
+      "loss": 0.4745,
+      "step": 74870
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016759767036517452,
+      "loss": 0.4553,
+      "step": 74880
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016751264719636102,
+      "loss": 0.4113,
+      "step": 74890
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0001674276240275475,
+      "loss": 0.4678,
+      "step": 74900
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.000167342600858734,
+      "loss": 0.4459,
+      "step": 74910
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016725757768992053,
+      "loss": 0.4772,
+      "step": 74920
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.000167172554521107,
+      "loss": 0.478,
+      "step": 74930
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016708753135229353,
+      "loss": 0.4992,
+      "step": 74940
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016700250818348,
+      "loss": 0.4441,
+      "step": 74950
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016691748501466652,
+      "loss": 0.4584,
+      "step": 74960
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.000166832461845853,
+      "loss": 0.5,
+      "step": 74970
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0001667474386770395,
+      "loss": 0.4654,
+      "step": 74980
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00016666241550822598,
+      "loss": 0.4409,
+      "step": 74990
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0001665773923394125,
+      "loss": 0.5219,
+      "step": 75000
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.000166492369170599,
+      "loss": 0.4433,
+      "step": 75010
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0001664073460017855,
+      "loss": 0.4927,
+      "step": 75020
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.000166322322832972,
+      "loss": 0.5489,
+      "step": 75030
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.0001662372996641585,
+      "loss": 0.4797,
+      "step": 75040
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.000166152276495345,
+      "loss": 0.4858,
+      "step": 75050
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016606725332653149,
+      "loss": 0.4709,
+      "step": 75060
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016598223015771798,
+      "loss": 0.4344,
+      "step": 75070
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016589720698890448,
+      "loss": 0.4628,
+      "step": 75080
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016581218382009098,
+      "loss": 0.425,
+      "step": 75090
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.0001657271606512775,
+      "loss": 0.49,
+      "step": 75100
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016564213748246397,
+      "loss": 0.4465,
+      "step": 75110
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.0001655571143136505,
+      "loss": 0.4974,
+      "step": 75120
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016547209114483696,
+      "loss": 0.4474,
+      "step": 75130
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016538706797602349,
+      "loss": 0.431,
+      "step": 75140
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016530204480720996,
+      "loss": 0.4594,
+      "step": 75150
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016521702163839648,
+      "loss": 0.529,
+      "step": 75160
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016513199846958295,
+      "loss": 0.4174,
+      "step": 75170
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016504697530076947,
+      "loss": 0.493,
+      "step": 75180
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016496195213195597,
+      "loss": 0.499,
+      "step": 75190
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016487692896314247,
+      "loss": 0.5104,
+      "step": 75200
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016479190579432896,
+      "loss": 0.4767,
+      "step": 75210
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016470688262551546,
+      "loss": 0.5021,
+      "step": 75220
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016462185945670196,
+      "loss": 0.5625,
+      "step": 75230
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016453683628788845,
+      "loss": 0.4306,
+      "step": 75240
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016445181311907495,
+      "loss": 0.4333,
+      "step": 75250
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016436678995026145,
+      "loss": 0.4702,
+      "step": 75260
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016428176678144794,
+      "loss": 0.4748,
+      "step": 75270
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016419674361263444,
+      "loss": 0.4862,
+      "step": 75280
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016411172044382096,
+      "loss": 0.4496,
+      "step": 75290
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016402669727500746,
+      "loss": 0.4844,
+      "step": 75300
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016394167410619396,
+      "loss": 0.4777,
+      "step": 75310
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016385665093738045,
+      "loss": 0.4301,
+      "step": 75320
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016377162776856695,
+      "loss": 0.415,
+      "step": 75330
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016368660459975345,
+      "loss": 0.4883,
+      "step": 75340
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016360158143093994,
+      "loss": 0.4706,
+      "step": 75350
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016351655826212644,
+      "loss": 0.4264,
+      "step": 75360
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016343153509331294,
+      "loss": 0.4193,
+      "step": 75370
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016334651192449943,
+      "loss": 0.456,
+      "step": 75380
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016326148875568596,
+      "loss": 0.4985,
+      "step": 75390
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016317646558687243,
+      "loss": 0.4813,
+      "step": 75400
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016309144241805895,
+      "loss": 0.4703,
+      "step": 75410
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016300641924924542,
+      "loss": 0.4114,
+      "step": 75420
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016292139608043194,
+      "loss": 0.4701,
+      "step": 75430
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0001628363729116184,
+      "loss": 0.4473,
+      "step": 75440
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016275134974280494,
+      "loss": 0.3923,
+      "step": 75450
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0001626663265739914,
+      "loss": 0.4928,
+      "step": 75460
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016258130340517793,
+      "loss": 0.4361,
+      "step": 75470
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0001624962802363644,
+      "loss": 0.4634,
+      "step": 75480
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016241125706755092,
+      "loss": 0.4193,
+      "step": 75490
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016232623389873742,
+      "loss": 0.4464,
+      "step": 75500
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016224121072992392,
+      "loss": 0.4846,
+      "step": 75510
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001621561875611104,
+      "loss": 0.4138,
+      "step": 75520
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001620711643922969,
+      "loss": 0.4466,
+      "step": 75530
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001619861412234834,
+      "loss": 0.527,
+      "step": 75540
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001619011180546699,
+      "loss": 0.479,
+      "step": 75550
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001618160948858564,
+      "loss": 0.4664,
+      "step": 75560
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001617310717170429,
+      "loss": 0.4603,
+      "step": 75570
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001616460485482294,
+      "loss": 0.5156,
+      "step": 75580
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016156102537941592,
+      "loss": 0.4873,
+      "step": 75590
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00016147600221060238,
+      "loss": 0.4892,
+      "step": 75600
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001613909790417889,
+      "loss": 0.4686,
+      "step": 75610
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016130595587297538,
+      "loss": 0.4961,
+      "step": 75620
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0001612209327041619,
+      "loss": 0.4584,
+      "step": 75630
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016113590953534837,
+      "loss": 0.5097,
+      "step": 75640
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0001610508863665349,
+      "loss": 0.4801,
+      "step": 75650
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016096586319772136,
+      "loss": 0.4488,
+      "step": 75660
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0001608808400289079,
+      "loss": 0.4965,
+      "step": 75670
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016079581686009439,
+      "loss": 0.4996,
+      "step": 75680
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016071079369128088,
+      "loss": 0.4808,
+      "step": 75690
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016062577052246738,
+      "loss": 0.4865,
+      "step": 75700
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016054074735365387,
+      "loss": 0.5004,
+      "step": 75710
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016045572418484037,
+      "loss": 0.4382,
+      "step": 75720
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016037070101602687,
+      "loss": 0.5307,
+      "step": 75730
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00016028567784721336,
+      "loss": 0.482,
+      "step": 75740
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00016020065467839986,
+      "loss": 0.5137,
+      "step": 75750
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00016011563150958639,
+      "loss": 0.4003,
+      "step": 75760
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00016003060834077285,
+      "loss": 0.5163,
+      "step": 75770
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015994558517195938,
+      "loss": 0.4738,
+      "step": 75780
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015986056200314588,
+      "loss": 0.4781,
+      "step": 75790
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015977553883433237,
+      "loss": 0.4565,
+      "step": 75800
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015969051566551887,
+      "loss": 0.4811,
+      "step": 75810
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015960549249670537,
+      "loss": 0.4527,
+      "step": 75820
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015952046932789186,
+      "loss": 0.5038,
+      "step": 75830
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015943544615907836,
+      "loss": 0.4885,
+      "step": 75840
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00015935042299026485,
+      "loss": 0.4995,
+      "step": 75850
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015926539982145135,
+      "loss": 0.4467,
+      "step": 75860
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015918037665263785,
+      "loss": 0.5182,
+      "step": 75870
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015909535348382434,
+      "loss": 0.4571,
+      "step": 75880
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015901033031501084,
+      "loss": 0.4673,
+      "step": 75890
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015892530714619737,
+      "loss": 0.4516,
+      "step": 75900
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015884028397738383,
+      "loss": 0.4216,
+      "step": 75910
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015875526080857036,
+      "loss": 0.5162,
+      "step": 75920
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015867023763975683,
+      "loss": 0.4724,
+      "step": 75930
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015858521447094335,
+      "loss": 0.4805,
+      "step": 75940
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015850019130212982,
+      "loss": 0.4323,
+      "step": 75950
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00015841516813331635,
+      "loss": 0.4209,
+      "step": 75960
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015833014496450281,
+      "loss": 0.4836,
+      "step": 75970
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015824512179568934,
+      "loss": 0.4747,
+      "step": 75980
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015816009862687583,
+      "loss": 0.4338,
+      "step": 75990
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015807507545806233,
+      "loss": 0.4773,
+      "step": 76000
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015799005228924883,
+      "loss": 0.4602,
+      "step": 76010
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015790502912043532,
+      "loss": 0.411,
+      "step": 76020
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015782000595162182,
+      "loss": 0.5058,
+      "step": 76030
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015773498278280832,
+      "loss": 0.4715,
+      "step": 76040
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015764995961399481,
+      "loss": 0.4574,
+      "step": 76050
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0001575649364451813,
+      "loss": 0.4556,
+      "step": 76060
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0001574799132763678,
+      "loss": 0.4538,
+      "step": 76070
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00015739489010755433,
+      "loss": 0.4746,
+      "step": 76080
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001573098669387408,
+      "loss": 0.4457,
+      "step": 76090
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015722484376992732,
+      "loss": 0.5026,
+      "step": 76100
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001571398206011138,
+      "loss": 0.4888,
+      "step": 76110
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015705479743230032,
+      "loss": 0.4745,
+      "step": 76120
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00015696977426348681,
+      "loss": 0.4858,
+      "step": 76130
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001568847510946733,
+      "loss": 0.4697,
+      "step": 76140
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001567997279258598,
+      "loss": 0.5139,
+      "step": 76150
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001567147047570463,
+      "loss": 0.4831,
+      "step": 76160
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001566296815882328,
+      "loss": 0.4615,
+      "step": 76170
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001565446584194193,
+      "loss": 0.5446,
+      "step": 76180
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001564596352506058,
+      "loss": 0.4846,
+      "step": 76190
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001563746120817923,
+      "loss": 0.4794,
+      "step": 76200
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015628958891297882,
+      "loss": 0.4291,
+      "step": 76210
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015620456574416528,
+      "loss": 0.4805,
+      "step": 76220
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0001561195425753518,
+      "loss": 0.4165,
+      "step": 76230
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015603451940653828,
+      "loss": 0.4753,
+      "step": 76240
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0001559494962377248,
+      "loss": 0.5178,
+      "step": 76250
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015586447306891127,
+      "loss": 0.527,
+      "step": 76260
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0001557794499000978,
+      "loss": 0.5085,
+      "step": 76270
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0001556944267312843,
+      "loss": 0.4936,
+      "step": 76280
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0001556094035624708,
+      "loss": 0.448,
+      "step": 76290
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015552438039365728,
+      "loss": 0.482,
+      "step": 76300
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015543935722484378,
+      "loss": 0.476,
+      "step": 76310
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00015535433405603028,
+      "loss": 0.4163,
+      "step": 76320
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015526931088721677,
+      "loss": 0.5037,
+      "step": 76330
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015518428771840327,
+      "loss": 0.4241,
+      "step": 76340
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015509926454958977,
+      "loss": 0.4746,
+      "step": 76350
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015501424138077626,
+      "loss": 0.45,
+      "step": 76360
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015492921821196276,
+      "loss": 0.4526,
+      "step": 76370
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015484419504314926,
+      "loss": 0.4208,
+      "step": 76380
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015475917187433578,
+      "loss": 0.4523,
+      "step": 76390
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015467414870552225,
+      "loss": 0.4685,
+      "step": 76400
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015458912553670877,
+      "loss": 0.5308,
+      "step": 76410
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015450410236789524,
+      "loss": 0.4915,
+      "step": 76420
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015441907919908177,
+      "loss": 0.432,
+      "step": 76430
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00015433405603026824,
+      "loss": 0.4573,
+      "step": 76440
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015424903286145476,
+      "loss": 0.4672,
+      "step": 76450
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015416400969264123,
+      "loss": 0.4977,
+      "step": 76460
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015407898652382775,
+      "loss": 0.4549,
+      "step": 76470
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015399396335501425,
+      "loss": 0.4808,
+      "step": 76480
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015390894018620075,
+      "loss": 0.5046,
+      "step": 76490
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015382391701738724,
+      "loss": 0.4401,
+      "step": 76500
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015373889384857374,
+      "loss": 0.484,
+      "step": 76510
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015365387067976024,
+      "loss": 0.4848,
+      "step": 76520
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015356884751094673,
+      "loss": 0.4679,
+      "step": 76530
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015348382434213323,
+      "loss": 0.4457,
+      "step": 76540
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015339880117331973,
+      "loss": 0.426,
+      "step": 76550
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00015331377800450622,
+      "loss": 0.4407,
+      "step": 76560
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015322875483569272,
+      "loss": 0.4446,
+      "step": 76570
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015314373166687922,
+      "loss": 0.4692,
+      "step": 76580
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015305870849806574,
+      "loss": 0.4535,
+      "step": 76590
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015297368532925224,
+      "loss": 0.4706,
+      "step": 76600
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015288866216043873,
+      "loss": 0.54,
+      "step": 76610
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015280363899162523,
+      "loss": 0.4764,
+      "step": 76620
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015271861582281173,
+      "loss": 0.4554,
+      "step": 76630
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015263359265399822,
+      "loss": 0.485,
+      "step": 76640
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015254856948518472,
+      "loss": 0.4437,
+      "step": 76650
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015246354631637122,
+      "loss": 0.4622,
+      "step": 76660
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00015237852314755771,
+      "loss": 0.4742,
+      "step": 76670
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015229349997874424,
+      "loss": 0.4742,
+      "step": 76680
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001522084768099307,
+      "loss": 0.4484,
+      "step": 76690
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015212345364111723,
+      "loss": 0.5452,
+      "step": 76700
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001520384304723037,
+      "loss": 0.5014,
+      "step": 76710
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015195340730349022,
+      "loss": 0.4461,
+      "step": 76720
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001518683841346767,
+      "loss": 0.4483,
+      "step": 76730
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00015178336096586322,
+      "loss": 0.498,
+      "step": 76740
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001516983377970497,
+      "loss": 0.4573,
+      "step": 76750
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001516133146282362,
+      "loss": 0.4762,
+      "step": 76760
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001515282914594227,
+      "loss": 0.4646,
+      "step": 76770
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001514432682906092,
+      "loss": 0.4592,
+      "step": 76780
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001513582451217957,
+      "loss": 0.5003,
+      "step": 76790
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001512732219529822,
+      "loss": 0.4743,
+      "step": 76800
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001511881987841687,
+      "loss": 0.4252,
+      "step": 76810
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001511031756153552,
+      "loss": 0.4768,
+      "step": 76820
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001510181524465417,
+      "loss": 0.4973,
+      "step": 76830
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015093312927772818,
+      "loss": 0.4593,
+      "step": 76840
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015084810610891468,
+      "loss": 0.4835,
+      "step": 76850
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015076308294010118,
+      "loss": 0.4483,
+      "step": 76860
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015067805977128767,
+      "loss": 0.4974,
+      "step": 76870
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001505930366024742,
+      "loss": 0.469,
+      "step": 76880
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015050801343366067,
+      "loss": 0.4787,
+      "step": 76890
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001504229902648472,
+      "loss": 0.4907,
+      "step": 76900
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00015033796709603366,
+      "loss": 0.4545,
+      "step": 76910
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00015025294392722018,
+      "loss": 0.502,
+      "step": 76920
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00015016792075840665,
+      "loss": 0.4744,
+      "step": 76930
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00015008289758959318,
+      "loss": 0.4634,
+      "step": 76940
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014999787442077965,
+      "loss": 0.4604,
+      "step": 76950
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014991285125196617,
+      "loss": 0.4705,
+      "step": 76960
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014982782808315267,
+      "loss": 0.4192,
+      "step": 76970
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014974280491433916,
+      "loss": 0.458,
+      "step": 76980
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014965778174552566,
+      "loss": 0.4533,
+      "step": 76990
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014957275857671216,
+      "loss": 0.5566,
+      "step": 77000
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014948773540789865,
+      "loss": 0.522,
+      "step": 77010
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014940271223908515,
+      "loss": 0.4612,
+      "step": 77020
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00014931768907027165,
+      "loss": 0.457,
+      "step": 77030
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014923266590145814,
+      "loss": 0.4517,
+      "step": 77040
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014914764273264467,
+      "loss": 0.471,
+      "step": 77050
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014906261956383114,
+      "loss": 0.4822,
+      "step": 77060
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014897759639501766,
+      "loss": 0.4693,
+      "step": 77070
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014889257322620416,
+      "loss": 0.4277,
+      "step": 77080
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014880755005739065,
+      "loss": 0.464,
+      "step": 77090
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014872252688857715,
+      "loss": 0.4533,
+      "step": 77100
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014863750371976365,
+      "loss": 0.4557,
+      "step": 77110
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014855248055095014,
+      "loss": 0.4914,
+      "step": 77120
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014846745738213664,
+      "loss": 0.4543,
+      "step": 77130
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014838243421332314,
+      "loss": 0.4856,
+      "step": 77140
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00014829741104450963,
+      "loss": 0.4833,
+      "step": 77150
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014821238787569613,
+      "loss": 0.5162,
+      "step": 77160
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014812736470688265,
+      "loss": 0.4502,
+      "step": 77170
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014804234153806912,
+      "loss": 0.4444,
+      "step": 77180
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014795731836925565,
+      "loss": 0.4963,
+      "step": 77190
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014787229520044212,
+      "loss": 0.4836,
+      "step": 77200
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014778727203162864,
+      "loss": 0.4774,
+      "step": 77210
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0001477022488628151,
+      "loss": 0.4813,
+      "step": 77220
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014761722569400163,
+      "loss": 0.4383,
+      "step": 77230
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0001475322025251881,
+      "loss": 0.4508,
+      "step": 77240
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00014744717935637463,
+      "loss": 0.4425,
+      "step": 77250
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0001473621561875611,
+      "loss": 0.4676,
+      "step": 77260
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014727713301874762,
+      "loss": 0.4557,
+      "step": 77270
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014719210984993412,
+      "loss": 0.4854,
+      "step": 77280
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014710708668112061,
+      "loss": 0.4024,
+      "step": 77290
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001470220635123071,
+      "loss": 0.5487,
+      "step": 77300
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001469370403434936,
+      "loss": 0.4675,
+      "step": 77310
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001468520171746801,
+      "loss": 0.438,
+      "step": 77320
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001467669940058666,
+      "loss": 0.4132,
+      "step": 77330
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001466819708370531,
+      "loss": 0.5235,
+      "step": 77340
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001465969476682396,
+      "loss": 0.4781,
+      "step": 77350
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001465119244994261,
+      "loss": 0.4759,
+      "step": 77360
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014642690133061261,
+      "loss": 0.4916,
+      "step": 77370
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00014634187816179908,
+      "loss": 0.5263,
+      "step": 77380
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0001462568549929856,
+      "loss": 0.4412,
+      "step": 77390
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014617183182417208,
+      "loss": 0.5014,
+      "step": 77400
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0001460868086553586,
+      "loss": 0.557,
+      "step": 77410
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0001460017854865451,
+      "loss": 0.5056,
+      "step": 77420
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0001459167623177316,
+      "loss": 0.485,
+      "step": 77430
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0001458317391489181,
+      "loss": 0.4721,
+      "step": 77440
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0001457467159801046,
+      "loss": 0.489,
+      "step": 77450
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014566169281129108,
+      "loss": 0.4855,
+      "step": 77460
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014557666964247758,
+      "loss": 0.4805,
+      "step": 77470
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014549164647366408,
+      "loss": 0.4796,
+      "step": 77480
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014540662330485057,
+      "loss": 0.4845,
+      "step": 77490
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00014532160013603707,
+      "loss": 0.5152,
+      "step": 77500
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014523657696722357,
+      "loss": 0.424,
+      "step": 77510
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0001451515537984101,
+      "loss": 0.4992,
+      "step": 77520
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014506653062959656,
+      "loss": 0.4447,
+      "step": 77530
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014498150746078308,
+      "loss": 0.4454,
+      "step": 77540
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014489648429196955,
+      "loss": 0.4443,
+      "step": 77550
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014481146112315608,
+      "loss": 0.4344,
+      "step": 77560
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014472643795434257,
+      "loss": 0.4614,
+      "step": 77570
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014464141478552907,
+      "loss": 0.473,
+      "step": 77580
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014455639161671557,
+      "loss": 0.4527,
+      "step": 77590
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014447136844790206,
+      "loss": 0.4754,
+      "step": 77600
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014438634527908856,
+      "loss": 0.5314,
+      "step": 77610
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00014430132211027506,
+      "loss": 0.5305,
+      "step": 77620
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014421629894146155,
+      "loss": 0.4943,
+      "step": 77630
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014413127577264805,
+      "loss": 0.4627,
+      "step": 77640
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014404625260383455,
+      "loss": 0.4423,
+      "step": 77650
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014396122943502107,
+      "loss": 0.433,
+      "step": 77660
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014387620626620754,
+      "loss": 0.4836,
+      "step": 77670
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014379118309739406,
+      "loss": 0.4669,
+      "step": 77680
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014370615992858053,
+      "loss": 0.438,
+      "step": 77690
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014362113675976706,
+      "loss": 0.4717,
+      "step": 77700
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014353611359095353,
+      "loss": 0.5034,
+      "step": 77710
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014345109042214005,
+      "loss": 0.4633,
+      "step": 77720
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014336606725332652,
+      "loss": 0.4845,
+      "step": 77730
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00014328104408451304,
+      "loss": 0.4928,
+      "step": 77740
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0001431960209156995,
+      "loss": 0.4744,
+      "step": 77750
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014311099774688604,
+      "loss": 0.4695,
+      "step": 77760
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014302597457807253,
+      "loss": 0.5374,
+      "step": 77770
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014294095140925903,
+      "loss": 0.5099,
+      "step": 77780
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014285592824044553,
+      "loss": 0.4764,
+      "step": 77790
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014277090507163202,
+      "loss": 0.4451,
+      "step": 77800
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014268588190281852,
+      "loss": 0.4544,
+      "step": 77810
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014260085873400502,
+      "loss": 0.4252,
+      "step": 77820
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014251583556519151,
+      "loss": 0.4783,
+      "step": 77830
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.000142430812396378,
+      "loss": 0.4585,
+      "step": 77840
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0001423457892275645,
+      "loss": 0.4837,
+      "step": 77850
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00014226076605875103,
+      "loss": 0.4603,
+      "step": 77860
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001421757428899375,
+      "loss": 0.4521,
+      "step": 77870
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014209071972112402,
+      "loss": 0.454,
+      "step": 77880
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014200569655231052,
+      "loss": 0.4778,
+      "step": 77890
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014192067338349702,
+      "loss": 0.4704,
+      "step": 77900
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014183565021468351,
+      "loss": 0.4732,
+      "step": 77910
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014175062704587,
+      "loss": 0.4819,
+      "step": 77920
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001416656038770565,
+      "loss": 0.4636,
+      "step": 77930
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.000141580580708243,
+      "loss": 0.4627,
+      "step": 77940
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001414955575394295,
+      "loss": 0.4661,
+      "step": 77950
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.000141410534370616,
+      "loss": 0.4324,
+      "step": 77960
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00014132551120180252,
+      "loss": 0.4881,
+      "step": 77970
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.000141240488032989,
+      "loss": 0.4232,
+      "step": 77980
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014115546486417551,
+      "loss": 0.4932,
+      "step": 77990
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014107044169536198,
+      "loss": 0.4501,
+      "step": 78000
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001409854185265485,
+      "loss": 0.4714,
+      "step": 78010
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014090039535773498,
+      "loss": 0.4409,
+      "step": 78020
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001408153721889215,
+      "loss": 0.5277,
+      "step": 78030
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014073034902010797,
+      "loss": 0.4525,
+      "step": 78040
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001406453258512945,
+      "loss": 0.4834,
+      "step": 78050
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.000140560302682481,
+      "loss": 0.4187,
+      "step": 78060
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001404752795136675,
+      "loss": 0.4624,
+      "step": 78070
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014039025634485398,
+      "loss": 0.4622,
+      "step": 78080
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00014030523317604048,
+      "loss": 0.421,
+      "step": 78090
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00014022021000722698,
+      "loss": 0.5195,
+      "step": 78100
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00014013518683841347,
+      "loss": 0.4555,
+      "step": 78110
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00014005016366959997,
+      "loss": 0.4422,
+      "step": 78120
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013996514050078647,
+      "loss": 0.462,
+      "step": 78130
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013988011733197296,
+      "loss": 0.4348,
+      "step": 78140
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013979509416315946,
+      "loss": 0.4874,
+      "step": 78150
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013971007099434596,
+      "loss": 0.5237,
+      "step": 78160
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013962504782553248,
+      "loss": 0.4519,
+      "step": 78170
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013954002465671895,
+      "loss": 0.4971,
+      "step": 78180
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013945500148790547,
+      "loss": 0.4661,
+      "step": 78190
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013936997831909194,
+      "loss": 0.4556,
+      "step": 78200
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013928495515027847,
+      "loss": 0.4464,
+      "step": 78210
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013919993198146494,
+      "loss": 0.4227,
+      "step": 78220
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013911490881265146,
+      "loss": 0.4518,
+      "step": 78230
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013902988564383793,
+      "loss": 0.4911,
+      "step": 78240
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013894486247502445,
+      "loss": 0.4691,
+      "step": 78250
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013885983930621095,
+      "loss": 0.4748,
+      "step": 78260
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013877481613739745,
+      "loss": 0.4694,
+      "step": 78270
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013868979296858394,
+      "loss": 0.4608,
+      "step": 78280
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013860476979977044,
+      "loss": 0.4716,
+      "step": 78290
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013851974663095694,
+      "loss": 0.4805,
+      "step": 78300
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013843472346214343,
+      "loss": 0.4453,
+      "step": 78310
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013834970029332993,
+      "loss": 0.4526,
+      "step": 78320
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013826467712451643,
+      "loss": 0.4224,
+      "step": 78330
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013817965395570295,
+      "loss": 0.4877,
+      "step": 78340
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013809463078688942,
+      "loss": 0.444,
+      "step": 78350
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013800960761807594,
+      "loss": 0.5149,
+      "step": 78360
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013792458444926244,
+      "loss": 0.4704,
+      "step": 78370
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013783956128044894,
+      "loss": 0.5403,
+      "step": 78380
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013775453811163543,
+      "loss": 0.4675,
+      "step": 78390
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013766951494282193,
+      "loss": 0.4685,
+      "step": 78400
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013758449177400843,
+      "loss": 0.4713,
+      "step": 78410
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013749946860519492,
+      "loss": 0.4553,
+      "step": 78420
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013741444543638142,
+      "loss": 0.4663,
+      "step": 78430
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013732942226756792,
+      "loss": 0.4465,
+      "step": 78440
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0001372443990987544,
+      "loss": 0.4471,
+      "step": 78450
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013715937592994094,
+      "loss": 0.4581,
+      "step": 78460
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001370743527611274,
+      "loss": 0.5034,
+      "step": 78470
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013698932959231393,
+      "loss": 0.5467,
+      "step": 78480
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001369043064235004,
+      "loss": 0.4381,
+      "step": 78490
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013681928325468692,
+      "loss": 0.48,
+      "step": 78500
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001367342600858734,
+      "loss": 0.4206,
+      "step": 78510
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013664923691705992,
+      "loss": 0.4638,
+      "step": 78520
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013656421374824639,
+      "loss": 0.5043,
+      "step": 78530
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001364791905794329,
+      "loss": 0.4282,
+      "step": 78540
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001363941674106194,
+      "loss": 0.511,
+      "step": 78550
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001363091442418059,
+      "loss": 0.478,
+      "step": 78560
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001362241210729924,
+      "loss": 0.439,
+      "step": 78570
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001361390979041789,
+      "loss": 0.5128,
+      "step": 78580
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001360540747353654,
+      "loss": 0.4364,
+      "step": 78590
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001359690515665519,
+      "loss": 0.5293,
+      "step": 78600
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001358840283977384,
+      "loss": 0.4718,
+      "step": 78610
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013579900522892488,
+      "loss": 0.5381,
+      "step": 78620
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013571398206011138,
+      "loss": 0.4953,
+      "step": 78630
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013562895889129788,
+      "loss": 0.5035,
+      "step": 78640
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013554393572248437,
+      "loss": 0.4836,
+      "step": 78650
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001354589125536709,
+      "loss": 0.4657,
+      "step": 78660
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013537388938485737,
+      "loss": 0.4648,
+      "step": 78670
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001352888662160439,
+      "loss": 0.4738,
+      "step": 78680
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013520384304723036,
+      "loss": 0.4213,
+      "step": 78690
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013511881987841688,
+      "loss": 0.4683,
+      "step": 78700
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013503379670960335,
+      "loss": 0.5144,
+      "step": 78710
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013494877354078988,
+      "loss": 0.4415,
+      "step": 78720
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013486375037197637,
+      "loss": 0.4433,
+      "step": 78730
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013477872720316287,
+      "loss": 0.4883,
+      "step": 78740
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013469370403434937,
+      "loss": 0.4316,
+      "step": 78750
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013460868086553586,
+      "loss": 0.4734,
+      "step": 78760
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013452365769672236,
+      "loss": 0.481,
+      "step": 78770
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013443863452790886,
+      "loss": 0.4771,
+      "step": 78780
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013435361135909535,
+      "loss": 0.4616,
+      "step": 78790
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00013426858819028185,
+      "loss": 0.5292,
+      "step": 78800
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013418356502146837,
+      "loss": 0.4869,
+      "step": 78810
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013409854185265484,
+      "loss": 0.494,
+      "step": 78820
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013401351868384137,
+      "loss": 0.4416,
+      "step": 78830
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013392849551502784,
+      "loss": 0.458,
+      "step": 78840
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013384347234621436,
+      "loss": 0.496,
+      "step": 78850
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013375844917740086,
+      "loss": 0.4605,
+      "step": 78860
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013367342600858735,
+      "loss": 0.4813,
+      "step": 78870
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013358840283977385,
+      "loss": 0.4857,
+      "step": 78880
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013350337967096035,
+      "loss": 0.4455,
+      "step": 78890
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013341835650214684,
+      "loss": 0.5053,
+      "step": 78900
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 0.4808,
+      "step": 78910
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00013324831016451984,
+      "loss": 0.4871,
+      "step": 78920
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013316328699570633,
+      "loss": 0.4675,
+      "step": 78930
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013307826382689283,
+      "loss": 0.4311,
+      "step": 78940
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013299324065807935,
+      "loss": 0.4995,
+      "step": 78950
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013290821748926582,
+      "loss": 0.4977,
+      "step": 78960
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013282319432045235,
+      "loss": 0.4505,
+      "step": 78970
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013273817115163882,
+      "loss": 0.4916,
+      "step": 78980
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013265314798282534,
+      "loss": 0.5253,
+      "step": 78990
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0001325681248140118,
+      "loss": 0.4459,
+      "step": 79000
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013248310164519833,
+      "loss": 0.4844,
+      "step": 79010
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0001323980784763848,
+      "loss": 0.4062,
+      "step": 79020
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013231305530757133,
+      "loss": 0.4346,
+      "step": 79030
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0001322280321387578,
+      "loss": 0.4568,
+      "step": 79040
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013214300896994432,
+      "loss": 0.4617,
+      "step": 79050
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013205798580113082,
+      "loss": 0.4774,
+      "step": 79060
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001319729626323173,
+      "loss": 0.4509,
+      "step": 79070
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001318879394635038,
+      "loss": 0.4469,
+      "step": 79080
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001318029162946903,
+      "loss": 0.4753,
+      "step": 79090
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001317178931258768,
+      "loss": 0.4725,
+      "step": 79100
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001316328699570633,
+      "loss": 0.4511,
+      "step": 79110
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001315478467882498,
+      "loss": 0.4881,
+      "step": 79120
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001314628236194363,
+      "loss": 0.4506,
+      "step": 79130
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001313778004506228,
+      "loss": 0.4922,
+      "step": 79140
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001312927772818093,
+      "loss": 0.5037,
+      "step": 79150
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00013120775411299578,
+      "loss": 0.4697,
+      "step": 79160
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001311227309441823,
+      "loss": 0.4458,
+      "step": 79170
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001310377077753688,
+      "loss": 0.4242,
+      "step": 79180
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001309526846065553,
+      "loss": 0.5274,
+      "step": 79190
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001308676614377418,
+      "loss": 0.5007,
+      "step": 79200
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001307826382689283,
+      "loss": 0.4881,
+      "step": 79210
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001306976151001148,
+      "loss": 0.5019,
+      "step": 79220
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013061259193130129,
+      "loss": 0.4745,
+      "step": 79230
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013052756876248778,
+      "loss": 0.4816,
+      "step": 79240
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013044254559367428,
+      "loss": 0.4621,
+      "step": 79250
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013035752242486078,
+      "loss": 0.4551,
+      "step": 79260
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013027249925604727,
+      "loss": 0.4892,
+      "step": 79270
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0001301874760872338,
+      "loss": 0.4834,
+      "step": 79280
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00013010245291842027,
+      "loss": 0.5052,
+      "step": 79290
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0001300174297496068,
+      "loss": 0.4984,
+      "step": 79300
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012993240658079326,
+      "loss": 0.4805,
+      "step": 79310
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012984738341197978,
+      "loss": 0.4872,
+      "step": 79320
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012976236024316625,
+      "loss": 0.4722,
+      "step": 79330
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012967733707435278,
+      "loss": 0.4246,
+      "step": 79340
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012959231390553927,
+      "loss": 0.4682,
+      "step": 79350
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012950729073672577,
+      "loss": 0.4451,
+      "step": 79360
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012942226756791227,
+      "loss": 0.4893,
+      "step": 79370
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012933724439909876,
+      "loss": 0.4971,
+      "step": 79380
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00012925222123028526,
+      "loss": 0.4349,
+      "step": 79390
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012916719806147176,
+      "loss": 0.4688,
+      "step": 79400
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012908217489265825,
+      "loss": 0.4833,
+      "step": 79410
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012899715172384475,
+      "loss": 0.4482,
+      "step": 79420
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012891212855503125,
+      "loss": 0.4963,
+      "step": 79430
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012882710538621777,
+      "loss": 0.474,
+      "step": 79440
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012874208221740424,
+      "loss": 0.4805,
+      "step": 79450
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012865705904859076,
+      "loss": 0.4497,
+      "step": 79460
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012857203587977723,
+      "loss": 0.5073,
+      "step": 79470
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012848701271096376,
+      "loss": 0.4827,
+      "step": 79480
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012840198954215023,
+      "loss": 0.4565,
+      "step": 79490
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012831696637333675,
+      "loss": 0.4659,
+      "step": 79500
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00012823194320452322,
+      "loss": 0.4384,
+      "step": 79510
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012814692003570974,
+      "loss": 0.4943,
+      "step": 79520
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001280618968668962,
+      "loss": 0.5008,
+      "step": 79530
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012797687369808274,
+      "loss": 0.4532,
+      "step": 79540
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012789185052926923,
+      "loss": 0.5051,
+      "step": 79550
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012780682736045573,
+      "loss": 0.5215,
+      "step": 79560
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012772180419164223,
+      "loss": 0.468,
+      "step": 79570
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012763678102282872,
+      "loss": 0.4397,
+      "step": 79580
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012755175785401522,
+      "loss": 0.5017,
+      "step": 79590
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012746673468520172,
+      "loss": 0.4144,
+      "step": 79600
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001273817115163882,
+      "loss": 0.4533,
+      "step": 79610
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001272966883475747,
+      "loss": 0.4858,
+      "step": 79620
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001272116651787612,
+      "loss": 0.4654,
+      "step": 79630
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012712664200994773,
+      "loss": 0.4248,
+      "step": 79640
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012704161884113423,
+      "loss": 0.4636,
+      "step": 79650
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012695659567232072,
+      "loss": 0.5111,
+      "step": 79660
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012687157250350722,
+      "loss": 0.5019,
+      "step": 79670
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012678654933469372,
+      "loss": 0.4847,
+      "step": 79680
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001267015261658802,
+      "loss": 0.4924,
+      "step": 79690
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001266165029970667,
+      "loss": 0.4646,
+      "step": 79700
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001265314798282532,
+      "loss": 0.4436,
+      "step": 79710
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001264464566594397,
+      "loss": 0.4684,
+      "step": 79720
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001263614334906262,
+      "loss": 0.4128,
+      "step": 79730
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001262764103218127,
+      "loss": 0.4883,
+      "step": 79740
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00012619138715299922,
+      "loss": 0.5142,
+      "step": 79750
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001261063639841857,
+      "loss": 0.4342,
+      "step": 79760
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001260213408153722,
+      "loss": 0.4648,
+      "step": 79770
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012593631764655868,
+      "loss": 0.4484,
+      "step": 79780
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001258512944777452,
+      "loss": 0.495,
+      "step": 79790
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012576627130893168,
+      "loss": 0.512,
+      "step": 79800
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001256812481401182,
+      "loss": 0.4787,
+      "step": 79810
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012559622497130467,
+      "loss": 0.4551,
+      "step": 79820
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001255112018024912,
+      "loss": 0.4676,
+      "step": 79830
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001254261786336777,
+      "loss": 0.4589,
+      "step": 79840
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012534115546486419,
+      "loss": 0.486,
+      "step": 79850
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012525613229605068,
+      "loss": 0.4734,
+      "step": 79860
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012517110912723718,
+      "loss": 0.5169,
+      "step": 79870
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012508608595842368,
+      "loss": 0.5044,
+      "step": 79880
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012500106278961017,
+      "loss": 0.4509,
+      "step": 79890
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012491603962079667,
+      "loss": 0.4379,
+      "step": 79900
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012483101645198317,
+      "loss": 0.4781,
+      "step": 79910
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012474599328316966,
+      "loss": 0.4922,
+      "step": 79920
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012466097011435616,
+      "loss": 0.4889,
+      "step": 79930
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012457594694554266,
+      "loss": 0.4701,
+      "step": 79940
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012449092377672918,
+      "loss": 0.4359,
+      "step": 79950
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012440590060791565,
+      "loss": 0.4172,
+      "step": 79960
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012432087743910217,
+      "loss": 0.4445,
+      "step": 79970
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00012423585427028864,
+      "loss": 0.4843,
+      "step": 79980
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012415083110147517,
+      "loss": 0.4576,
+      "step": 79990
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012406580793266164,
+      "loss": 0.4499,
+      "step": 80000
+    },
+    {
+      "epoch": 6.77,
+      "eval_loss": 0.5771708488464355,
+      "eval_runtime": 312.5656,
+      "eval_samples_per_second": 16.812,
+      "eval_steps_per_second": 2.102,
+      "step": 80000
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012398078476384816,
+      "loss": 0.4332,
+      "step": 80010
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012389576159503466,
+      "loss": 0.5419,
+      "step": 80020
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012381073842622115,
+      "loss": 0.5019,
+      "step": 80030
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012372571525740765,
+      "loss": 0.4262,
+      "step": 80040
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012364069208859415,
+      "loss": 0.4986,
+      "step": 80050
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012355566891978064,
+      "loss": 0.4618,
+      "step": 80060
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012347064575096714,
+      "loss": 0.4538,
+      "step": 80070
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012338562258215364,
+      "loss": 0.4875,
+      "step": 80080
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012330059941334013,
+      "loss": 0.4681,
+      "step": 80090
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012321557624452666,
+      "loss": 0.4427,
+      "step": 80100
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012313055307571313,
+      "loss": 0.5063,
+      "step": 80110
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012304552990689965,
+      "loss": 0.4841,
+      "step": 80120
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012296050673808615,
+      "loss": 0.474,
+      "step": 80130
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012287548356927264,
+      "loss": 0.4756,
+      "step": 80140
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012279046040045914,
+      "loss": 0.5035,
+      "step": 80150
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012270543723164564,
+      "loss": 0.521,
+      "step": 80160
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012262041406283213,
+      "loss": 0.4306,
+      "step": 80170
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012253539089401863,
+      "loss": 0.5009,
+      "step": 80180
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012245036772520513,
+      "loss": 0.4849,
+      "step": 80190
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012236534455639162,
+      "loss": 0.4718,
+      "step": 80200
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012228032138757812,
+      "loss": 0.5018,
+      "step": 80210
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012219529821876462,
+      "loss": 0.4572,
+      "step": 80220
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0001221102750499511,
+      "loss": 0.5188,
+      "step": 80230
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012202525188113762,
+      "loss": 0.4592,
+      "step": 80240
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0001219402287123241,
+      "loss": 0.4929,
+      "step": 80250
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012185520554351062,
+      "loss": 0.4659,
+      "step": 80260
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012177018237469711,
+      "loss": 0.4732,
+      "step": 80270
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012168515920588361,
+      "loss": 0.4437,
+      "step": 80280
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0001216001360370701,
+      "loss": 0.442,
+      "step": 80290
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0001215151128682566,
+      "loss": 0.5136,
+      "step": 80300
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0001214300896994431,
+      "loss": 0.4095,
+      "step": 80310
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012134506653062961,
+      "loss": 0.4184,
+      "step": 80320
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0001212600433618161,
+      "loss": 0.4461,
+      "step": 80330
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0001211750201930026,
+      "loss": 0.4408,
+      "step": 80340
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012108999702418911,
+      "loss": 0.541,
+      "step": 80350
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0001210049738553756,
+      "loss": 0.4856,
+      "step": 80360
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0001209199506865621,
+      "loss": 0.4819,
+      "step": 80370
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012083492751774859,
+      "loss": 0.5081,
+      "step": 80380
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0001207499043489351,
+      "loss": 0.5032,
+      "step": 80390
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012066488118012158,
+      "loss": 0.489,
+      "step": 80400
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012057985801130809,
+      "loss": 0.5099,
+      "step": 80410
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012049483484249458,
+      "loss": 0.4511,
+      "step": 80420
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012040981167368109,
+      "loss": 0.4804,
+      "step": 80430
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0001203247885048676,
+      "loss": 0.4917,
+      "step": 80440
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012023976533605408,
+      "loss": 0.5543,
+      "step": 80450
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00012015474216724059,
+      "loss": 0.4769,
+      "step": 80460
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00012006971899842707,
+      "loss": 0.557,
+      "step": 80470
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011998469582961358,
+      "loss": 0.4717,
+      "step": 80480
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011989967266080007,
+      "loss": 0.4312,
+      "step": 80490
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011981464949198658,
+      "loss": 0.4494,
+      "step": 80500
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011972962632317306,
+      "loss": 0.4569,
+      "step": 80510
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011964460315435957,
+      "loss": 0.4846,
+      "step": 80520
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011955957998554608,
+      "loss": 0.5011,
+      "step": 80530
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011947455681673256,
+      "loss": 0.4077,
+      "step": 80540
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011938953364791907,
+      "loss": 0.4085,
+      "step": 80550
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011930451047910556,
+      "loss": 0.447,
+      "step": 80560
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011921948731029207,
+      "loss": 0.5159,
+      "step": 80570
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00011913446414147855,
+      "loss": 0.4276,
+      "step": 80580
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011904944097266506,
+      "loss": 0.44,
+      "step": 80590
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011896441780385154,
+      "loss": 0.4436,
+      "step": 80600
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011887939463503805,
+      "loss": 0.5109,
+      "step": 80610
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011879437146622454,
+      "loss": 0.4878,
+      "step": 80620
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011870934829741105,
+      "loss": 0.5007,
+      "step": 80630
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011862432512859756,
+      "loss": 0.4692,
+      "step": 80640
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011853930195978404,
+      "loss": 0.4858,
+      "step": 80650
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011845427879097055,
+      "loss": 0.4991,
+      "step": 80660
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011836925562215703,
+      "loss": 0.4435,
+      "step": 80670
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011828423245334354,
+      "loss": 0.4553,
+      "step": 80680
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00011819920928453004,
+      "loss": 0.5304,
+      "step": 80690
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011811418611571654,
+      "loss": 0.4886,
+      "step": 80700
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011802916294690303,
+      "loss": 0.4771,
+      "step": 80710
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011794413977808953,
+      "loss": 0.4732,
+      "step": 80720
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011785911660927604,
+      "loss": 0.489,
+      "step": 80730
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011777409344046254,
+      "loss": 0.4958,
+      "step": 80740
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011768907027164903,
+      "loss": 0.4811,
+      "step": 80750
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011760404710283553,
+      "loss": 0.529,
+      "step": 80760
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011751902393402204,
+      "loss": 0.4733,
+      "step": 80770
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011743400076520852,
+      "loss": 0.4652,
+      "step": 80780
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011734897759639503,
+      "loss": 0.4071,
+      "step": 80790
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011726395442758152,
+      "loss": 0.4318,
+      "step": 80800
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00011717893125876803,
+      "loss": 0.5108,
+      "step": 80810
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011709390808995451,
+      "loss": 0.5465,
+      "step": 80820
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011700888492114102,
+      "loss": 0.4406,
+      "step": 80830
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011692386175232753,
+      "loss": 0.4658,
+      "step": 80840
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011683883858351401,
+      "loss": 0.4437,
+      "step": 80850
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011675381541470052,
+      "loss": 0.4559,
+      "step": 80860
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.000116668792245887,
+      "loss": 0.481,
+      "step": 80870
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011658376907707352,
+      "loss": 0.4932,
+      "step": 80880
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011649874590826,
+      "loss": 0.4813,
+      "step": 80890
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011641372273944651,
+      "loss": 0.4409,
+      "step": 80900
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011632869957063299,
+      "loss": 0.4382,
+      "step": 80910
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.0001162436764018195,
+      "loss": 0.5431,
+      "step": 80920
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00011615865323300601,
+      "loss": 0.4376,
+      "step": 80930
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.0001160736300641925,
+      "loss": 0.4745,
+      "step": 80940
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.000115988606895379,
+      "loss": 0.4671,
+      "step": 80950
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011590358372656549,
+      "loss": 0.4497,
+      "step": 80960
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.000115818560557752,
+      "loss": 0.468,
+      "step": 80970
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011573353738893848,
+      "loss": 0.4805,
+      "step": 80980
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011564851422012499,
+      "loss": 0.4651,
+      "step": 80990
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011556349105131147,
+      "loss": 0.5288,
+      "step": 81000
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011547846788249799,
+      "loss": 0.4455,
+      "step": 81010
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.0001153934447136845,
+      "loss": 0.4593,
+      "step": 81020
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011530842154487098,
+      "loss": 0.417,
+      "step": 81030
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011522339837605749,
+      "loss": 0.4913,
+      "step": 81040
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00011513837520724397,
+      "loss": 0.5375,
+      "step": 81050
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011505335203843048,
+      "loss": 0.4244,
+      "step": 81060
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011496832886961696,
+      "loss": 0.441,
+      "step": 81070
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011488330570080348,
+      "loss": 0.4699,
+      "step": 81080
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011479828253198996,
+      "loss": 0.4311,
+      "step": 81090
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011471325936317647,
+      "loss": 0.5104,
+      "step": 81100
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011462823619436296,
+      "loss": 0.4413,
+      "step": 81110
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011454321302554946,
+      "loss": 0.4431,
+      "step": 81120
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011445818985673597,
+      "loss": 0.5149,
+      "step": 81130
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011437316668792247,
+      "loss": 0.471,
+      "step": 81140
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011428814351910897,
+      "loss": 0.4624,
+      "step": 81150
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011420312035029546,
+      "loss": 0.5135,
+      "step": 81160
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00011411809718148196,
+      "loss": 0.488,
+      "step": 81170
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011403307401266845,
+      "loss": 0.5301,
+      "step": 81180
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011394805084385497,
+      "loss": 0.4484,
+      "step": 81190
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011386302767504145,
+      "loss": 0.4974,
+      "step": 81200
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011377800450622796,
+      "loss": 0.4766,
+      "step": 81210
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011369298133741446,
+      "loss": 0.4187,
+      "step": 81220
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011360795816860095,
+      "loss": 0.4913,
+      "step": 81230
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011352293499978746,
+      "loss": 0.4274,
+      "step": 81240
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011343791183097394,
+      "loss": 0.4361,
+      "step": 81250
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011335288866216046,
+      "loss": 0.4476,
+      "step": 81260
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011326786549334694,
+      "loss": 0.4725,
+      "step": 81270
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00011318284232453345,
+      "loss": 0.3999,
+      "step": 81280
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011309781915571993,
+      "loss": 0.467,
+      "step": 81290
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011301279598690644,
+      "loss": 0.474,
+      "step": 81300
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011292777281809292,
+      "loss": 0.4721,
+      "step": 81310
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011284274964927943,
+      "loss": 0.4609,
+      "step": 81320
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011275772648046595,
+      "loss": 0.5312,
+      "step": 81330
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011267270331165243,
+      "loss": 0.5048,
+      "step": 81340
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011258768014283894,
+      "loss": 0.4508,
+      "step": 81350
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011250265697402542,
+      "loss": 0.4764,
+      "step": 81360
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011241763380521193,
+      "loss": 0.4815,
+      "step": 81370
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011233261063639841,
+      "loss": 0.4669,
+      "step": 81380
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011224758746758492,
+      "loss": 0.5285,
+      "step": 81390
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00011216256429877141,
+      "loss": 0.5192,
+      "step": 81400
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011207754112995792,
+      "loss": 0.4551,
+      "step": 81410
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011199251796114443,
+      "loss": 0.4223,
+      "step": 81420
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011190749479233091,
+      "loss": 0.476,
+      "step": 81430
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011182247162351742,
+      "loss": 0.5081,
+      "step": 81440
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0001117374484547039,
+      "loss": 0.4942,
+      "step": 81450
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011165242528589041,
+      "loss": 0.4197,
+      "step": 81460
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0001115674021170769,
+      "loss": 0.4133,
+      "step": 81470
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011148237894826341,
+      "loss": 0.4641,
+      "step": 81480
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011139735577944989,
+      "loss": 0.4799,
+      "step": 81490
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0001113123326106364,
+      "loss": 0.5106,
+      "step": 81500
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00011122730944182288,
+      "loss": 0.5299,
+      "step": 81510
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0001111422862730094,
+      "loss": 0.4363,
+      "step": 81520
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0001110572631041959,
+      "loss": 0.4571,
+      "step": 81530
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011097223993538239,
+      "loss": 0.4861,
+      "step": 81540
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0001108872167665689,
+      "loss": 0.4267,
+      "step": 81550
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0001108021935977554,
+      "loss": 0.466,
+      "step": 81560
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011071717042894189,
+      "loss": 0.5278,
+      "step": 81570
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011063214726012839,
+      "loss": 0.4308,
+      "step": 81580
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011054712409131488,
+      "loss": 0.491,
+      "step": 81590
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011046210092250138,
+      "loss": 0.4682,
+      "step": 81600
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011037707775368789,
+      "loss": 0.504,
+      "step": 81610
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011029205458487439,
+      "loss": 0.4231,
+      "step": 81620
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011020703141606088,
+      "loss": 0.5023,
+      "step": 81630
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00011012200824724738,
+      "loss": 0.4241,
+      "step": 81640
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00011003698507843388,
+      "loss": 0.469,
+      "step": 81650
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010995196190962039,
+      "loss": 0.5361,
+      "step": 81660
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010986693874080687,
+      "loss": 0.4889,
+      "step": 81670
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010978191557199338,
+      "loss": 0.4605,
+      "step": 81680
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010969689240317986,
+      "loss": 0.4814,
+      "step": 81690
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010961186923436637,
+      "loss": 0.4103,
+      "step": 81700
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010952684606555288,
+      "loss": 0.4695,
+      "step": 81710
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010944182289673937,
+      "loss": 0.4595,
+      "step": 81720
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010935679972792588,
+      "loss": 0.4298,
+      "step": 81730
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010927177655911236,
+      "loss": 0.4853,
+      "step": 81740
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010918675339029887,
+      "loss": 0.4523,
+      "step": 81750
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010910173022148535,
+      "loss": 0.5212,
+      "step": 81760
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010901670705267186,
+      "loss": 0.4329,
+      "step": 81770
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010893168388385835,
+      "loss": 0.4408,
+      "step": 81780
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010884666071504486,
+      "loss": 0.4664,
+      "step": 81790
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010876163754623134,
+      "loss": 0.4956,
+      "step": 81800
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010867661437741785,
+      "loss": 0.498,
+      "step": 81810
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010859159120860436,
+      "loss": 0.4439,
+      "step": 81820
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010850656803979084,
+      "loss": 0.4812,
+      "step": 81830
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010842154487097735,
+      "loss": 0.4691,
+      "step": 81840
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010833652170216384,
+      "loss": 0.4677,
+      "step": 81850
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010825149853335035,
+      "loss": 0.4175,
+      "step": 81860
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010816647536453683,
+      "loss": 0.49,
+      "step": 81870
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010808145219572334,
+      "loss": 0.5169,
+      "step": 81880
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010799642902690982,
+      "loss": 0.4866,
+      "step": 81890
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010791140585809633,
+      "loss": 0.5272,
+      "step": 81900
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010782638268928284,
+      "loss": 0.4266,
+      "step": 81910
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010774135952046933,
+      "loss": 0.4483,
+      "step": 81920
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010765633635165584,
+      "loss": 0.4655,
+      "step": 81930
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010757131318284232,
+      "loss": 0.5188,
+      "step": 81940
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010748629001402883,
+      "loss": 0.4868,
+      "step": 81950
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010740126684521531,
+      "loss": 0.4866,
+      "step": 81960
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010731624367640182,
+      "loss": 0.4552,
+      "step": 81970
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010723122050758832,
+      "loss": 0.4839,
+      "step": 81980
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00010714619733877482,
+      "loss": 0.5123,
+      "step": 81990
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010706117416996131,
+      "loss": 0.4429,
+      "step": 82000
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010697615100114781,
+      "loss": 0.4423,
+      "step": 82010
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010689112783233432,
+      "loss": 0.4693,
+      "step": 82020
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010680610466352082,
+      "loss": 0.4754,
+      "step": 82030
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010672108149470731,
+      "loss": 0.4765,
+      "step": 82040
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010663605832589381,
+      "loss": 0.4848,
+      "step": 82050
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010655103515708031,
+      "loss": 0.4042,
+      "step": 82060
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.0001064660119882668,
+      "loss": 0.4744,
+      "step": 82070
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010638098881945331,
+      "loss": 0.493,
+      "step": 82080
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.0001062959656506398,
+      "loss": 0.4903,
+      "step": 82090
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010621094248182631,
+      "loss": 0.4523,
+      "step": 82100
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010612591931301282,
+      "loss": 0.4387,
+      "step": 82110
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0001060408961441993,
+      "loss": 0.5024,
+      "step": 82120
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010595587297538581,
+      "loss": 0.4828,
+      "step": 82130
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0001058708498065723,
+      "loss": 0.5199,
+      "step": 82140
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0001057858266377588,
+      "loss": 0.5035,
+      "step": 82150
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010570080346894529,
+      "loss": 0.5055,
+      "step": 82160
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0001056157803001318,
+      "loss": 0.4136,
+      "step": 82170
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010553075713131828,
+      "loss": 0.4705,
+      "step": 82180
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010544573396250479,
+      "loss": 0.5084,
+      "step": 82190
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010536071079369127,
+      "loss": 0.5079,
+      "step": 82200
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010527568762487778,
+      "loss": 0.4567,
+      "step": 82210
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0001051906644560643,
+      "loss": 0.4367,
+      "step": 82220
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010510564128725078,
+      "loss": 0.4915,
+      "step": 82230
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010502061811843729,
+      "loss": 0.5148,
+      "step": 82240
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010493559494962377,
+      "loss": 0.4516,
+      "step": 82250
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010485057178081028,
+      "loss": 0.5174,
+      "step": 82260
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010476554861199676,
+      "loss": 0.4813,
+      "step": 82270
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010468052544318327,
+      "loss": 0.4667,
+      "step": 82280
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010459550227436976,
+      "loss": 0.4859,
+      "step": 82290
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010451047910555627,
+      "loss": 0.5174,
+      "step": 82300
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010442545593674278,
+      "loss": 0.4724,
+      "step": 82310
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010434043276792926,
+      "loss": 0.456,
+      "step": 82320
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010425540959911577,
+      "loss": 0.5067,
+      "step": 82330
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010417038643030225,
+      "loss": 0.4947,
+      "step": 82340
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010408536326148876,
+      "loss": 0.4133,
+      "step": 82350
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010400034009267525,
+      "loss": 0.5182,
+      "step": 82360
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010391531692386176,
+      "loss": 0.5106,
+      "step": 82370
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010383029375504824,
+      "loss": 0.4926,
+      "step": 82380
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010374527058623475,
+      "loss": 0.5318,
+      "step": 82390
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010366024741742125,
+      "loss": 0.4638,
+      "step": 82400
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010357522424860774,
+      "loss": 0.4491,
+      "step": 82410
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010349020107979425,
+      "loss": 0.4543,
+      "step": 82420
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010340517791098074,
+      "loss": 0.4686,
+      "step": 82430
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010332015474216725,
+      "loss": 0.4359,
+      "step": 82440
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010323513157335374,
+      "loss": 0.4699,
+      "step": 82450
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010315010840454024,
+      "loss": 0.4645,
+      "step": 82460
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010306508523572674,
+      "loss": 0.503,
+      "step": 82470
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010298006206691325,
+      "loss": 0.4844,
+      "step": 82480
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010289503889809973,
+      "loss": 0.5775,
+      "step": 82490
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010281001572928624,
+      "loss": 0.4581,
+      "step": 82500
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010272499256047274,
+      "loss": 0.4789,
+      "step": 82510
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010263996939165923,
+      "loss": 0.4874,
+      "step": 82520
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010255494622284574,
+      "loss": 0.4467,
+      "step": 82530
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010246992305403223,
+      "loss": 0.4504,
+      "step": 82540
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010238489988521874,
+      "loss": 0.454,
+      "step": 82550
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010229987671640522,
+      "loss": 0.4735,
+      "step": 82560
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010221485354759173,
+      "loss": 0.4517,
+      "step": 82570
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010212983037877821,
+      "loss": 0.4938,
+      "step": 82580
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00010204480720996472,
+      "loss": 0.4977,
+      "step": 82590
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010195978404115123,
+      "loss": 0.4435,
+      "step": 82600
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010187476087233772,
+      "loss": 0.5022,
+      "step": 82610
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010178973770352423,
+      "loss": 0.4598,
+      "step": 82620
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010170471453471071,
+      "loss": 0.466,
+      "step": 82630
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010161969136589722,
+      "loss": 0.4743,
+      "step": 82640
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0001015346681970837,
+      "loss": 0.4286,
+      "step": 82650
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010144964502827021,
+      "loss": 0.4601,
+      "step": 82660
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0001013646218594567,
+      "loss": 0.4606,
+      "step": 82670
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010127959869064321,
+      "loss": 0.4974,
+      "step": 82680
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010119457552182969,
+      "loss": 0.4241,
+      "step": 82690
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0001011095523530162,
+      "loss": 0.4423,
+      "step": 82700
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010102452918420271,
+      "loss": 0.5469,
+      "step": 82710
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.0001009395060153892,
+      "loss": 0.424,
+      "step": 82720
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.0001008544828465757,
+      "loss": 0.4917,
+      "step": 82730
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010076945967776219,
+      "loss": 0.4384,
+      "step": 82740
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.0001006844365089487,
+      "loss": 0.4302,
+      "step": 82750
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010059941334013518,
+      "loss": 0.4757,
+      "step": 82760
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010051439017132169,
+      "loss": 0.4922,
+      "step": 82770
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010042936700250817,
+      "loss": 0.5116,
+      "step": 82780
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010034434383369468,
+      "loss": 0.4431,
+      "step": 82790
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.0001002593206648812,
+      "loss": 0.4129,
+      "step": 82800
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010017429749606768,
+      "loss": 0.4633,
+      "step": 82810
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010008927432725419,
+      "loss": 0.4425,
+      "step": 82820
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.00010000425115844067,
+      "loss": 0.4291,
+      "step": 82830
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.991922798962717e-05,
+      "loss": 0.4442,
+      "step": 82840
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.983420482081368e-05,
+      "loss": 0.4328,
+      "step": 82850
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.974918165200017e-05,
+      "loss": 0.415,
+      "step": 82860
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.966415848318667e-05,
+      "loss": 0.4204,
+      "step": 82870
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.957913531437317e-05,
+      "loss": 0.4808,
+      "step": 82880
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.949411214555966e-05,
+      "loss": 0.5045,
+      "step": 82890
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.940908897674617e-05,
+      "loss": 0.4679,
+      "step": 82900
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.932406580793267e-05,
+      "loss": 0.4627,
+      "step": 82910
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.923904263911917e-05,
+      "loss": 0.4432,
+      "step": 82920
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.915401947030566e-05,
+      "loss": 0.4673,
+      "step": 82930
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.906899630149216e-05,
+      "loss": 0.3941,
+      "step": 82940
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.898397313267867e-05,
+      "loss": 0.429,
+      "step": 82950
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.889894996386517e-05,
+      "loss": 0.4531,
+      "step": 82960
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.881392679505166e-05,
+      "loss": 0.4783,
+      "step": 82970
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.872890362623816e-05,
+      "loss": 0.438,
+      "step": 82980
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.864388045742466e-05,
+      "loss": 0.4395,
+      "step": 82990
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.855885728861115e-05,
+      "loss": 0.4684,
+      "step": 83000
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.847383411979765e-05,
+      "loss": 0.4401,
+      "step": 83010
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.838881095098415e-05,
+      "loss": 0.4601,
+      "step": 83020
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.830378778217064e-05,
+      "loss": 0.4505,
+      "step": 83030
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.821876461335714e-05,
+      "loss": 0.4516,
+      "step": 83040
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.813374144454365e-05,
+      "loss": 0.4439,
+      "step": 83050
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.804871827573015e-05,
+      "loss": 0.4249,
+      "step": 83060
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.796369510691664e-05,
+      "loss": 0.445,
+      "step": 83070
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.787867193810314e-05,
+      "loss": 0.4413,
+      "step": 83080
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.779364876928964e-05,
+      "loss": 0.4456,
+      "step": 83090
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.770862560047613e-05,
+      "loss": 0.4815,
+      "step": 83100
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.762360243166263e-05,
+      "loss": 0.4112,
+      "step": 83110
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.753857926284913e-05,
+      "loss": 0.4696,
+      "step": 83120
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.745355609403562e-05,
+      "loss": 0.4112,
+      "step": 83130
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.736853292522212e-05,
+      "loss": 0.4346,
+      "step": 83140
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.728350975640863e-05,
+      "loss": 0.4367,
+      "step": 83150
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.719848658759513e-05,
+      "loss": 0.4846,
+      "step": 83160
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.711346341878162e-05,
+      "loss": 0.4847,
+      "step": 83170
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.702844024996812e-05,
+      "loss": 0.4759,
+      "step": 83180
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.694341708115462e-05,
+      "loss": 0.4709,
+      "step": 83190
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.685839391234111e-05,
+      "loss": 0.4259,
+      "step": 83200
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.677337074352761e-05,
+      "loss": 0.4473,
+      "step": 83210
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.668834757471411e-05,
+      "loss": 0.4324,
+      "step": 83220
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.66033244059006e-05,
+      "loss": 0.4032,
+      "step": 83230
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.65183012370871e-05,
+      "loss": 0.4832,
+      "step": 83240
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.643327806827361e-05,
+      "loss": 0.4245,
+      "step": 83250
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.634825489946011e-05,
+      "loss": 0.4632,
+      "step": 83260
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.62632317306466e-05,
+      "loss": 0.4242,
+      "step": 83270
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.61782085618331e-05,
+      "loss": 0.4376,
+      "step": 83280
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.60931853930196e-05,
+      "loss": 0.4347,
+      "step": 83290
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.60081622242061e-05,
+      "loss": 0.4752,
+      "step": 83300
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.592313905539259e-05,
+      "loss": 0.4117,
+      "step": 83310
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.58381158865791e-05,
+      "loss": 0.4451,
+      "step": 83320
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.57530927177656e-05,
+      "loss": 0.4581,
+      "step": 83330
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.56680695489521e-05,
+      "loss": 0.4383,
+      "step": 83340
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.558304638013859e-05,
+      "loss": 0.4093,
+      "step": 83350
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.54980232113251e-05,
+      "loss": 0.4664,
+      "step": 83360
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.54130000425116e-05,
+      "loss": 0.4269,
+      "step": 83370
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.53279768736981e-05,
+      "loss": 0.4996,
+      "step": 83380
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.524295370488459e-05,
+      "loss": 0.472,
+      "step": 83390
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.515793053607109e-05,
+      "loss": 0.4761,
+      "step": 83400
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.507290736725758e-05,
+      "loss": 0.4514,
+      "step": 83410
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.498788419844408e-05,
+      "loss": 0.4556,
+      "step": 83420
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.490286102963058e-05,
+      "loss": 0.4048,
+      "step": 83430
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.481783786081709e-05,
+      "loss": 0.4897,
+      "step": 83440
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.473281469200358e-05,
+      "loss": 0.4905,
+      "step": 83450
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.464779152319008e-05,
+      "loss": 0.4532,
+      "step": 83460
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.456276835437658e-05,
+      "loss": 0.4478,
+      "step": 83470
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.447774518556307e-05,
+      "loss": 0.4784,
+      "step": 83480
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.439272201674957e-05,
+      "loss": 0.5155,
+      "step": 83490
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.430769884793607e-05,
+      "loss": 0.4686,
+      "step": 83500
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.422267567912256e-05,
+      "loss": 0.4952,
+      "step": 83510
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.413765251030906e-05,
+      "loss": 0.4457,
+      "step": 83520
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.405262934149556e-05,
+      "loss": 0.4845,
+      "step": 83530
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.396760617268207e-05,
+      "loss": 0.4571,
+      "step": 83540
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.388258300386856e-05,
+      "loss": 0.5148,
+      "step": 83550
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.379755983505506e-05,
+      "loss": 0.4837,
+      "step": 83560
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.371253666624156e-05,
+      "loss": 0.4506,
+      "step": 83570
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.362751349742805e-05,
+      "loss": 0.4636,
+      "step": 83580
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.354249032861455e-05,
+      "loss": 0.4362,
+      "step": 83590
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.345746715980105e-05,
+      "loss": 0.4826,
+      "step": 83600
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.337244399098754e-05,
+      "loss": 0.4551,
+      "step": 83610
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.328742082217404e-05,
+      "loss": 0.4652,
+      "step": 83620
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.320239765336054e-05,
+      "loss": 0.4611,
+      "step": 83630
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.311737448454705e-05,
+      "loss": 0.4348,
+      "step": 83640
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.303235131573354e-05,
+      "loss": 0.4955,
+      "step": 83650
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.294732814692004e-05,
+      "loss": 0.4087,
+      "step": 83660
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.286230497810654e-05,
+      "loss": 0.4183,
+      "step": 83670
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.277728180929303e-05,
+      "loss": 0.4352,
+      "step": 83680
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.269225864047953e-05,
+      "loss": 0.4783,
+      "step": 83690
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.260723547166603e-05,
+      "loss": 0.453,
+      "step": 83700
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.252221230285252e-05,
+      "loss": 0.4107,
+      "step": 83710
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.243718913403902e-05,
+      "loss": 0.4135,
+      "step": 83720
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.235216596522553e-05,
+      "loss": 0.437,
+      "step": 83730
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.226714279641203e-05,
+      "loss": 0.4864,
+      "step": 83740
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.218211962759852e-05,
+      "loss": 0.4635,
+      "step": 83750
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.209709645878502e-05,
+      "loss": 0.4161,
+      "step": 83760
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.201207328997152e-05,
+      "loss": 0.4261,
+      "step": 83770
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.192705012115803e-05,
+      "loss": 0.4467,
+      "step": 83780
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.184202695234452e-05,
+      "loss": 0.4645,
+      "step": 83790
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.175700378353102e-05,
+      "loss": 0.4439,
+      "step": 83800
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.167198061471752e-05,
+      "loss": 0.5187,
+      "step": 83810
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.158695744590401e-05,
+      "loss": 0.4192,
+      "step": 83820
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.150193427709051e-05,
+      "loss": 0.4598,
+      "step": 83830
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.141691110827702e-05,
+      "loss": 0.4633,
+      "step": 83840
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.133188793946352e-05,
+      "loss": 0.4029,
+      "step": 83850
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.124686477065001e-05,
+      "loss": 0.4222,
+      "step": 83860
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.116184160183651e-05,
+      "loss": 0.4463,
+      "step": 83870
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.1076818433023e-05,
+      "loss": 0.4813,
+      "step": 83880
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.09917952642095e-05,
+      "loss": 0.4576,
+      "step": 83890
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.0906772095396e-05,
+      "loss": 0.4284,
+      "step": 83900
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.08217489265825e-05,
+      "loss": 0.4595,
+      "step": 83910
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.0736725757769e-05,
+      "loss": 0.4419,
+      "step": 83920
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.065170258895549e-05,
+      "loss": 0.4295,
+      "step": 83930
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.0566679420142e-05,
+      "loss": 0.4517,
+      "step": 83940
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.04816562513285e-05,
+      "loss": 0.5001,
+      "step": 83950
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.0396633082515e-05,
+      "loss": 0.3672,
+      "step": 83960
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.031160991370149e-05,
+      "loss": 0.4611,
+      "step": 83970
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.022658674488799e-05,
+      "loss": 0.4231,
+      "step": 83980
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.014156357607448e-05,
+      "loss": 0.4641,
+      "step": 83990
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.005654040726098e-05,
+      "loss": 0.4612,
+      "step": 84000
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.997151723844748e-05,
+      "loss": 0.4874,
+      "step": 84010
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.988649406963397e-05,
+      "loss": 0.433,
+      "step": 84020
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.980147090082047e-05,
+      "loss": 0.4139,
+      "step": 84030
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.971644773200698e-05,
+      "loss": 0.4425,
+      "step": 84040
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.963142456319348e-05,
+      "loss": 0.4483,
+      "step": 84050
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.954640139437997e-05,
+      "loss": 0.4401,
+      "step": 84060
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.946137822556647e-05,
+      "loss": 0.5062,
+      "step": 84070
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.937635505675297e-05,
+      "loss": 0.4725,
+      "step": 84080
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.929133188793946e-05,
+      "loss": 0.4732,
+      "step": 84090
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.920630871912596e-05,
+      "loss": 0.4248,
+      "step": 84100
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.912128555031246e-05,
+      "loss": 0.5115,
+      "step": 84110
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.903626238149895e-05,
+      "loss": 0.4513,
+      "step": 84120
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.895123921268546e-05,
+      "loss": 0.4804,
+      "step": 84130
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.886621604387196e-05,
+      "loss": 0.5058,
+      "step": 84140
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.878119287505846e-05,
+      "loss": 0.466,
+      "step": 84150
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.869616970624495e-05,
+      "loss": 0.4323,
+      "step": 84160
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.861114653743145e-05,
+      "loss": 0.4802,
+      "step": 84170
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.852612336861795e-05,
+      "loss": 0.4174,
+      "step": 84180
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.844110019980444e-05,
+      "loss": 0.5147,
+      "step": 84190
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.835607703099095e-05,
+      "loss": 0.4383,
+      "step": 84200
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.827105386217745e-05,
+      "loss": 0.4347,
+      "step": 84210
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.818603069336395e-05,
+      "loss": 0.436,
+      "step": 84220
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.810100752455044e-05,
+      "loss": 0.485,
+      "step": 84230
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.801598435573695e-05,
+      "loss": 0.4289,
+      "step": 84240
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.793096118692345e-05,
+      "loss": 0.4143,
+      "step": 84250
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.784593801810995e-05,
+      "loss": 0.4853,
+      "step": 84260
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.776091484929644e-05,
+      "loss": 0.431,
+      "step": 84270
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.767589168048294e-05,
+      "loss": 0.4742,
+      "step": 84280
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.759086851166944e-05,
+      "loss": 0.4767,
+      "step": 84290
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.750584534285593e-05,
+      "loss": 0.3991,
+      "step": 84300
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.742082217404243e-05,
+      "loss": 0.4188,
+      "step": 84310
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.733579900522893e-05,
+      "loss": 0.4715,
+      "step": 84320
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.725077583641544e-05,
+      "loss": 0.455,
+      "step": 84330
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.716575266760193e-05,
+      "loss": 0.4545,
+      "step": 84340
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.708072949878843e-05,
+      "loss": 0.5022,
+      "step": 84350
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.699570632997493e-05,
+      "loss": 0.4522,
+      "step": 84360
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.691068316116142e-05,
+      "loss": 0.4614,
+      "step": 84370
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.682565999234792e-05,
+      "loss": 0.443,
+      "step": 84380
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.674063682353442e-05,
+      "loss": 0.4325,
+      "step": 84390
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.665561365472091e-05,
+      "loss": 0.4978,
+      "step": 84400
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.657059048590741e-05,
+      "loss": 0.4811,
+      "step": 84410
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.64855673170939e-05,
+      "loss": 0.4331,
+      "step": 84420
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.640054414828042e-05,
+      "loss": 0.4555,
+      "step": 84430
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.631552097946691e-05,
+      "loss": 0.4369,
+      "step": 84440
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.623049781065341e-05,
+      "loss": 0.46,
+      "step": 84450
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.61454746418399e-05,
+      "loss": 0.4657,
+      "step": 84460
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.60604514730264e-05,
+      "loss": 0.441,
+      "step": 84470
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.59754283042129e-05,
+      "loss": 0.4219,
+      "step": 84480
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.58904051353994e-05,
+      "loss": 0.4233,
+      "step": 84490
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.580538196658589e-05,
+      "loss": 0.46,
+      "step": 84500
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.572035879777239e-05,
+      "loss": 0.5212,
+      "step": 84510
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.563533562895889e-05,
+      "loss": 0.4424,
+      "step": 84520
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.55503124601454e-05,
+      "loss": 0.4428,
+      "step": 84530
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.546528929133189e-05,
+      "loss": 0.4958,
+      "step": 84540
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.538026612251839e-05,
+      "loss": 0.4611,
+      "step": 84550
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.529524295370489e-05,
+      "loss": 0.466,
+      "step": 84560
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.521021978489138e-05,
+      "loss": 0.4316,
+      "step": 84570
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.512519661607788e-05,
+      "loss": 0.4776,
+      "step": 84580
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.504017344726438e-05,
+      "loss": 0.4457,
+      "step": 84590
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.495515027845087e-05,
+      "loss": 0.4261,
+      "step": 84600
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.487012710963738e-05,
+      "loss": 0.4316,
+      "step": 84610
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.478510394082388e-05,
+      "loss": 0.4659,
+      "step": 84620
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.470008077201038e-05,
+      "loss": 0.4394,
+      "step": 84630
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.461505760319687e-05,
+      "loss": 0.4465,
+      "step": 84640
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.453003443438337e-05,
+      "loss": 0.4443,
+      "step": 84650
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.444501126556988e-05,
+      "loss": 0.4549,
+      "step": 84660
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.435998809675638e-05,
+      "loss": 0.4555,
+      "step": 84670
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.427496492794287e-05,
+      "loss": 0.4521,
+      "step": 84680
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.418994175912937e-05,
+      "loss": 0.4518,
+      "step": 84690
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.410491859031587e-05,
+      "loss": 0.4177,
+      "step": 84700
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.401989542150236e-05,
+      "loss": 0.4388,
+      "step": 84710
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.393487225268886e-05,
+      "loss": 0.4899,
+      "step": 84720
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.384984908387537e-05,
+      "loss": 0.4809,
+      "step": 84730
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.376482591506187e-05,
+      "loss": 0.4538,
+      "step": 84740
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.367980274624836e-05,
+      "loss": 0.4879,
+      "step": 84750
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.359477957743486e-05,
+      "loss": 0.4671,
+      "step": 84760
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.350975640862136e-05,
+      "loss": 0.4898,
+      "step": 84770
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.342473323980785e-05,
+      "loss": 0.4997,
+      "step": 84780
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.333971007099435e-05,
+      "loss": 0.4604,
+      "step": 84790
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.325468690218085e-05,
+      "loss": 0.4478,
+      "step": 84800
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.316966373336734e-05,
+      "loss": 0.4232,
+      "step": 84810
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.308464056455384e-05,
+      "loss": 0.4136,
+      "step": 84820
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.299961739574035e-05,
+      "loss": 0.446,
+      "step": 84830
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.291459422692685e-05,
+      "loss": 0.4771,
+      "step": 84840
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.282957105811334e-05,
+      "loss": 0.4556,
+      "step": 84850
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.274454788929984e-05,
+      "loss": 0.4027,
+      "step": 84860
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.265952472048634e-05,
+      "loss": 0.4379,
+      "step": 84870
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.257450155167283e-05,
+      "loss": 0.4218,
+      "step": 84880
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.248947838285933e-05,
+      "loss": 0.4248,
+      "step": 84890
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.240445521404583e-05,
+      "loss": 0.4192,
+      "step": 84900
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.231943204523232e-05,
+      "loss": 0.5094,
+      "step": 84910
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.223440887641882e-05,
+      "loss": 0.4735,
+      "step": 84920
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.214938570760533e-05,
+      "loss": 0.4231,
+      "step": 84930
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.206436253879183e-05,
+      "loss": 0.4531,
+      "step": 84940
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.197933936997832e-05,
+      "loss": 0.5112,
+      "step": 84950
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.189431620116482e-05,
+      "loss": 0.4541,
+      "step": 84960
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.180929303235132e-05,
+      "loss": 0.4727,
+      "step": 84970
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.172426986353781e-05,
+      "loss": 0.4206,
+      "step": 84980
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.163924669472431e-05,
+      "loss": 0.4139,
+      "step": 84990
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.15542235259108e-05,
+      "loss": 0.4639,
+      "step": 85000
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.14692003570973e-05,
+      "loss": 0.4954,
+      "step": 85010
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.138417718828381e-05,
+      "loss": 0.4751,
+      "step": 85020
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.129915401947031e-05,
+      "loss": 0.5065,
+      "step": 85030
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.12141308506568e-05,
+      "loss": 0.4984,
+      "step": 85040
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.11291076818433e-05,
+      "loss": 0.4652,
+      "step": 85050
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.10440845130298e-05,
+      "loss": 0.446,
+      "step": 85060
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.095906134421631e-05,
+      "loss": 0.4602,
+      "step": 85070
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.08740381754028e-05,
+      "loss": 0.5102,
+      "step": 85080
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.07890150065893e-05,
+      "loss": 0.4065,
+      "step": 85090
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.07039918377758e-05,
+      "loss": 0.4515,
+      "step": 85100
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.06189686689623e-05,
+      "loss": 0.4594,
+      "step": 85110
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.05339455001488e-05,
+      "loss": 0.4488,
+      "step": 85120
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.04489223313353e-05,
+      "loss": 0.5323,
+      "step": 85130
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.03638991625218e-05,
+      "loss": 0.3827,
+      "step": 85140
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.02788759937083e-05,
+      "loss": 0.4834,
+      "step": 85150
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.019385282489479e-05,
+      "loss": 0.4253,
+      "step": 85160
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.010882965608129e-05,
+      "loss": 0.4681,
+      "step": 85170
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.002380648726779e-05,
+      "loss": 0.4411,
+      "step": 85180
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 7.993878331845428e-05,
+      "loss": 0.4119,
+      "step": 85190
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.985376014964078e-05,
+      "loss": 0.4781,
+      "step": 85200
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.976873698082728e-05,
+      "loss": 0.4064,
+      "step": 85210
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.968371381201379e-05,
+      "loss": 0.4383,
+      "step": 85220
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.959869064320028e-05,
+      "loss": 0.4268,
+      "step": 85230
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.951366747438678e-05,
+      "loss": 0.4883,
+      "step": 85240
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.942864430557328e-05,
+      "loss": 0.4314,
+      "step": 85250
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.934362113675977e-05,
+      "loss": 0.4796,
+      "step": 85260
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.925859796794627e-05,
+      "loss": 0.4574,
+      "step": 85270
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.917357479913277e-05,
+      "loss": 0.4044,
+      "step": 85280
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.908855163031926e-05,
+      "loss": 0.5022,
+      "step": 85290
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.900352846150576e-05,
+      "loss": 0.5294,
+      "step": 85300
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.891850529269226e-05,
+      "loss": 0.4641,
+      "step": 85310
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.883348212387877e-05,
+      "loss": 0.4507,
+      "step": 85320
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.874845895506526e-05,
+      "loss": 0.4829,
+      "step": 85330
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.866343578625176e-05,
+      "loss": 0.4299,
+      "step": 85340
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.857841261743826e-05,
+      "loss": 0.452,
+      "step": 85350
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.849338944862475e-05,
+      "loss": 0.4237,
+      "step": 85360
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.840836627981125e-05,
+      "loss": 0.4069,
+      "step": 85370
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.832334311099775e-05,
+      "loss": 0.4203,
+      "step": 85380
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.823831994218424e-05,
+      "loss": 0.4192,
+      "step": 85390
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.815329677337074e-05,
+      "loss": 0.455,
+      "step": 85400
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.806827360455724e-05,
+      "loss": 0.4751,
+      "step": 85410
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.798325043574375e-05,
+      "loss": 0.4836,
+      "step": 85420
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.789822726693024e-05,
+      "loss": 0.4542,
+      "step": 85430
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.781320409811674e-05,
+      "loss": 0.469,
+      "step": 85440
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.772818092930324e-05,
+      "loss": 0.4857,
+      "step": 85450
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.764315776048973e-05,
+      "loss": 0.4552,
+      "step": 85460
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.755813459167623e-05,
+      "loss": 0.4656,
+      "step": 85470
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.747311142286273e-05,
+      "loss": 0.4414,
+      "step": 85480
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.738808825404924e-05,
+      "loss": 0.4472,
+      "step": 85490
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.730306508523573e-05,
+      "loss": 0.4225,
+      "step": 85500
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.721804191642223e-05,
+      "loss": 0.4587,
+      "step": 85510
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.713301874760873e-05,
+      "loss": 0.4347,
+      "step": 85520
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.704799557879522e-05,
+      "loss": 0.4596,
+      "step": 85530
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.696297240998173e-05,
+      "loss": 0.4188,
+      "step": 85540
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.687794924116823e-05,
+      "loss": 0.4044,
+      "step": 85550
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.679292607235473e-05,
+      "loss": 0.4424,
+      "step": 85560
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.670790290354122e-05,
+      "loss": 0.4781,
+      "step": 85570
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.662287973472772e-05,
+      "loss": 0.4158,
+      "step": 85580
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.653785656591422e-05,
+      "loss": 0.3681,
+      "step": 85590
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.645283339710071e-05,
+      "loss": 0.441,
+      "step": 85600
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.636781022828721e-05,
+      "loss": 0.517,
+      "step": 85610
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.628278705947372e-05,
+      "loss": 0.455,
+      "step": 85620
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.619776389066022e-05,
+      "loss": 0.4439,
+      "step": 85630
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.611274072184671e-05,
+      "loss": 0.4613,
+      "step": 85640
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.602771755303321e-05,
+      "loss": 0.4097,
+      "step": 85650
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.59426943842197e-05,
+      "loss": 0.4065,
+      "step": 85660
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.58576712154062e-05,
+      "loss": 0.4906,
+      "step": 85670
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.57726480465927e-05,
+      "loss": 0.4738,
+      "step": 85680
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.56876248777792e-05,
+      "loss": 0.4348,
+      "step": 85690
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.560260170896569e-05,
+      "loss": 0.4868,
+      "step": 85700
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.551757854015219e-05,
+      "loss": 0.4343,
+      "step": 85710
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.54325553713387e-05,
+      "loss": 0.4769,
+      "step": 85720
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.53475322025252e-05,
+      "loss": 0.454,
+      "step": 85730
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.526250903371169e-05,
+      "loss": 0.4199,
+      "step": 85740
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.517748586489819e-05,
+      "loss": 0.4145,
+      "step": 85750
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.509246269608469e-05,
+      "loss": 0.4753,
+      "step": 85760
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.500743952727118e-05,
+      "loss": 0.5056,
+      "step": 85770
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.492241635845768e-05,
+      "loss": 0.4696,
+      "step": 85780
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.483739318964418e-05,
+      "loss": 0.4496,
+      "step": 85790
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.475237002083067e-05,
+      "loss": 0.4452,
+      "step": 85800
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.466734685201718e-05,
+      "loss": 0.442,
+      "step": 85810
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.458232368320368e-05,
+      "loss": 0.4162,
+      "step": 85820
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.449730051439018e-05,
+      "loss": 0.4054,
+      "step": 85830
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.441227734557667e-05,
+      "loss": 0.4287,
+      "step": 85840
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.432725417676317e-05,
+      "loss": 0.4422,
+      "step": 85850
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.424223100794967e-05,
+      "loss": 0.4679,
+      "step": 85860
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.415720783913616e-05,
+      "loss": 0.4573,
+      "step": 85870
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.407218467032266e-05,
+      "loss": 0.391,
+      "step": 85880
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.398716150150916e-05,
+      "loss": 0.4445,
+      "step": 85890
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.390213833269565e-05,
+      "loss": 0.3981,
+      "step": 85900
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.381711516388216e-05,
+      "loss": 0.4409,
+      "step": 85910
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.373209199506866e-05,
+      "loss": 0.4418,
+      "step": 85920
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.364706882625516e-05,
+      "loss": 0.4477,
+      "step": 85930
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.356204565744165e-05,
+      "loss": 0.4283,
+      "step": 85940
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.347702248862816e-05,
+      "loss": 0.3977,
+      "step": 85950
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.339199931981466e-05,
+      "loss": 0.4791,
+      "step": 85960
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.330697615100116e-05,
+      "loss": 0.4136,
+      "step": 85970
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.322195298218765e-05,
+      "loss": 0.4491,
+      "step": 85980
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.313692981337415e-05,
+      "loss": 0.4434,
+      "step": 85990
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.305190664456065e-05,
+      "loss": 0.4453,
+      "step": 86000
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.296688347574716e-05,
+      "loss": 0.4577,
+      "step": 86010
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.288186030693365e-05,
+      "loss": 0.4507,
+      "step": 86020
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.279683713812015e-05,
+      "loss": 0.4723,
+      "step": 86030
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.271181396930665e-05,
+      "loss": 0.4298,
+      "step": 86040
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.262679080049314e-05,
+      "loss": 0.503,
+      "step": 86050
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.254176763167964e-05,
+      "loss": 0.4251,
+      "step": 86060
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.245674446286614e-05,
+      "loss": 0.4481,
+      "step": 86070
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.237172129405263e-05,
+      "loss": 0.4347,
+      "step": 86080
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.228669812523913e-05,
+      "loss": 0.4283,
+      "step": 86090
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.220167495642563e-05,
+      "loss": 0.4232,
+      "step": 86100
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.211665178761214e-05,
+      "loss": 0.4104,
+      "step": 86110
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.203162861879863e-05,
+      "loss": 0.4212,
+      "step": 86120
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.194660544998513e-05,
+      "loss": 0.4876,
+      "step": 86130
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.186158228117163e-05,
+      "loss": 0.5087,
+      "step": 86140
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.177655911235812e-05,
+      "loss": 0.4507,
+      "step": 86150
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.169153594354462e-05,
+      "loss": 0.4911,
+      "step": 86160
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.160651277473112e-05,
+      "loss": 0.4383,
+      "step": 86170
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.152148960591761e-05,
+      "loss": 0.4737,
+      "step": 86180
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.143646643710411e-05,
+      "loss": 0.4321,
+      "step": 86190
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.13514432682906e-05,
+      "loss": 0.4743,
+      "step": 86200
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.126642009947712e-05,
+      "loss": 0.5253,
+      "step": 86210
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.118139693066361e-05,
+      "loss": 0.4287,
+      "step": 86220
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.109637376185011e-05,
+      "loss": 0.4928,
+      "step": 86230
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.10113505930366e-05,
+      "loss": 0.4335,
+      "step": 86240
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.09263274242231e-05,
+      "loss": 0.453,
+      "step": 86250
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.08413042554096e-05,
+      "loss": 0.4275,
+      "step": 86260
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.07562810865961e-05,
+      "loss": 0.4456,
+      "step": 86270
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.067125791778259e-05,
+      "loss": 0.4519,
+      "step": 86280
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.058623474896909e-05,
+      "loss": 0.4735,
+      "step": 86290
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.050121158015558e-05,
+      "loss": 0.4665,
+      "step": 86300
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.04161884113421e-05,
+      "loss": 0.4397,
+      "step": 86310
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.033116524252859e-05,
+      "loss": 0.4414,
+      "step": 86320
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.024614207371509e-05,
+      "loss": 0.4094,
+      "step": 86330
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.016111890490159e-05,
+      "loss": 0.4622,
+      "step": 86340
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.007609573608808e-05,
+      "loss": 0.4272,
+      "step": 86350
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.999107256727458e-05,
+      "loss": 0.4345,
+      "step": 86360
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.990604939846109e-05,
+      "loss": 0.462,
+      "step": 86370
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.982102622964759e-05,
+      "loss": 0.4628,
+      "step": 86380
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.973600306083408e-05,
+      "loss": 0.5108,
+      "step": 86390
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.965097989202058e-05,
+      "loss": 0.421,
+      "step": 86400
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.956595672320709e-05,
+      "loss": 0.426,
+      "step": 86410
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.948093355439359e-05,
+      "loss": 0.4594,
+      "step": 86420
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.939591038558008e-05,
+      "loss": 0.4824,
+      "step": 86430
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.931088721676658e-05,
+      "loss": 0.4539,
+      "step": 86440
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.922586404795308e-05,
+      "loss": 0.4708,
+      "step": 86450
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.914084087913957e-05,
+      "loss": 0.4293,
+      "step": 86460
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.905581771032607e-05,
+      "loss": 0.461,
+      "step": 86470
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.897079454151257e-05,
+      "loss": 0.5229,
+      "step": 86480
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.888577137269906e-05,
+      "loss": 0.4567,
+      "step": 86490
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.880074820388556e-05,
+      "loss": 0.4024,
+      "step": 86500
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.871572503507207e-05,
+      "loss": 0.4394,
+      "step": 86510
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.863070186625857e-05,
+      "loss": 0.4883,
+      "step": 86520
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.854567869744506e-05,
+      "loss": 0.4484,
+      "step": 86530
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.846065552863156e-05,
+      "loss": 0.4517,
+      "step": 86540
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.837563235981806e-05,
+      "loss": 0.4786,
+      "step": 86550
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.829060919100455e-05,
+      "loss": 0.4629,
+      "step": 86560
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.820558602219105e-05,
+      "loss": 0.3818,
+      "step": 86570
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.812056285337754e-05,
+      "loss": 0.4741,
+      "step": 86580
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.803553968456404e-05,
+      "loss": 0.4347,
+      "step": 86590
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.795051651575055e-05,
+      "loss": 0.4563,
+      "step": 86600
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 6.786549334693705e-05,
+      "loss": 0.45,
+      "step": 86610
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.778047017812354e-05,
+      "loss": 0.473,
+      "step": 86620
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.769544700931004e-05,
+      "loss": 0.48,
+      "step": 86630
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.761042384049654e-05,
+      "loss": 0.4561,
+      "step": 86640
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.752540067168303e-05,
+      "loss": 0.417,
+      "step": 86650
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.744037750286953e-05,
+      "loss": 0.4482,
+      "step": 86660
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.735535433405603e-05,
+      "loss": 0.4926,
+      "step": 86670
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.727033116524252e-05,
+      "loss": 0.4653,
+      "step": 86680
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.718530799642902e-05,
+      "loss": 0.4022,
+      "step": 86690
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.710028482761553e-05,
+      "loss": 0.4558,
+      "step": 86700
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.701526165880203e-05,
+      "loss": 0.5264,
+      "step": 86710
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.693023848998852e-05,
+      "loss": 0.4777,
+      "step": 86720
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.684521532117502e-05,
+      "loss": 0.4545,
+      "step": 86730
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.676019215236152e-05,
+      "loss": 0.4598,
+      "step": 86740
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.667516898354801e-05,
+      "loss": 0.406,
+      "step": 86750
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.659014581473451e-05,
+      "loss": 0.4403,
+      "step": 86760
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.650512264592101e-05,
+      "loss": 0.4631,
+      "step": 86770
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.64200994771075e-05,
+      "loss": 0.4195,
+      "step": 86780
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.633507630829401e-05,
+      "loss": 0.4487,
+      "step": 86790
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.625005313948051e-05,
+      "loss": 0.4813,
+      "step": 86800
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.616502997066701e-05,
+      "loss": 0.4214,
+      "step": 86810
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.60800068018535e-05,
+      "loss": 0.4961,
+      "step": 86820
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.599498363304001e-05,
+      "loss": 0.4765,
+      "step": 86830
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.590996046422651e-05,
+      "loss": 0.5267,
+      "step": 86840
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.582493729541301e-05,
+      "loss": 0.4529,
+      "step": 86850
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.57399141265995e-05,
+      "loss": 0.4673,
+      "step": 86860
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.5654890957786e-05,
+      "loss": 0.4269,
+      "step": 86870
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.55698677889725e-05,
+      "loss": 0.4745,
+      "step": 86880
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.5484844620159e-05,
+      "loss": 0.3991,
+      "step": 86890
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.53998214513455e-05,
+      "loss": 0.4551,
+      "step": 86900
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.5314798282532e-05,
+      "loss": 0.492,
+      "step": 86910
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.52297751137185e-05,
+      "loss": 0.4565,
+      "step": 86920
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.5144751944905e-05,
+      "loss": 0.4375,
+      "step": 86930
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.505972877609149e-05,
+      "loss": 0.4317,
+      "step": 86940
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.497470560727799e-05,
+      "loss": 0.4893,
+      "step": 86950
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 6.488968243846448e-05,
+      "loss": 0.4515,
+      "step": 86960
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.480465926965098e-05,
+      "loss": 0.4781,
+      "step": 86970
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.471963610083748e-05,
+      "loss": 0.4033,
+      "step": 86980
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.463461293202397e-05,
+      "loss": 0.4406,
+      "step": 86990
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.454958976321048e-05,
+      "loss": 0.412,
+      "step": 87000
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.446456659439698e-05,
+      "loss": 0.4284,
+      "step": 87010
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.437954342558348e-05,
+      "loss": 0.5165,
+      "step": 87020
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.429452025676997e-05,
+      "loss": 0.4381,
+      "step": 87030
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.420949708795647e-05,
+      "loss": 0.4104,
+      "step": 87040
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.412447391914297e-05,
+      "loss": 0.4724,
+      "step": 87050
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.403945075032946e-05,
+      "loss": 0.4993,
+      "step": 87060
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.395442758151596e-05,
+      "loss": 0.492,
+      "step": 87070
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.386940441270246e-05,
+      "loss": 0.4933,
+      "step": 87080
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.378438124388895e-05,
+      "loss": 0.4736,
+      "step": 87090
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.369935807507546e-05,
+      "loss": 0.4507,
+      "step": 87100
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.361433490626196e-05,
+      "loss": 0.416,
+      "step": 87110
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.352931173744846e-05,
+      "loss": 0.4708,
+      "step": 87120
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.344428856863495e-05,
+      "loss": 0.5012,
+      "step": 87130
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.335926539982145e-05,
+      "loss": 0.466,
+      "step": 87140
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.327424223100795e-05,
+      "loss": 0.4881,
+      "step": 87150
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.318921906219444e-05,
+      "loss": 0.4363,
+      "step": 87160
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.310419589338094e-05,
+      "loss": 0.4321,
+      "step": 87170
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.301917272456744e-05,
+      "loss": 0.4228,
+      "step": 87180
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.293414955575393e-05,
+      "loss": 0.4904,
+      "step": 87190
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 6.284912638694044e-05,
+      "loss": 0.4665,
+      "step": 87200
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.276410321812694e-05,
+      "loss": 0.402,
+      "step": 87210
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.267908004931344e-05,
+      "loss": 0.4554,
+      "step": 87220
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.259405688049993e-05,
+      "loss": 0.4649,
+      "step": 87230
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.250903371168643e-05,
+      "loss": 0.4649,
+      "step": 87240
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.242401054287294e-05,
+      "loss": 0.4105,
+      "step": 87250
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.233898737405944e-05,
+      "loss": 0.4635,
+      "step": 87260
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.225396420524593e-05,
+      "loss": 0.4715,
+      "step": 87270
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.216894103643243e-05,
+      "loss": 0.4471,
+      "step": 87280
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.208391786761893e-05,
+      "loss": 0.4705,
+      "step": 87290
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.199889469880544e-05,
+      "loss": 0.4285,
+      "step": 87300
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.191387152999193e-05,
+      "loss": 0.4203,
+      "step": 87310
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.182884836117843e-05,
+      "loss": 0.4206,
+      "step": 87320
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.174382519236493e-05,
+      "loss": 0.4798,
+      "step": 87330
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.165880202355142e-05,
+      "loss": 0.4181,
+      "step": 87340
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.157377885473792e-05,
+      "loss": 0.4481,
+      "step": 87350
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.148875568592442e-05,
+      "loss": 0.4509,
+      "step": 87360
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.140373251711091e-05,
+      "loss": 0.4102,
+      "step": 87370
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.131870934829741e-05,
+      "loss": 0.4482,
+      "step": 87380
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.123368617948392e-05,
+      "loss": 0.4387,
+      "step": 87390
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.114866301067042e-05,
+      "loss": 0.4332,
+      "step": 87400
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.106363984185691e-05,
+      "loss": 0.4012,
+      "step": 87410
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.097861667304341e-05,
+      "loss": 0.466,
+      "step": 87420
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.089359350422991e-05,
+      "loss": 0.4482,
+      "step": 87430
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.0808570335416404e-05,
+      "loss": 0.4271,
+      "step": 87440
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.07235471666029e-05,
+      "loss": 0.3824,
+      "step": 87450
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.06385239977894e-05,
+      "loss": 0.4336,
+      "step": 87460
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.0553500828975894e-05,
+      "loss": 0.4595,
+      "step": 87470
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.046847766016239e-05,
+      "loss": 0.3969,
+      "step": 87480
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.03834544913489e-05,
+      "loss": 0.4846,
+      "step": 87490
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.02984313225354e-05,
+      "loss": 0.4275,
+      "step": 87500
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.0213408153721894e-05,
+      "loss": 0.4465,
+      "step": 87510
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.012838498490839e-05,
+      "loss": 0.4901,
+      "step": 87520
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 6.004336181609489e-05,
+      "loss": 0.4811,
+      "step": 87530
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.9958338647281384e-05,
+      "loss": 0.4411,
+      "step": 87540
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.987331547846788e-05,
+      "loss": 0.4819,
+      "step": 87550
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.9788292309654384e-05,
+      "loss": 0.4197,
+      "step": 87560
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.970326914084088e-05,
+      "loss": 0.4139,
+      "step": 87570
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.961824597202738e-05,
+      "loss": 0.4524,
+      "step": 87580
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.953322280321388e-05,
+      "loss": 0.4103,
+      "step": 87590
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.9448199634400384e-05,
+      "loss": 0.4683,
+      "step": 87600
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.936317646558688e-05,
+      "loss": 0.4384,
+      "step": 87610
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.927815329677338e-05,
+      "loss": 0.444,
+      "step": 87620
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.9193130127959874e-05,
+      "loss": 0.4111,
+      "step": 87630
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.910810695914637e-05,
+      "loss": 0.4113,
+      "step": 87640
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.902308379033287e-05,
+      "loss": 0.4386,
+      "step": 87650
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.8938060621519364e-05,
+      "loss": 0.4452,
+      "step": 87660
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.885303745270586e-05,
+      "loss": 0.4566,
+      "step": 87670
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.876801428389236e-05,
+      "loss": 0.4744,
+      "step": 87680
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.868299111507887e-05,
+      "loss": 0.4397,
+      "step": 87690
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.8597967946265364e-05,
+      "loss": 0.4864,
+      "step": 87700
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.851294477745186e-05,
+      "loss": 0.4214,
+      "step": 87710
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.842792160863836e-05,
+      "loss": 0.4789,
+      "step": 87720
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.8342898439824854e-05,
+      "loss": 0.4318,
+      "step": 87730
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.825787527101135e-05,
+      "loss": 0.3858,
+      "step": 87740
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.817285210219785e-05,
+      "loss": 0.4741,
+      "step": 87750
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.8087828933384344e-05,
+      "loss": 0.401,
+      "step": 87760
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.800280576457085e-05,
+      "loss": 0.4502,
+      "step": 87770
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.7917782595757344e-05,
+      "loss": 0.4226,
+      "step": 87780
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.783275942694385e-05,
+      "loss": 0.4862,
+      "step": 87790
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.7747736258130344e-05,
+      "loss": 0.4215,
+      "step": 87800
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.766271308931685e-05,
+      "loss": 0.3554,
+      "step": 87810
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.7577689920503344e-05,
+      "loss": 0.4045,
+      "step": 87820
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.749266675168984e-05,
+      "loss": 0.4534,
+      "step": 87830
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.740764358287634e-05,
+      "loss": 0.393,
+      "step": 87840
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.7322620414062834e-05,
+      "loss": 0.4279,
+      "step": 87850
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.723759724524933e-05,
+      "loss": 0.4272,
+      "step": 87860
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.715257407643583e-05,
+      "loss": 0.4947,
+      "step": 87870
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.7067550907622324e-05,
+      "loss": 0.4628,
+      "step": 87880
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.6982527738808834e-05,
+      "loss": 0.4713,
+      "step": 87890
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.689750456999533e-05,
+      "loss": 0.4872,
+      "step": 87900
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.681248140118183e-05,
+      "loss": 0.4994,
+      "step": 87910
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.6727458232368324e-05,
+      "loss": 0.4036,
+      "step": 87920
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.664243506355482e-05,
+      "loss": 0.4296,
+      "step": 87930
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.655741189474132e-05,
+      "loss": 0.4557,
+      "step": 87940
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.6472388725927814e-05,
+      "loss": 0.4519,
+      "step": 87950
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.638736555711431e-05,
+      "loss": 0.4318,
+      "step": 87960
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.630234238830081e-05,
+      "loss": 0.4882,
+      "step": 87970
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.621731921948731e-05,
+      "loss": 0.4537,
+      "step": 87980
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.6132296050673814e-05,
+      "loss": 0.4174,
+      "step": 87990
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.604727288186031e-05,
+      "loss": 0.4353,
+      "step": 88000
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.596224971304681e-05,
+      "loss": 0.4403,
+      "step": 88010
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.587722654423331e-05,
+      "loss": 0.5376,
+      "step": 88020
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.579220337541981e-05,
+      "loss": 0.4429,
+      "step": 88030
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.5707180206606304e-05,
+      "loss": 0.4183,
+      "step": 88040
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.56221570377928e-05,
+      "loss": 0.5015,
+      "step": 88050
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.55371338689793e-05,
+      "loss": 0.4888,
+      "step": 88060
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.5452110700165794e-05,
+      "loss": 0.4528,
+      "step": 88070
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.536708753135229e-05,
+      "loss": 0.4273,
+      "step": 88080
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.52820643625388e-05,
+      "loss": 0.4657,
+      "step": 88090
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.51970411937253e-05,
+      "loss": 0.4511,
+      "step": 88100
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.5112018024911794e-05,
+      "loss": 0.4044,
+      "step": 88110
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.502699485609829e-05,
+      "loss": 0.4821,
+      "step": 88120
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.494197168728479e-05,
+      "loss": 0.41,
+      "step": 88130
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.4856948518471284e-05,
+      "loss": 0.4377,
+      "step": 88140
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.477192534965778e-05,
+      "loss": 0.4575,
+      "step": 88150
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.468690218084428e-05,
+      "loss": 0.3842,
+      "step": 88160
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.4601879012030774e-05,
+      "loss": 0.4512,
+      "step": 88170
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.451685584321727e-05,
+      "loss": 0.4322,
+      "step": 88180
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.443183267440378e-05,
+      "loss": 0.3837,
+      "step": 88190
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.434680950559028e-05,
+      "loss": 0.4386,
+      "step": 88200
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.4261786336776774e-05,
+      "loss": 0.4778,
+      "step": 88210
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.417676316796327e-05,
+      "loss": 0.4156,
+      "step": 88220
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.4091739999149774e-05,
+      "loss": 0.4744,
+      "step": 88230
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.400671683033627e-05,
+      "loss": 0.4553,
+      "step": 88240
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.392169366152277e-05,
+      "loss": 0.473,
+      "step": 88250
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.3836670492709264e-05,
+      "loss": 0.4368,
+      "step": 88260
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.375164732389576e-05,
+      "loss": 0.4426,
+      "step": 88270
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.366662415508227e-05,
+      "loss": 0.4497,
+      "step": 88280
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.358160098626877e-05,
+      "loss": 0.4721,
+      "step": 88290
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.3496577817455264e-05,
+      "loss": 0.4009,
+      "step": 88300
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.341155464864176e-05,
+      "loss": 0.4359,
+      "step": 88310
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.332653147982826e-05,
+      "loss": 0.4431,
+      "step": 88320
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.3241508311014754e-05,
+      "loss": 0.4485,
+      "step": 88330
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.315648514220125e-05,
+      "loss": 0.4635,
+      "step": 88340
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.307146197338775e-05,
+      "loss": 0.461,
+      "step": 88350
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.2986438804574244e-05,
+      "loss": 0.4179,
+      "step": 88360
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.290141563576074e-05,
+      "loss": 0.4842,
+      "step": 88370
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.281639246694725e-05,
+      "loss": 0.4433,
+      "step": 88380
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.273136929813375e-05,
+      "loss": 0.4526,
+      "step": 88390
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.2646346129320244e-05,
+      "loss": 0.4244,
+      "step": 88400
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.256132296050674e-05,
+      "loss": 0.4065,
+      "step": 88410
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.247629979169324e-05,
+      "loss": 0.4816,
+      "step": 88420
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.2391276622879734e-05,
+      "loss": 0.4408,
+      "step": 88430
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.230625345406624e-05,
+      "loss": 0.4795,
+      "step": 88440
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.2221230285252734e-05,
+      "loss": 0.4336,
+      "step": 88450
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.213620711643923e-05,
+      "loss": 0.5003,
+      "step": 88460
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.205118394762573e-05,
+      "loss": 0.4595,
+      "step": 88470
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.196616077881224e-05,
+      "loss": 0.4299,
+      "step": 88480
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.1881137609998734e-05,
+      "loss": 0.4518,
+      "step": 88490
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.179611444118523e-05,
+      "loss": 0.4509,
+      "step": 88500
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.171109127237173e-05,
+      "loss": 0.4333,
+      "step": 88510
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.1626068103558224e-05,
+      "loss": 0.3992,
+      "step": 88520
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.154104493474472e-05,
+      "loss": 0.4516,
+      "step": 88530
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.145602176593122e-05,
+      "loss": 0.4306,
+      "step": 88540
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.1370998597117714e-05,
+      "loss": 0.4385,
+      "step": 88550
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.128597542830421e-05,
+      "loss": 0.4407,
+      "step": 88560
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.120095225949071e-05,
+      "loss": 0.4534,
+      "step": 88570
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.111592909067722e-05,
+      "loss": 0.4752,
+      "step": 88580
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.1030905921863714e-05,
+      "loss": 0.4621,
+      "step": 88590
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.094588275305021e-05,
+      "loss": 0.4584,
+      "step": 88600
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.086085958423671e-05,
+      "loss": 0.4215,
+      "step": 88610
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.0775836415423204e-05,
+      "loss": 0.4463,
+      "step": 88620
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.06908132466097e-05,
+      "loss": 0.4525,
+      "step": 88630
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.06057900777962e-05,
+      "loss": 0.4446,
+      "step": 88640
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.05207669089827e-05,
+      "loss": 0.4715,
+      "step": 88650
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.04357437401692e-05,
+      "loss": 0.4343,
+      "step": 88660
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.0350720571355694e-05,
+      "loss": 0.5009,
+      "step": 88670
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.02656974025422e-05,
+      "loss": 0.4706,
+      "step": 88680
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.01806742337287e-05,
+      "loss": 0.4797,
+      "step": 88690
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.00956510649152e-05,
+      "loss": 0.4247,
+      "step": 88700
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.0010627896101694e-05,
+      "loss": 0.4736,
+      "step": 88710
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.992560472728819e-05,
+      "loss": 0.4175,
+      "step": 88720
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.984058155847469e-05,
+      "loss": 0.4437,
+      "step": 88730
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.9755558389661184e-05,
+      "loss": 0.3876,
+      "step": 88740
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.967053522084769e-05,
+      "loss": 0.4759,
+      "step": 88750
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.9585512052034184e-05,
+      "loss": 0.4597,
+      "step": 88760
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.950048888322068e-05,
+      "loss": 0.4805,
+      "step": 88770
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.941546571440718e-05,
+      "loss": 0.4295,
+      "step": 88780
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.9330442545593674e-05,
+      "loss": 0.459,
+      "step": 88790
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.924541937678018e-05,
+      "loss": 0.4529,
+      "step": 88800
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.9160396207966674e-05,
+      "loss": 0.4757,
+      "step": 88810
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.907537303915317e-05,
+      "loss": 0.466,
+      "step": 88820
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.899034987033967e-05,
+      "loss": 0.4778,
+      "step": 88830
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.8905326701526163e-05,
+      "loss": 0.4981,
+      "step": 88840
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.882030353271267e-05,
+      "loss": 0.5009,
+      "step": 88850
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.8735280363899164e-05,
+      "loss": 0.4245,
+      "step": 88860
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.865025719508566e-05,
+      "loss": 0.4615,
+      "step": 88870
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.8565234026272164e-05,
+      "loss": 0.4282,
+      "step": 88880
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.848021085745866e-05,
+      "loss": 0.4231,
+      "step": 88890
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.8395187688645164e-05,
+      "loss": 0.4434,
+      "step": 88900
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.831016451983166e-05,
+      "loss": 0.4416,
+      "step": 88910
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.822514135101816e-05,
+      "loss": 0.4083,
+      "step": 88920
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.8140118182204654e-05,
+      "loss": 0.4831,
+      "step": 88930
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.805509501339115e-05,
+      "loss": 0.4515,
+      "step": 88940
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.7970071844577654e-05,
+      "loss": 0.4216,
+      "step": 88950
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.788504867576415e-05,
+      "loss": 0.4364,
+      "step": 88960
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.780002550695065e-05,
+      "loss": 0.437,
+      "step": 88970
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.7715002338137143e-05,
+      "loss": 0.4624,
+      "step": 88980
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.762997916932364e-05,
+      "loss": 0.4277,
+      "step": 88990
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.7544956000510144e-05,
+      "loss": 0.4086,
+      "step": 89000
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.745993283169664e-05,
+      "loss": 0.4304,
+      "step": 89010
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.737490966288314e-05,
+      "loss": 0.4594,
+      "step": 89020
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.7289886494069633e-05,
+      "loss": 0.4869,
+      "step": 89030
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.720486332525613e-05,
+      "loss": 0.4492,
+      "step": 89040
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.7119840156442633e-05,
+      "loss": 0.4663,
+      "step": 89050
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.703481698762913e-05,
+      "loss": 0.4536,
+      "step": 89060
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.694979381881563e-05,
+      "loss": 0.4361,
+      "step": 89070
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.686477065000212e-05,
+      "loss": 0.4795,
+      "step": 89080
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.677974748118863e-05,
+      "loss": 0.4333,
+      "step": 89090
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.669472431237512e-05,
+      "loss": 0.4658,
+      "step": 89100
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.660970114356163e-05,
+      "loss": 0.4165,
+      "step": 89110
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.6524677974748123e-05,
+      "loss": 0.4272,
+      "step": 89120
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.643965480593462e-05,
+      "loss": 0.4833,
+      "step": 89130
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.635463163712112e-05,
+      "loss": 0.4953,
+      "step": 89140
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.626960846830762e-05,
+      "loss": 0.5134,
+      "step": 89150
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.618458529949412e-05,
+      "loss": 0.4407,
+      "step": 89160
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.609956213068061e-05,
+      "loss": 0.4611,
+      "step": 89170
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.601453896186711e-05,
+      "loss": 0.4181,
+      "step": 89180
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.592951579305361e-05,
+      "loss": 0.4458,
+      "step": 89190
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.584449262424011e-05,
+      "loss": 0.4533,
+      "step": 89200
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.575946945542661e-05,
+      "loss": 0.4982,
+      "step": 89210
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.56744462866131e-05,
+      "loss": 0.4655,
+      "step": 89220
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.55894231177996e-05,
+      "loss": 0.4814,
+      "step": 89230
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.5504399948986097e-05,
+      "loss": 0.4569,
+      "step": 89240
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.54193767801726e-05,
+      "loss": 0.4751,
+      "step": 89250
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.53343536113591e-05,
+      "loss": 0.4838,
+      "step": 89260
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.524933044254559e-05,
+      "loss": 0.4328,
+      "step": 89270
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.516430727373209e-05,
+      "loss": 0.4446,
+      "step": 89280
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.5079284104918587e-05,
+      "loss": 0.4409,
+      "step": 89290
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.499426093610509e-05,
+      "loss": 0.4598,
+      "step": 89300
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.4909237767291587e-05,
+      "loss": 0.4217,
+      "step": 89310
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.482421459847809e-05,
+      "loss": 0.4811,
+      "step": 89320
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.4739191429664587e-05,
+      "loss": 0.4224,
+      "step": 89330
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.465416826085108e-05,
+      "loss": 0.3986,
+      "step": 89340
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.456914509203759e-05,
+      "loss": 0.4369,
+      "step": 89350
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.448412192322408e-05,
+      "loss": 0.4601,
+      "step": 89360
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.439909875441058e-05,
+      "loss": 0.4622,
+      "step": 89370
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.4314075585597077e-05,
+      "loss": 0.446,
+      "step": 89380
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.422905241678357e-05,
+      "loss": 0.4639,
+      "step": 89390
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.4144029247970077e-05,
+      "loss": 0.4878,
+      "step": 89400
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.405900607915657e-05,
+      "loss": 0.4571,
+      "step": 89410
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.397398291034307e-05,
+      "loss": 0.407,
+      "step": 89420
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.3888959741529566e-05,
+      "loss": 0.4951,
+      "step": 89430
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.380393657271607e-05,
+      "loss": 0.4823,
+      "step": 89440
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.3718913403902567e-05,
+      "loss": 0.4335,
+      "step": 89450
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.363389023508906e-05,
+      "loss": 0.4436,
+      "step": 89460
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.354886706627556e-05,
+      "loss": 0.4202,
+      "step": 89470
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.3463843897462056e-05,
+      "loss": 0.44,
+      "step": 89480
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.337882072864856e-05,
+      "loss": 0.3842,
+      "step": 89490
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.3293797559835056e-05,
+      "loss": 0.4738,
+      "step": 89500
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.320877439102155e-05,
+      "loss": 0.4514,
+      "step": 89510
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.312375122220805e-05,
+      "loss": 0.4182,
+      "step": 89520
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.303872805339455e-05,
+      "loss": 0.4914,
+      "step": 89530
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.295370488458105e-05,
+      "loss": 0.467,
+      "step": 89540
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.286868171576755e-05,
+      "loss": 0.4326,
+      "step": 89550
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.278365854695405e-05,
+      "loss": 0.4506,
+      "step": 89560
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.2698635378140546e-05,
+      "loss": 0.4917,
+      "step": 89570
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.261361220932704e-05,
+      "loss": 0.4105,
+      "step": 89580
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.2528589040513547e-05,
+      "loss": 0.4529,
+      "step": 89590
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.244356587170004e-05,
+      "loss": 0.4513,
+      "step": 89600
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.235854270288654e-05,
+      "loss": 0.4706,
+      "step": 89610
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.2273519534073036e-05,
+      "loss": 0.4839,
+      "step": 89620
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.218849636525953e-05,
+      "loss": 0.4496,
+      "step": 89630
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.2103473196446036e-05,
+      "loss": 0.5001,
+      "step": 89640
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.201845002763253e-05,
+      "loss": 0.4301,
+      "step": 89650
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.193342685881903e-05,
+      "loss": 0.4111,
+      "step": 89660
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.1848403690005526e-05,
+      "loss": 0.4217,
+      "step": 89670
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.176338052119202e-05,
+      "loss": 0.4323,
+      "step": 89680
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.1678357352378526e-05,
+      "loss": 0.4092,
+      "step": 89690
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.159333418356502e-05,
+      "loss": 0.4181,
+      "step": 89700
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.150831101475152e-05,
+      "loss": 0.4389,
+      "step": 89710
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.1423287845938016e-05,
+      "loss": 0.453,
+      "step": 89720
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.133826467712452e-05,
+      "loss": 0.4532,
+      "step": 89730
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.1253241508311016e-05,
+      "loss": 0.4446,
+      "step": 89740
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.116821833949751e-05,
+      "loss": 0.4578,
+      "step": 89750
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.1083195170684016e-05,
+      "loss": 0.4206,
+      "step": 89760
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.099817200187051e-05,
+      "loss": 0.4509,
+      "step": 89770
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.091314883305701e-05,
+      "loss": 0.4373,
+      "step": 89780
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.082812566424351e-05,
+      "loss": 0.4377,
+      "step": 89790
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.074310249543001e-05,
+      "loss": 0.3976,
+      "step": 89800
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.0658079326616506e-05,
+      "loss": 0.4472,
+      "step": 89810
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.0573056157803e-05,
+      "loss": 0.4511,
+      "step": 89820
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.04880329889895e-05,
+      "loss": 0.4621,
+      "step": 89830
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.0403009820176e-05,
+      "loss": 0.4834,
+      "step": 89840
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.03179866513625e-05,
+      "loss": 0.4437,
+      "step": 89850
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.0232963482548996e-05,
+      "loss": 0.4512,
+      "step": 89860
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.014794031373549e-05,
+      "loss": 0.4158,
+      "step": 89870
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.006291714492199e-05,
+      "loss": 0.4457,
+      "step": 89880
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.997789397610849e-05,
+      "loss": 0.4664,
+      "step": 89890
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.989287080729499e-05,
+      "loss": 0.4054,
+      "step": 89900
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.9807847638481486e-05,
+      "loss": 0.4678,
+      "step": 89910
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.972282446966798e-05,
+      "loss": 0.447,
+      "step": 89920
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.963780130085448e-05,
+      "loss": 0.451,
+      "step": 89930
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.955277813204098e-05,
+      "loss": 0.4579,
+      "step": 89940
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.946775496322748e-05,
+      "loss": 0.4598,
+      "step": 89950
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.9382731794413976e-05,
+      "loss": 0.4447,
+      "step": 89960
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.929770862560048e-05,
+      "loss": 0.493,
+      "step": 89970
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.9212685456786976e-05,
+      "loss": 0.4051,
+      "step": 89980
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.912766228797348e-05,
+      "loss": 0.4513,
+      "step": 89990
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.9042639119159976e-05,
+      "loss": 0.428,
+      "step": 90000
+    },
+    {
+      "epoch": 7.61,
+      "eval_loss": 0.5812339186668396,
+      "eval_runtime": 312.2316,
+      "eval_samples_per_second": 16.83,
+      "eval_steps_per_second": 2.104,
+      "step": 90000
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.895761595034647e-05,
+      "loss": 0.5039,
+      "step": 90010
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.887259278153297e-05,
+      "loss": 0.419,
+      "step": 90020
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.8787569612719466e-05,
+      "loss": 0.4615,
+      "step": 90030
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.870254644390597e-05,
+      "loss": 0.4258,
+      "step": 90040
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.8617523275092466e-05,
+      "loss": 0.4916,
+      "step": 90050
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.853250010627896e-05,
+      "loss": 0.4471,
+      "step": 90060
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.844747693746546e-05,
+      "loss": 0.4613,
+      "step": 90070
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.8362453768651956e-05,
+      "loss": 0.472,
+      "step": 90080
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.827743059983846e-05,
+      "loss": 0.4841,
+      "step": 90090
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.8192407431024956e-05,
+      "loss": 0.5057,
+      "step": 90100
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.810738426221145e-05,
+      "loss": 0.4485,
+      "step": 90110
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.802236109339795e-05,
+      "loss": 0.4429,
+      "step": 90120
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.7937337924584446e-05,
+      "loss": 0.4231,
+      "step": 90130
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.785231475577095e-05,
+      "loss": 0.4345,
+      "step": 90140
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.7767291586957446e-05,
+      "loss": 0.4329,
+      "step": 90150
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.768226841814394e-05,
+      "loss": 0.4208,
+      "step": 90160
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.7597245249330446e-05,
+      "loss": 0.4625,
+      "step": 90170
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.751222208051694e-05,
+      "loss": 0.4833,
+      "step": 90180
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.742719891170344e-05,
+      "loss": 0.4481,
+      "step": 90190
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.734217574288994e-05,
+      "loss": 0.4855,
+      "step": 90200
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.725715257407644e-05,
+      "loss": 0.4334,
+      "step": 90210
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.7172129405262936e-05,
+      "loss": 0.4445,
+      "step": 90220
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.708710623644943e-05,
+      "loss": 0.47,
+      "step": 90230
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.7002083067635936e-05,
+      "loss": 0.4672,
+      "step": 90240
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.691705989882243e-05,
+      "loss": 0.4539,
+      "step": 90250
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.683203673000893e-05,
+      "loss": 0.4332,
+      "step": 90260
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.6747013561195426e-05,
+      "loss": 0.4296,
+      "step": 90270
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.666199039238193e-05,
+      "loss": 0.436,
+      "step": 90280
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.6576967223568426e-05,
+      "loss": 0.4595,
+      "step": 90290
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.649194405475492e-05,
+      "loss": 0.489,
+      "step": 90300
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.640692088594142e-05,
+      "loss": 0.419,
+      "step": 90310
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.6321897717127916e-05,
+      "loss": 0.4138,
+      "step": 90320
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.623687454831442e-05,
+      "loss": 0.4594,
+      "step": 90330
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.6151851379500916e-05,
+      "loss": 0.4871,
+      "step": 90340
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.606682821068741e-05,
+      "loss": 0.4954,
+      "step": 90350
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.598180504187391e-05,
+      "loss": 0.4685,
+      "step": 90360
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.5896781873060406e-05,
+      "loss": 0.4113,
+      "step": 90370
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.581175870424691e-05,
+      "loss": 0.494,
+      "step": 90380
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.5726735535433406e-05,
+      "loss": 0.4946,
+      "step": 90390
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.564171236661991e-05,
+      "loss": 0.4551,
+      "step": 90400
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.5556689197806406e-05,
+      "loss": 0.4217,
+      "step": 90410
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.54716660289929e-05,
+      "loss": 0.4708,
+      "step": 90420
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.5386642860179406e-05,
+      "loss": 0.464,
+      "step": 90430
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.53016196913659e-05,
+      "loss": 0.4615,
+      "step": 90440
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.52165965225524e-05,
+      "loss": 0.4497,
+      "step": 90450
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.5131573353738896e-05,
+      "loss": 0.4206,
+      "step": 90460
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.504655018492539e-05,
+      "loss": 0.4623,
+      "step": 90470
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.4961527016111896e-05,
+      "loss": 0.4253,
+      "step": 90480
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.487650384729839e-05,
+      "loss": 0.4247,
+      "step": 90490
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.479148067848489e-05,
+      "loss": 0.4231,
+      "step": 90500
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.4706457509671386e-05,
+      "loss": 0.446,
+      "step": 90510
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.462143434085788e-05,
+      "loss": 0.4614,
+      "step": 90520
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.4536411172044386e-05,
+      "loss": 0.5001,
+      "step": 90530
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.445138800323088e-05,
+      "loss": 0.3757,
+      "step": 90540
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.436636483441738e-05,
+      "loss": 0.4759,
+      "step": 90550
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.4281341665603876e-05,
+      "loss": 0.4265,
+      "step": 90560
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.419631849679037e-05,
+      "loss": 0.4474,
+      "step": 90570
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.4111295327976876e-05,
+      "loss": 0.4566,
+      "step": 90580
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.402627215916337e-05,
+      "loss": 0.3826,
+      "step": 90590
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.394124899034987e-05,
+      "loss": 0.4145,
+      "step": 90600
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.385622582153637e-05,
+      "loss": 0.5058,
+      "step": 90610
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.377120265272287e-05,
+      "loss": 0.4206,
+      "step": 90620
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.368617948390937e-05,
+      "loss": 0.4853,
+      "step": 90630
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.360115631509587e-05,
+      "loss": 0.4493,
+      "step": 90640
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.3516133146282366e-05,
+      "loss": 0.4228,
+      "step": 90650
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.343110997746886e-05,
+      "loss": 0.3622,
+      "step": 90660
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.334608680865536e-05,
+      "loss": 0.3963,
+      "step": 90670
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.326106363984186e-05,
+      "loss": 0.4307,
+      "step": 90680
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.317604047102836e-05,
+      "loss": 0.4493,
+      "step": 90690
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.3091017302214856e-05,
+      "loss": 0.4477,
+      "step": 90700
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.300599413340135e-05,
+      "loss": 0.4858,
+      "step": 90710
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.292097096458785e-05,
+      "loss": 0.4645,
+      "step": 90720
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.283594779577435e-05,
+      "loss": 0.4448,
+      "step": 90730
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.275092462696085e-05,
+      "loss": 0.3821,
+      "step": 90740
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.2665901458147346e-05,
+      "loss": 0.46,
+      "step": 90750
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.258087828933384e-05,
+      "loss": 0.434,
+      "step": 90760
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.249585512052034e-05,
+      "loss": 0.4732,
+      "step": 90770
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.241083195170684e-05,
+      "loss": 0.4581,
+      "step": 90780
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.232580878289334e-05,
+      "loss": 0.4878,
+      "step": 90790
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.2240785614079836e-05,
+      "loss": 0.4628,
+      "step": 90800
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.215576244526633e-05,
+      "loss": 0.3924,
+      "step": 90810
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.2070739276452836e-05,
+      "loss": 0.4586,
+      "step": 90820
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.198571610763933e-05,
+      "loss": 0.4426,
+      "step": 90830
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.1900692938825836e-05,
+      "loss": 0.4388,
+      "step": 90840
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.181566977001233e-05,
+      "loss": 0.4102,
+      "step": 90850
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.173064660119883e-05,
+      "loss": 0.4691,
+      "step": 90860
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.1645623432385326e-05,
+      "loss": 0.4025,
+      "step": 90870
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.156060026357183e-05,
+      "loss": 0.4122,
+      "step": 90880
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.1475577094758326e-05,
+      "loss": 0.4705,
+      "step": 90890
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.139055392594482e-05,
+      "loss": 0.4633,
+      "step": 90900
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.130553075713132e-05,
+      "loss": 0.4933,
+      "step": 90910
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.1220507588317816e-05,
+      "loss": 0.4467,
+      "step": 90920
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.113548441950432e-05,
+      "loss": 0.5137,
+      "step": 90930
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.1050461250690816e-05,
+      "loss": 0.4235,
+      "step": 90940
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.096543808187731e-05,
+      "loss": 0.4016,
+      "step": 90950
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.088041491306381e-05,
+      "loss": 0.4205,
+      "step": 90960
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.0795391744250306e-05,
+      "loss": 0.4411,
+      "step": 90970
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.071036857543681e-05,
+      "loss": 0.4699,
+      "step": 90980
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.0625345406623306e-05,
+      "loss": 0.4288,
+      "step": 90990
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.05403222378098e-05,
+      "loss": 0.4449,
+      "step": 91000
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.0455299068996302e-05,
+      "loss": 0.4194,
+      "step": 91010
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.03702759001828e-05,
+      "loss": 0.4324,
+      "step": 91020
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.0285252731369302e-05,
+      "loss": 0.4811,
+      "step": 91030
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.02002295625558e-05,
+      "loss": 0.4255,
+      "step": 91040
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.0115206393742296e-05,
+      "loss": 0.4467,
+      "step": 91050
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.0030183224928792e-05,
+      "loss": 0.4008,
+      "step": 91060
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.9945160056115296e-05,
+      "loss": 0.4343,
+      "step": 91070
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.9860136887301792e-05,
+      "loss": 0.5129,
+      "step": 91080
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.977511371848829e-05,
+      "loss": 0.5039,
+      "step": 91090
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.969009054967479e-05,
+      "loss": 0.4139,
+      "step": 91100
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.9605067380861285e-05,
+      "loss": 0.4515,
+      "step": 91110
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.9520044212047786e-05,
+      "loss": 0.4817,
+      "step": 91120
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.9435021043234286e-05,
+      "loss": 0.4622,
+      "step": 91130
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.9349997874420782e-05,
+      "loss": 0.4624,
+      "step": 91140
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.926497470560728e-05,
+      "loss": 0.4468,
+      "step": 91150
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.9179951536793775e-05,
+      "loss": 0.4585,
+      "step": 91160
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.909492836798028e-05,
+      "loss": 0.4352,
+      "step": 91170
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.9009905199166775e-05,
+      "loss": 0.4525,
+      "step": 91180
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.8924882030353272e-05,
+      "loss": 0.5179,
+      "step": 91190
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.883985886153977e-05,
+      "loss": 0.4474,
+      "step": 91200
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.875483569272627e-05,
+      "loss": 0.4617,
+      "step": 91210
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.866981252391277e-05,
+      "loss": 0.4602,
+      "step": 91220
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.858478935509927e-05,
+      "loss": 0.4188,
+      "step": 91230
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.8499766186285765e-05,
+      "loss": 0.404,
+      "step": 91240
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.8414743017472262e-05,
+      "loss": 0.4137,
+      "step": 91250
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.832971984865876e-05,
+      "loss": 0.4563,
+      "step": 91260
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.8244696679845262e-05,
+      "loss": 0.4056,
+      "step": 91270
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.815967351103176e-05,
+      "loss": 0.4268,
+      "step": 91280
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.8074650342218255e-05,
+      "loss": 0.4341,
+      "step": 91290
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.7989627173404752e-05,
+      "loss": 0.5105,
+      "step": 91300
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.7904604004591252e-05,
+      "loss": 0.4458,
+      "step": 91310
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.7819580835777752e-05,
+      "loss": 0.4883,
+      "step": 91320
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.7734557666964252e-05,
+      "loss": 0.4161,
+      "step": 91330
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.764953449815075e-05,
+      "loss": 0.3872,
+      "step": 91340
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7564511329337245e-05,
+      "loss": 0.446,
+      "step": 91350
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7479488160523742e-05,
+      "loss": 0.4916,
+      "step": 91360
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7394464991710245e-05,
+      "loss": 0.4724,
+      "step": 91370
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7309441822896742e-05,
+      "loss": 0.47,
+      "step": 91380
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.722441865408324e-05,
+      "loss": 0.4271,
+      "step": 91390
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7139395485269735e-05,
+      "loss": 0.467,
+      "step": 91400
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.7054372316456232e-05,
+      "loss": 0.4468,
+      "step": 91410
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.6969349147642735e-05,
+      "loss": 0.4104,
+      "step": 91420
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.6884325978829232e-05,
+      "loss": 0.4186,
+      "step": 91430
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.6799302810015732e-05,
+      "loss": 0.4838,
+      "step": 91440
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.671427964120223e-05,
+      "loss": 0.452,
+      "step": 91450
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.6629256472388725e-05,
+      "loss": 0.5001,
+      "step": 91460
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.654423330357523e-05,
+      "loss": 0.4673,
+      "step": 91470
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.6459210134761725e-05,
+      "loss": 0.4704,
+      "step": 91480
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.6374186965948222e-05,
+      "loss": 0.4655,
+      "step": 91490
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.628916379713472e-05,
+      "loss": 0.4463,
+      "step": 91500
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.6204140628321215e-05,
+      "loss": 0.4236,
+      "step": 91510
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.611911745950772e-05,
+      "loss": 0.461,
+      "step": 91520
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.6034094290694215e-05,
+      "loss": 0.4819,
+      "step": 91530
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.5949071121880715e-05,
+      "loss": 0.4186,
+      "step": 91540
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.5864047953067212e-05,
+      "loss": 0.399,
+      "step": 91550
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.577902478425371e-05,
+      "loss": 0.4068,
+      "step": 91560
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.5694001615440212e-05,
+      "loss": 0.4131,
+      "step": 91570
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.560897844662671e-05,
+      "loss": 0.4822,
+      "step": 91580
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.5523955277813205e-05,
+      "loss": 0.3793,
+      "step": 91590
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.5438932108999702e-05,
+      "loss": 0.43,
+      "step": 91600
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.53539089401862e-05,
+      "loss": 0.4477,
+      "step": 91610
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.5268885771372702e-05,
+      "loss": 0.4141,
+      "step": 91620
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.51838626025592e-05,
+      "loss": 0.4897,
+      "step": 91630
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.5098839433745695e-05,
+      "loss": 0.4204,
+      "step": 91640
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.5013816264932195e-05,
+      "loss": 0.3975,
+      "step": 91650
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4928793096118695e-05,
+      "loss": 0.4316,
+      "step": 91660
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4843769927305192e-05,
+      "loss": 0.4465,
+      "step": 91670
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4758746758491692e-05,
+      "loss": 0.4376,
+      "step": 91680
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.467372358967819e-05,
+      "loss": 0.4633,
+      "step": 91690
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.458870042086469e-05,
+      "loss": 0.4274,
+      "step": 91700
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4503677252051185e-05,
+      "loss": 0.3883,
+      "step": 91710
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4418654083237682e-05,
+      "loss": 0.4746,
+      "step": 91720
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4333630914424182e-05,
+      "loss": 0.436,
+      "step": 91730
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.424860774561068e-05,
+      "loss": 0.4635,
+      "step": 91740
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.416358457679718e-05,
+      "loss": 0.4329,
+      "step": 91750
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.407856140798368e-05,
+      "loss": 0.4823,
+      "step": 91760
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.3993538239170175e-05,
+      "loss": 0.4496,
+      "step": 91770
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.3908515070356675e-05,
+      "loss": 0.5018,
+      "step": 91780
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.3823491901543172e-05,
+      "loss": 0.4434,
+      "step": 91790
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.3738468732729672e-05,
+      "loss": 0.4297,
+      "step": 91800
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.365344556391617e-05,
+      "loss": 0.4445,
+      "step": 91810
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.3568422395102665e-05,
+      "loss": 0.4475,
+      "step": 91820
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.3483399226289165e-05,
+      "loss": 0.462,
+      "step": 91830
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.339837605747566e-05,
+      "loss": 0.4726,
+      "step": 91840
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.3313352888662162e-05,
+      "loss": 0.4719,
+      "step": 91850
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.322832971984866e-05,
+      "loss": 0.4462,
+      "step": 91860
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.314330655103516e-05,
+      "loss": 0.4292,
+      "step": 91870
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.305828338222166e-05,
+      "loss": 0.4535,
+      "step": 91880
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.2973260213408155e-05,
+      "loss": 0.4316,
+      "step": 91890
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.2888237044594655e-05,
+      "loss": 0.4314,
+      "step": 91900
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.280321387578115e-05,
+      "loss": 0.4546,
+      "step": 91910
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.2718190706967648e-05,
+      "loss": 0.4555,
+      "step": 91920
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.263316753815415e-05,
+      "loss": 0.4479,
+      "step": 91930
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.2548144369340645e-05,
+      "loss": 0.444,
+      "step": 91940
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.2463121200527145e-05,
+      "loss": 0.5207,
+      "step": 91950
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.237809803171364e-05,
+      "loss": 0.4537,
+      "step": 91960
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.229307486290014e-05,
+      "loss": 0.4183,
+      "step": 91970
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.220805169408664e-05,
+      "loss": 0.435,
+      "step": 91980
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.2123028525273138e-05,
+      "loss": 0.3769,
+      "step": 91990
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.203800535645964e-05,
+      "loss": 0.4553,
+      "step": 92000
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.1952982187646135e-05,
+      "loss": 0.4257,
+      "step": 92010
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.186795901883263e-05,
+      "loss": 0.4533,
+      "step": 92020
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.178293585001913e-05,
+      "loss": 0.4039,
+      "step": 92030
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.1697912681205628e-05,
+      "loss": 0.4521,
+      "step": 92040
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.1612889512392128e-05,
+      "loss": 0.4754,
+      "step": 92050
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.1527866343578625e-05,
+      "loss": 0.4145,
+      "step": 92060
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.144284317476512e-05,
+      "loss": 0.4907,
+      "step": 92070
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.135782000595162e-05,
+      "loss": 0.4347,
+      "step": 92080
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.127279683713812e-05,
+      "loss": 0.4518,
+      "step": 92090
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.118777366832462e-05,
+      "loss": 0.4134,
+      "step": 92100
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.1102750499511118e-05,
+      "loss": 0.4322,
+      "step": 92110
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.1017727330697618e-05,
+      "loss": 0.4817,
+      "step": 92120
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.0932704161884115e-05,
+      "loss": 0.4051,
+      "step": 92130
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.084768099307061e-05,
+      "loss": 0.4339,
+      "step": 92140
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.076265782425711e-05,
+      "loss": 0.4389,
+      "step": 92150
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.0677634655443608e-05,
+      "loss": 0.4257,
+      "step": 92160
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.0592611486630108e-05,
+      "loss": 0.4302,
+      "step": 92170
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.0507588317816605e-05,
+      "loss": 0.432,
+      "step": 92180
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.0422565149003105e-05,
+      "loss": 0.469,
+      "step": 92190
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.0337541980189605e-05,
+      "loss": 0.4519,
+      "step": 92200
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.02525188113761e-05,
+      "loss": 0.4494,
+      "step": 92210
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.01674956425626e-05,
+      "loss": 0.4513,
+      "step": 92220
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.0082472473749098e-05,
+      "loss": 0.4396,
+      "step": 92230
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.9997449304935595e-05,
+      "loss": 0.4307,
+      "step": 92240
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.9912426136122095e-05,
+      "loss": 0.413,
+      "step": 92250
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.982740296730859e-05,
+      "loss": 0.4465,
+      "step": 92260
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.974237979849509e-05,
+      "loss": 0.4421,
+      "step": 92270
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.9657356629681588e-05,
+      "loss": 0.4653,
+      "step": 92280
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.9572333460868085e-05,
+      "loss": 0.4793,
+      "step": 92290
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.9487310292054585e-05,
+      "loss": 0.476,
+      "step": 92300
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.9402287123241085e-05,
+      "loss": 0.4399,
+      "step": 92310
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.9317263954427585e-05,
+      "loss": 0.4395,
+      "step": 92320
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.923224078561408e-05,
+      "loss": 0.4225,
+      "step": 92330
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.9147217616800578e-05,
+      "loss": 0.4412,
+      "step": 92340
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.9062194447987078e-05,
+      "loss": 0.4224,
+      "step": 92350
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.8977171279173575e-05,
+      "loss": 0.4483,
+      "step": 92360
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.8892148110360075e-05,
+      "loss": 0.4832,
+      "step": 92370
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.880712494154657e-05,
+      "loss": 0.4296,
+      "step": 92380
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.8722101772733068e-05,
+      "loss": 0.4542,
+      "step": 92390
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.8637078603919568e-05,
+      "loss": 0.4514,
+      "step": 92400
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.8552055435106068e-05,
+      "loss": 0.491,
+      "step": 92410
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.8467032266292568e-05,
+      "loss": 0.4434,
+      "step": 92420
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.8382009097479065e-05,
+      "loss": 0.4471,
+      "step": 92430
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.829698592866556e-05,
+      "loss": 0.4569,
+      "step": 92440
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.821196275985206e-05,
+      "loss": 0.4997,
+      "step": 92450
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.8126939591038558e-05,
+      "loss": 0.4747,
+      "step": 92460
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.8041916422225058e-05,
+      "loss": 0.424,
+      "step": 92470
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.7956893253411555e-05,
+      "loss": 0.4604,
+      "step": 92480
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.787187008459805e-05,
+      "loss": 0.5319,
+      "step": 92490
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.778684691578455e-05,
+      "loss": 0.4582,
+      "step": 92500
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.770182374697105e-05,
+      "loss": 0.4661,
+      "step": 92510
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.7616800578157548e-05,
+      "loss": 0.4463,
+      "step": 92520
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.7531777409344048e-05,
+      "loss": 0.4981,
+      "step": 92530
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.7446754240530548e-05,
+      "loss": 0.4238,
+      "step": 92540
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.7361731071717045e-05,
+      "loss": 0.474,
+      "step": 92550
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.727670790290354e-05,
+      "loss": 0.4556,
+      "step": 92560
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.719168473409004e-05,
+      "loss": 0.4743,
+      "step": 92570
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.7106661565276538e-05,
+      "loss": 0.4296,
+      "step": 92580
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.7021638396463038e-05,
+      "loss": 0.4855,
+      "step": 92590
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.6936615227649535e-05,
+      "loss": 0.4348,
+      "step": 92600
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.685159205883603e-05,
+      "loss": 0.5032,
+      "step": 92610
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.676656889002253e-05,
+      "loss": 0.4129,
+      "step": 92620
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.668154572120903e-05,
+      "loss": 0.4477,
+      "step": 92630
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.659652255239553e-05,
+      "loss": 0.4936,
+      "step": 92640
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.6511499383582028e-05,
+      "loss": 0.4647,
+      "step": 92650
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.6426476214768525e-05,
+      "loss": 0.4325,
+      "step": 92660
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.6341453045955025e-05,
+      "loss": 0.4381,
+      "step": 92670
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.625642987714152e-05,
+      "loss": 0.4567,
+      "step": 92680
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.617140670832802e-05,
+      "loss": 0.4729,
+      "step": 92690
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.6086383539514518e-05,
+      "loss": 0.436,
+      "step": 92700
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.6001360370701014e-05,
+      "loss": 0.4286,
+      "step": 92710
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.5916337201887514e-05,
+      "loss": 0.5115,
+      "step": 92720
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.5831314033074015e-05,
+      "loss": 0.4121,
+      "step": 92730
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.574629086426051e-05,
+      "loss": 0.4712,
+      "step": 92740
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.566126769544701e-05,
+      "loss": 0.483,
+      "step": 92750
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.5576244526633508e-05,
+      "loss": 0.4051,
+      "step": 92760
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.5491221357820008e-05,
+      "loss": 0.5133,
+      "step": 92770
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.5406198189006504e-05,
+      "loss": 0.4232,
+      "step": 92780
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.5321175020193004e-05,
+      "loss": 0.4225,
+      "step": 92790
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.5236151851379501e-05,
+      "loss": 0.4527,
+      "step": 92800
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.5151128682566e-05,
+      "loss": 0.4292,
+      "step": 92810
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.50661055137525e-05,
+      "loss": 0.4452,
+      "step": 92820
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4981082344938996e-05,
+      "loss": 0.419,
+      "step": 92830
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4896059176125496e-05,
+      "loss": 0.4765,
+      "step": 92840
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4811036007311993e-05,
+      "loss": 0.4507,
+      "step": 92850
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.4726012838498491e-05,
+      "loss": 0.4744,
+      "step": 92860
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.464098966968499e-05,
+      "loss": 0.4259,
+      "step": 92870
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4555966500871488e-05,
+      "loss": 0.4171,
+      "step": 92880
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4470943332057988e-05,
+      "loss": 0.455,
+      "step": 92890
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4385920163244484e-05,
+      "loss": 0.4825,
+      "step": 92900
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4300896994430981e-05,
+      "loss": 0.4297,
+      "step": 92910
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.4215873825617481e-05,
+      "loss": 0.5146,
+      "step": 92920
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.413085065680398e-05,
+      "loss": 0.5054,
+      "step": 92930
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.404582748799048e-05,
+      "loss": 0.4093,
+      "step": 92940
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.3960804319176976e-05,
+      "loss": 0.4726,
+      "step": 92950
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.3875781150363476e-05,
+      "loss": 0.5176,
+      "step": 92960
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.3790757981549973e-05,
+      "loss": 0.4894,
+      "step": 92970
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.3705734812736471e-05,
+      "loss": 0.4007,
+      "step": 92980
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.3620711643922971e-05,
+      "loss": 0.446,
+      "step": 92990
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3535688475109468e-05,
+      "loss": 0.4136,
+      "step": 93000
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3450665306295968e-05,
+      "loss": 0.4142,
+      "step": 93010
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3365642137482464e-05,
+      "loss": 0.4531,
+      "step": 93020
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3280618968668963e-05,
+      "loss": 0.4654,
+      "step": 93030
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.3195595799855463e-05,
+      "loss": 0.4914,
+      "step": 93040
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.311057263104196e-05,
+      "loss": 0.4601,
+      "step": 93050
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.302554946222846e-05,
+      "loss": 0.4513,
+      "step": 93060
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.2940526293414956e-05,
+      "loss": 0.5003,
+      "step": 93070
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.2855503124601454e-05,
+      "loss": 0.4727,
+      "step": 93080
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.2770479955787953e-05,
+      "loss": 0.3855,
+      "step": 93090
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.2685456786974451e-05,
+      "loss": 0.4311,
+      "step": 93100
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.2600433618160951e-05,
+      "loss": 0.4019,
+      "step": 93110
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.2515410449347448e-05,
+      "loss": 0.462,
+      "step": 93120
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.2430387280533946e-05,
+      "loss": 0.4569,
+      "step": 93130
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.2345364111720444e-05,
+      "loss": 0.4496,
+      "step": 93140
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.2260340942906943e-05,
+      "loss": 0.4979,
+      "step": 93150
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.2175317774093441e-05,
+      "loss": 0.4919,
+      "step": 93160
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.209029460527994e-05,
+      "loss": 0.4223,
+      "step": 93170
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.2005271436466438e-05,
+      "loss": 0.4325,
+      "step": 93180
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.1920248267652936e-05,
+      "loss": 0.4898,
+      "step": 93190
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.1835225098839434e-05,
+      "loss": 0.4559,
+      "step": 93200
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.1750201930025933e-05,
+      "loss": 0.4137,
+      "step": 93210
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.1665178761212431e-05,
+      "loss": 0.4241,
+      "step": 93220
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.158015559239893e-05,
+      "loss": 0.4901,
+      "step": 93230
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1495132423585428e-05,
+      "loss": 0.4256,
+      "step": 93240
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1410109254771926e-05,
+      "loss": 0.4382,
+      "step": 93250
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1325086085958424e-05,
+      "loss": 0.4456,
+      "step": 93260
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1240062917144922e-05,
+      "loss": 0.4192,
+      "step": 93270
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.115503974833142e-05,
+      "loss": 0.3933,
+      "step": 93280
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1070016579517919e-05,
+      "loss": 0.4474,
+      "step": 93290
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0984993410704417e-05,
+      "loss": 0.4111,
+      "step": 93300
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0899970241890916e-05,
+      "loss": 0.441,
+      "step": 93310
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0814947073077414e-05,
+      "loss": 0.4753,
+      "step": 93320
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0729923904263912e-05,
+      "loss": 0.4395,
+      "step": 93330
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.064490073545041e-05,
+      "loss": 0.4644,
+      "step": 93340
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0559877566636909e-05,
+      "loss": 0.4243,
+      "step": 93350
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0474854397823407e-05,
+      "loss": 0.4775,
+      "step": 93360
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0389831229009906e-05,
+      "loss": 0.4815,
+      "step": 93370
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0304808060196404e-05,
+      "loss": 0.4055,
+      "step": 93380
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0219784891382902e-05,
+      "loss": 0.4964,
+      "step": 93390
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.01347617225694e-05,
+      "loss": 0.4924,
+      "step": 93400
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0049738553755899e-05,
+      "loss": 0.4715,
+      "step": 93410
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.964715384942397e-06,
+      "loss": 0.4237,
+      "step": 93420
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.879692216128896e-06,
+      "loss": 0.4025,
+      "step": 93430
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.794669047315394e-06,
+      "loss": 0.4407,
+      "step": 93440
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.709645878501892e-06,
+      "loss": 0.3996,
+      "step": 93450
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.62462270968839e-06,
+      "loss": 0.3875,
+      "step": 93460
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.539599540874889e-06,
+      "loss": 0.4809,
+      "step": 93470
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.454576372061387e-06,
+      "loss": 0.5201,
+      "step": 93480
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.369553203247886e-06,
+      "loss": 0.4683,
+      "step": 93490
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.284530034434384e-06,
+      "loss": 0.4493,
+      "step": 93500
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.199506865620882e-06,
+      "loss": 0.4644,
+      "step": 93510
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.11448369680738e-06,
+      "loss": 0.4301,
+      "step": 93520
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.029460527993879e-06,
+      "loss": 0.4659,
+      "step": 93530
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 8.944437359180377e-06,
+      "loss": 0.4435,
+      "step": 93540
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 8.859414190366876e-06,
+      "loss": 0.4437,
+      "step": 93550
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 8.774391021553374e-06,
+      "loss": 0.4535,
+      "step": 93560
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 8.68936785273987e-06,
+      "loss": 0.452,
+      "step": 93570
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 8.60434468392637e-06,
+      "loss": 0.4507,
+      "step": 93580
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.519321515112869e-06,
+      "loss": 0.4472,
+      "step": 93590
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.434298346299367e-06,
+      "loss": 0.4803,
+      "step": 93600
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.349275177485866e-06,
+      "loss": 0.4232,
+      "step": 93610
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.264252008672362e-06,
+      "loss": 0.4921,
+      "step": 93620
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.179228839858862e-06,
+      "loss": 0.3874,
+      "step": 93630
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.09420567104536e-06,
+      "loss": 0.4414,
+      "step": 93640
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.009182502231859e-06,
+      "loss": 0.4693,
+      "step": 93650
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.924159333418357e-06,
+      "loss": 0.471,
+      "step": 93660
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.839136164604856e-06,
+      "loss": 0.4747,
+      "step": 93670
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.754112995791352e-06,
+      "loss": 0.454,
+      "step": 93680
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.669089826977852e-06,
+      "loss": 0.4822,
+      "step": 93690
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 7.5840666581643506e-06,
+      "loss": 0.4196,
+      "step": 93700
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.499043489350849e-06,
+      "loss": 0.4578,
+      "step": 93710
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.414020320537347e-06,
+      "loss": 0.4302,
+      "step": 93720
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.328997151723845e-06,
+      "loss": 0.4442,
+      "step": 93730
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.243973982910343e-06,
+      "loss": 0.3902,
+      "step": 93740
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.158950814096841e-06,
+      "loss": 0.4424,
+      "step": 93750
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 7.0739276452833405e-06,
+      "loss": 0.4025,
+      "step": 93760
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.988904476469839e-06,
+      "loss": 0.3716,
+      "step": 93770
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.903881307656336e-06,
+      "loss": 0.4438,
+      "step": 93780
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.818858138842835e-06,
+      "loss": 0.4907,
+      "step": 93790
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.733834970029333e-06,
+      "loss": 0.4198,
+      "step": 93800
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.648811801215832e-06,
+      "loss": 0.4378,
+      "step": 93810
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.5637886324023305e-06,
+      "loss": 0.487,
+      "step": 93820
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.478765463588828e-06,
+      "loss": 0.4017,
+      "step": 93830
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.393742294775326e-06,
+      "loss": 0.4379,
+      "step": 93840
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.308719125961825e-06,
+      "loss": 0.43,
+      "step": 93850
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.223695957148323e-06,
+      "loss": 0.4552,
+      "step": 93860
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.138672788334821e-06,
+      "loss": 0.4409,
+      "step": 93870
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.05364961952132e-06,
+      "loss": 0.4301,
+      "step": 93880
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.968626450707819e-06,
+      "loss": 0.4619,
+      "step": 93890
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.883603281894316e-06,
+      "loss": 0.4983,
+      "step": 93900
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.798580113080815e-06,
+      "loss": 0.4027,
+      "step": 93910
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.713556944267314e-06,
+      "loss": 0.3951,
+      "step": 93920
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.628533775453811e-06,
+      "loss": 0.4449,
+      "step": 93930
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.5435106066403096e-06,
+      "loss": 0.4592,
+      "step": 93940
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.458487437826808e-06,
+      "loss": 0.4083,
+      "step": 93950
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.373464269013306e-06,
+      "loss": 0.49,
+      "step": 93960
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.2884411001998045e-06,
+      "loss": 0.4149,
+      "step": 93970
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.203417931386303e-06,
+      "loss": 0.4211,
+      "step": 93980
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.118394762572801e-06,
+      "loss": 0.4695,
+      "step": 93990
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.0333715937593e-06,
+      "loss": 0.4552,
+      "step": 94000
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.948348424945798e-06,
+      "loss": 0.4532,
+      "step": 94010
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.863325256132296e-06,
+      "loss": 0.4906,
+      "step": 94020
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.7783020873187945e-06,
+      "loss": 0.4303,
+      "step": 94030
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.693278918505293e-06,
+      "loss": 0.4965,
+      "step": 94040
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.608255749691791e-06,
+      "loss": 0.4592,
+      "step": 94050
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.5232325808782895e-06,
+      "loss": 0.4168,
+      "step": 94060
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.438209412064788e-06,
+      "loss": 0.4044,
+      "step": 94070
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.353186243251286e-06,
+      "loss": 0.4498,
+      "step": 94080
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.2681630744377845e-06,
+      "loss": 0.4512,
+      "step": 94090
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.183139905624283e-06,
+      "loss": 0.498,
+      "step": 94100
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.098116736810781e-06,
+      "loss": 0.485,
+      "step": 94110
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.0130935679972795e-06,
+      "loss": 0.3993,
+      "step": 94120
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.928070399183778e-06,
+      "loss": 0.4418,
+      "step": 94130
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.843047230370276e-06,
+      "loss": 0.398,
+      "step": 94140
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.7580240615567744e-06,
+      "loss": 0.4658,
+      "step": 94150
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.6730008927432723e-06,
+      "loss": 0.4189,
+      "step": 94160
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.587977723929771e-06,
+      "loss": 0.4566,
+      "step": 94170
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.5029545551162694e-06,
+      "loss": 0.4739,
+      "step": 94180
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.4179313863027673e-06,
+      "loss": 0.4158,
+      "step": 94190
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.332908217489266e-06,
+      "loss": 0.4717,
+      "step": 94200
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.247885048675765e-06,
+      "loss": 0.3807,
+      "step": 94210
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.1628618798622623e-06,
+      "loss": 0.4482,
+      "step": 94220
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.077838711048761e-06,
+      "loss": 0.4644,
+      "step": 94230
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.9928155422352594e-06,
+      "loss": 0.4098,
+      "step": 94240
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.9077923734217577e-06,
+      "loss": 0.4333,
+      "step": 94250
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.8227692046082556e-06,
+      "loss": 0.4316,
+      "step": 94260
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.7377460357947544e-06,
+      "loss": 0.4393,
+      "step": 94270
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.6527228669812527e-06,
+      "loss": 0.433,
+      "step": 94280
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.567699698167751e-06,
+      "loss": 0.4792,
+      "step": 94290
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.482676529354249e-06,
+      "loss": 0.4011,
+      "step": 94300
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.3976533605407473e-06,
+      "loss": 0.4026,
+      "step": 94310
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.312630191727246e-06,
+      "loss": 0.4044,
+      "step": 94320
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.2276070229137443e-06,
+      "loss": 0.4494,
+      "step": 94330
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.1425838541002422e-06,
+      "loss": 0.4625,
+      "step": 94340
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.0575606852867406e-06,
+      "loss": 0.5127,
+      "step": 94350
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.9725375164732393e-06,
+      "loss": 0.5322,
+      "step": 94360
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.8875143476597374e-06,
+      "loss": 0.4295,
+      "step": 94370
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.8024911788462355e-06,
+      "loss": 0.4933,
+      "step": 94380
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.7174680100327339e-06,
+      "loss": 0.4859,
+      "step": 94390
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.6324448412192324e-06,
+      "loss": 0.4029,
+      "step": 94400
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.5474216724057307e-06,
+      "loss": 0.4206,
+      "step": 94410
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.4623985035922289e-06,
+      "loss": 0.4541,
+      "step": 94420
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.3773753347787274e-06,
+      "loss": 0.4267,
+      "step": 94430
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.2923521659652255e-06,
+      "loss": 0.4443,
+      "step": 94440
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.207328997151724e-06,
+      "loss": 0.4147,
+      "step": 94450
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.1223058283382222e-06,
+      "loss": 0.4217,
+      "step": 94460
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.0372826595247207e-06,
+      "loss": 0.4037,
+      "step": 94470
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 9.522594907112188e-07,
+      "loss": 0.4631,
+      "step": 94480
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 8.672363218977173e-07,
+      "loss": 0.4352,
+      "step": 94490
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 7.822131530842155e-07,
+      "loss": 0.4771,
+      "step": 94500
+    }
+  ],
+  "max_steps": 94592,
+  "num_train_epochs": 8,
+  "total_flos": 9.214365240970445e+17,
+  "trial_name": null,
+  "trial_params": null
+}