diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,52618 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "global_step": 87599,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 10.6468,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 10.6492,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 10.6458,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 10.6415,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 10.6404,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 10.6303,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 10.6289,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 10.6157,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 10.6121,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.6048,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 10.5845,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2e-05,
+      "loss": 10.5538,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 10.5247,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 10.4147,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-05,
+      "loss": 10.2692,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 10.0956,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 9.9124,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-05,
+      "loss": 9.7317,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9e-05,
+      "loss": 9.4866,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 9.2719,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1e-05,
+      "loss": 9.0888,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 8.8357,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 8.6098,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-05,
+      "loss": 8.3784,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-05,
+      "loss": 8.045,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 7.8679,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 7.6852,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 7.4301,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9e-05,
+      "loss": 7.1625,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 7.0188,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1e-05,
+      "loss": 6.8507,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 6.6495,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3e-05,
+      "loss": 6.7173,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 6.4475,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5e-05,
+      "loss": 6.3956,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6e-05,
+      "loss": 6.2667,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7e-05,
+      "loss": 6.3509,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8e-05,
+      "loss": 6.2078,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 6.2955,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4e-05,
+      "loss": 6.2176,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1e-05,
+      "loss": 6.2656,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2e-05,
+      "loss": 6.0814,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3e-05,
+      "loss": 6.0821,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 6.1788,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5e-05,
+      "loss": 6.0505,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 6.0381,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7e-05,
+      "loss": 6.0533,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8e-05,
+      "loss": 6.0918,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9e-05,
+      "loss": 6.1718,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 6.0889,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9997711565746717e-05,
+      "loss": 5.9879,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999542313149344e-05,
+      "loss": 6.0499,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999313469724015e-05,
+      "loss": 6.0488,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9990846262986866e-05,
+      "loss": 6.0059,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998855782873359e-05,
+      "loss": 6.0262,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99862693944803e-05,
+      "loss": 6.0835,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9983980960227015e-05,
+      "loss": 6.0749,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9981692525973736e-05,
+      "loss": 5.991,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997940409172045e-05,
+      "loss": 5.951,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997711565746716e-05,
+      "loss": 6.0269,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997482722321388e-05,
+      "loss": 6.0419,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997253878896059e-05,
+      "loss": 6.0267,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9970250354707306e-05,
+      "loss": 6.03,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996796192045403e-05,
+      "loss": 6.0709,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996567348620074e-05,
+      "loss": 6.0702,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9963385051947456e-05,
+      "loss": 5.9611,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9961096617694177e-05,
+      "loss": 6.0348,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995880818344089e-05,
+      "loss": 5.9904,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9956519749187605e-05,
+      "loss": 5.9468,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9954231314934326e-05,
+      "loss": 6.0501,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995194288068104e-05,
+      "loss": 6.1255,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9949654446427754e-05,
+      "loss": 5.9189,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9947366012174475e-05,
+      "loss": 5.9645,
+      "step": 730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994507757792119e-05,
+      "loss": 5.9799,
+      "step": 740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99427891436679e-05,
+      "loss": 5.9145,
+      "step": 750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9940500709414624e-05,
+      "loss": 5.9911,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993821227516134e-05,
+      "loss": 5.8975,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993592384090805e-05,
+      "loss": 5.9823,
+      "step": 780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993363540665477e-05,
+      "loss": 5.9283,
+      "step": 790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993134697240148e-05,
+      "loss": 5.8387,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99290585381482e-05,
+      "loss": 5.9164,
+      "step": 810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9926770103894916e-05,
+      "loss": 5.8926,
+      "step": 820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992448166964163e-05,
+      "loss": 5.8451,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992219323538835e-05,
+      "loss": 5.8911,
+      "step": 840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9919904801135065e-05,
+      "loss": 5.9424,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991761636688178e-05,
+      "loss": 5.9003,
+      "step": 860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99153279326285e-05,
+      "loss": 5.8693,
+      "step": 870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9913039498375214e-05,
+      "loss": 5.9383,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991075106412193e-05,
+      "loss": 5.867,
+      "step": 890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990846262986865e-05,
+      "loss": 5.7465,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990617419561536e-05,
+      "loss": 5.7667,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990388576136208e-05,
+      "loss": 5.8488,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99015973271088e-05,
+      "loss": 5.7592,
+      "step": 930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989930889285551e-05,
+      "loss": 5.8281,
+      "step": 940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9897020458602226e-05,
+      "loss": 5.8211,
+      "step": 950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989473202434895e-05,
+      "loss": 5.6165,
+      "step": 960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989244359009566e-05,
+      "loss": 5.8037,
+      "step": 970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9890155155842376e-05,
+      "loss": 5.7826,
+      "step": 980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9887866721589097e-05,
+      "loss": 5.7144,
+      "step": 990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9885578287335804e-05,
+      "loss": 5.6998,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988328985308252e-05,
+      "loss": 5.764,
+      "step": 1010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988100141882924e-05,
+      "loss": 5.6322,
+      "step": 1020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987871298457595e-05,
+      "loss": 5.7329,
+      "step": 1030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987642455032267e-05,
+      "loss": 5.6375,
+      "step": 1040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987413611606939e-05,
+      "loss": 5.5558,
+      "step": 1050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98718476818161e-05,
+      "loss": 5.521,
+      "step": 1060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9869559247562816e-05,
+      "loss": 5.717,
+      "step": 1070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986727081330954e-05,
+      "loss": 5.5194,
+      "step": 1080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986498237905625e-05,
+      "loss": 5.5116,
+      "step": 1090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9862693944802966e-05,
+      "loss": 5.6348,
+      "step": 1100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9860405510549687e-05,
+      "loss": 5.3917,
+      "step": 1110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98581170762964e-05,
+      "loss": 5.6154,
+      "step": 1120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9855828642043115e-05,
+      "loss": 5.4854,
+      "step": 1130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9853540207789836e-05,
+      "loss": 5.564,
+      "step": 1140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985125177353655e-05,
+      "loss": 5.4132,
+      "step": 1150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9848963339283264e-05,
+      "loss": 5.4725,
+      "step": 1160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9846674905029985e-05,
+      "loss": 5.4318,
+      "step": 1170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98443864707767e-05,
+      "loss": 5.3892,
+      "step": 1180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984209803652341e-05,
+      "loss": 5.3275,
+      "step": 1190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983980960227013e-05,
+      "loss": 5.3651,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.983752116801684e-05,
+      "loss": 5.4367,
+      "step": 1210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.983523273376356e-05,
+      "loss": 5.3669,
+      "step": 1220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9832944299510276e-05,
+      "loss": 5.2643,
+      "step": 1230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.983065586525699e-05,
+      "loss": 5.3319,
+      "step": 1240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.982836743100371e-05,
+      "loss": 5.4298,
+      "step": 1250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9826078996750426e-05,
+      "loss": 5.1351,
+      "step": 1260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.982379056249714e-05,
+      "loss": 5.2959,
+      "step": 1270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.982150212824386e-05,
+      "loss": 5.3613,
+      "step": 1280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9819213693990575e-05,
+      "loss": 5.1739,
+      "step": 1290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.981692525973729e-05,
+      "loss": 5.2899,
+      "step": 1300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.981463682548401e-05,
+      "loss": 5.1881,
+      "step": 1310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9812348391230724e-05,
+      "loss": 5.2533,
+      "step": 1320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.981005995697744e-05,
+      "loss": 5.1671,
+      "step": 1330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980777152272416e-05,
+      "loss": 5.193,
+      "step": 1340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980548308847087e-05,
+      "loss": 5.5243,
+      "step": 1350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980319465421759e-05,
+      "loss": 4.9803,
+      "step": 1360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980090621996431e-05,
+      "loss": 5.0336,
+      "step": 1370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979861778571102e-05,
+      "loss": 4.9251,
+      "step": 1380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979632935145773e-05,
+      "loss": 5.0601,
+      "step": 1390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979404091720445e-05,
+      "loss": 4.8293,
+      "step": 1400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9791752482951165e-05,
+      "loss": 5.1071,
+      "step": 1410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978946404869788e-05,
+      "loss": 5.0192,
+      "step": 1420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.97871756144446e-05,
+      "loss": 4.974,
+      "step": 1430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9784887180191314e-05,
+      "loss": 4.9292,
+      "step": 1440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.978259874593803e-05,
+      "loss": 4.9188,
+      "step": 1450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.978031031168475e-05,
+      "loss": 4.9091,
+      "step": 1460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977802187743146e-05,
+      "loss": 4.9007,
+      "step": 1470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977573344317818e-05,
+      "loss": 4.7506,
+      "step": 1480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.97734450089249e-05,
+      "loss": 4.9873,
+      "step": 1490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977115657467161e-05,
+      "loss": 4.9096,
+      "step": 1500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9768868140418326e-05,
+      "loss": 4.8511,
+      "step": 1510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976657970616505e-05,
+      "loss": 4.8182,
+      "step": 1520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976429127191176e-05,
+      "loss": 4.6981,
+      "step": 1530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9762002837658476e-05,
+      "loss": 4.7869,
+      "step": 1540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9759714403405196e-05,
+      "loss": 4.5999,
+      "step": 1550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975742596915191e-05,
+      "loss": 4.7654,
+      "step": 1560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9755137534898625e-05,
+      "loss": 4.8185,
+      "step": 1570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9752849100645346e-05,
+      "loss": 5.0029,
+      "step": 1580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975056066639205e-05,
+      "loss": 4.8317,
+      "step": 1590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9748272232138774e-05,
+      "loss": 4.7292,
+      "step": 1600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.974598379788549e-05,
+      "loss": 4.6309,
+      "step": 1610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.97436953636322e-05,
+      "loss": 4.6402,
+      "step": 1620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.974140692937892e-05,
+      "loss": 4.6918,
+      "step": 1630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.973911849512564e-05,
+      "loss": 4.5851,
+      "step": 1640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.973683006087235e-05,
+      "loss": 4.6675,
+      "step": 1650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.973454162661907e-05,
+      "loss": 4.5958,
+      "step": 1660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9732253192365786e-05,
+      "loss": 4.689,
+      "step": 1670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.97299647581125e-05,
+      "loss": 4.4157,
+      "step": 1680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9727676323859215e-05,
+      "loss": 4.5446,
+      "step": 1690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9725387889605936e-05,
+      "loss": 4.4268,
+      "step": 1700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.972309945535265e-05,
+      "loss": 4.4364,
+      "step": 1710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9720811021099364e-05,
+      "loss": 4.5106,
+      "step": 1720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9718522586846085e-05,
+      "loss": 4.6897,
+      "step": 1730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.97162341525928e-05,
+      "loss": 4.4631,
+      "step": 1740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.971394571833951e-05,
+      "loss": 4.5942,
+      "step": 1750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9711657284086234e-05,
+      "loss": 4.4797,
+      "step": 1760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970936884983295e-05,
+      "loss": 4.3681,
+      "step": 1770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970708041557966e-05,
+      "loss": 4.4252,
+      "step": 1780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9704791981326376e-05,
+      "loss": 4.3536,
+      "step": 1790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970250354707309e-05,
+      "loss": 4.5191,
+      "step": 1800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970021511281981e-05,
+      "loss": 4.2711,
+      "step": 1810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9697926678566526e-05,
+      "loss": 4.5769,
+      "step": 1820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.969563824431324e-05,
+      "loss": 4.2891,
+      "step": 1830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.969334981005996e-05,
+      "loss": 4.3045,
+      "step": 1840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9691061375806675e-05,
+      "loss": 4.4258,
+      "step": 1850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.968877294155339e-05,
+      "loss": 4.2687,
+      "step": 1860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.968648450730011e-05,
+      "loss": 4.4235,
+      "step": 1870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9684196073046824e-05,
+      "loss": 4.1402,
+      "step": 1880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.968190763879354e-05,
+      "loss": 4.209,
+      "step": 1890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.967961920454026e-05,
+      "loss": 4.2591,
+      "step": 1900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.967733077028697e-05,
+      "loss": 4.144,
+      "step": 1910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.967504233603369e-05,
+      "loss": 4.283,
+      "step": 1920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.967275390178041e-05,
+      "loss": 4.3179,
+      "step": 1930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.967046546752712e-05,
+      "loss": 4.1564,
+      "step": 1940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9668177033273836e-05,
+      "loss": 4.1144,
+      "step": 1950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.966588859902056e-05,
+      "loss": 4.3768,
+      "step": 1960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.966360016476727e-05,
+      "loss": 4.3718,
+      "step": 1970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9661311730513986e-05,
+      "loss": 4.1446,
+      "step": 1980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.96590232962607e-05,
+      "loss": 4.1684,
+      "step": 1990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9656734862007414e-05,
+      "loss": 4.1453,
+      "step": 2000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.965444642775413e-05,
+      "loss": 4.0921,
+      "step": 2010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.965215799350085e-05,
+      "loss": 4.115,
+      "step": 2020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.964986955924756e-05,
+      "loss": 4.094,
+      "step": 2030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.964758112499428e-05,
+      "loss": 4.0735,
+      "step": 2040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9645292690741e-05,
+      "loss": 4.0245,
+      "step": 2050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.964300425648771e-05,
+      "loss": 3.9754,
+      "step": 2060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9640715822234426e-05,
+      "loss": 4.1716,
+      "step": 2070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.963842738798115e-05,
+      "loss": 3.8179,
+      "step": 2080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.963613895372786e-05,
+      "loss": 4.036,
+      "step": 2090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9633850519474575e-05,
+      "loss": 4.2522,
+      "step": 2100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9631562085221296e-05,
+      "loss": 3.9801,
+      "step": 2110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.962927365096801e-05,
+      "loss": 3.9878,
+      "step": 2120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9626985216714725e-05,
+      "loss": 3.9668,
+      "step": 2130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9624696782461446e-05,
+      "loss": 4.0614,
+      "step": 2140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.962240834820816e-05,
+      "loss": 4.0036,
+      "step": 2150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9620119913954874e-05,
+      "loss": 3.8624,
+      "step": 2160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9617831479701595e-05,
+      "loss": 3.9028,
+      "step": 2170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.96155430454483e-05,
+      "loss": 3.982,
+      "step": 2180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.961325461119502e-05,
+      "loss": 3.9641,
+      "step": 2190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.961096617694174e-05,
+      "loss": 3.7962,
+      "step": 2200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.960867774268845e-05,
+      "loss": 4.0071,
+      "step": 2210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.960638930843517e-05,
+      "loss": 3.9433,
+      "step": 2220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9604100874181886e-05,
+      "loss": 3.943,
+      "step": 2230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.96018124399286e-05,
+      "loss": 3.6835,
+      "step": 2240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.959952400567532e-05,
+      "loss": 3.9436,
+      "step": 2250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9597235571422035e-05,
+      "loss": 3.6303,
+      "step": 2260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.959494713716875e-05,
+      "loss": 3.6082,
+      "step": 2270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.959265870291547e-05,
+      "loss": 3.9698,
+      "step": 2280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9590370268662185e-05,
+      "loss": 3.7346,
+      "step": 2290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.95880818344089e-05,
+      "loss": 3.785,
+      "step": 2300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.958579340015562e-05,
+      "loss": 3.634,
+      "step": 2310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9583504965902334e-05,
+      "loss": 3.8224,
+      "step": 2320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.958121653164905e-05,
+      "loss": 3.7121,
+      "step": 2330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.957892809739577e-05,
+      "loss": 3.6801,
+      "step": 2340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.957663966314248e-05,
+      "loss": 3.6434,
+      "step": 2350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.95743512288892e-05,
+      "loss": 3.5632,
+      "step": 2360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.957206279463592e-05,
+      "loss": 3.7956,
+      "step": 2370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9569774360382625e-05,
+      "loss": 3.7147,
+      "step": 2380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.956748592612934e-05,
+      "loss": 3.9062,
+      "step": 2390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.956519749187606e-05,
+      "loss": 3.7015,
+      "step": 2400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9562909057622775e-05,
+      "loss": 3.7737,
+      "step": 2410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.956062062336949e-05,
+      "loss": 3.6225,
+      "step": 2420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.955833218911621e-05,
+      "loss": 3.8234,
+      "step": 2430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9556043754862924e-05,
+      "loss": 3.6219,
+      "step": 2440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.955375532060964e-05,
+      "loss": 3.4964,
+      "step": 2450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.955146688635636e-05,
+      "loss": 3.5518,
+      "step": 2460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.954917845210307e-05,
+      "loss": 3.4895,
+      "step": 2470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.954689001784979e-05,
+      "loss": 3.5502,
+      "step": 2480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.954460158359651e-05,
+      "loss": 3.629,
+      "step": 2490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.954231314934322e-05,
+      "loss": 3.6208,
+      "step": 2500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9540024715089936e-05,
+      "loss": 3.5531,
+      "step": 2510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.953773628083666e-05,
+      "loss": 3.4852,
+      "step": 2520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.953544784658337e-05,
+      "loss": 3.446,
+      "step": 2530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9533159412330085e-05,
+      "loss": 3.5893,
+      "step": 2540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9530870978076806e-05,
+      "loss": 3.5965,
+      "step": 2550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.952858254382352e-05,
+      "loss": 3.5799,
+      "step": 2560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9526294109570235e-05,
+      "loss": 3.5579,
+      "step": 2570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.952400567531695e-05,
+      "loss": 3.5597,
+      "step": 2580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.952171724106366e-05,
+      "loss": 3.64,
+      "step": 2590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9519428806810384e-05,
+      "loss": 3.6476,
+      "step": 2600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.95171403725571e-05,
+      "loss": 3.4483,
+      "step": 2610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.951485193830381e-05,
+      "loss": 3.6166,
+      "step": 2620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.951256350405053e-05,
+      "loss": 3.4921,
+      "step": 2630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.951027506979725e-05,
+      "loss": 3.5374,
+      "step": 2640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.950798663554396e-05,
+      "loss": 3.6803,
+      "step": 2650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.950569820129068e-05,
+      "loss": 3.594,
+      "step": 2660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9503409767037396e-05,
+      "loss": 3.4511,
+      "step": 2670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.950112133278411e-05,
+      "loss": 3.5751,
+      "step": 2680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.949883289853083e-05,
+      "loss": 3.4753,
+      "step": 2690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9496544464277545e-05,
+      "loss": 3.4152,
+      "step": 2700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.949425603002426e-05,
+      "loss": 3.4295,
+      "step": 2710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.949196759577098e-05,
+      "loss": 3.4824,
+      "step": 2720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9489679161517695e-05,
+      "loss": 3.4024,
+      "step": 2730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.948739072726441e-05,
+      "loss": 3.5077,
+      "step": 2740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.948510229301113e-05,
+      "loss": 3.5525,
+      "step": 2750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9482813858757844e-05,
+      "loss": 3.3653,
+      "step": 2760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.948052542450456e-05,
+      "loss": 3.4213,
+      "step": 2770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.947823699025127e-05,
+      "loss": 3.2256,
+      "step": 2780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9475948555997986e-05,
+      "loss": 3.3265,
+      "step": 2790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.94736601217447e-05,
+      "loss": 3.5103,
+      "step": 2800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.947137168749142e-05,
+      "loss": 3.2936,
+      "step": 2810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9469083253238135e-05,
+      "loss": 3.2385,
+      "step": 2820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.946679481898485e-05,
+      "loss": 3.2628,
+      "step": 2830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.946450638473157e-05,
+      "loss": 3.2475,
+      "step": 2840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9462217950478285e-05,
+      "loss": 3.3101,
+      "step": 2850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9459929516225e-05,
+      "loss": 3.275,
+      "step": 2860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.945764108197172e-05,
+      "loss": 3.2514,
+      "step": 2870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9455352647718434e-05,
+      "loss": 3.4532,
+      "step": 2880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.945306421346515e-05,
+      "loss": 3.349,
+      "step": 2890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.945077577921187e-05,
+      "loss": 3.1803,
+      "step": 2900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.944848734495858e-05,
+      "loss": 3.3639,
+      "step": 2910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.94461989107053e-05,
+      "loss": 3.3155,
+      "step": 2920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.944391047645202e-05,
+      "loss": 3.3371,
+      "step": 2930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.944162204219873e-05,
+      "loss": 3.2016,
+      "step": 2940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9439333607945446e-05,
+      "loss": 3.1036,
+      "step": 2950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.943704517369217e-05,
+      "loss": 3.3109,
+      "step": 2960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9434756739438874e-05,
+      "loss": 3.3041,
+      "step": 2970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.943246830518559e-05,
+      "loss": 3.2101,
+      "step": 2980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.943017987093231e-05,
+      "loss": 3.2372,
+      "step": 2990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9427891436679024e-05,
+      "loss": 3.1914,
+      "step": 3000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.942560300242574e-05,
+      "loss": 3.2797,
+      "step": 3010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.942331456817246e-05,
+      "loss": 3.3601,
+      "step": 3020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.942102613391917e-05,
+      "loss": 3.4813,
+      "step": 3030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.941873769966589e-05,
+      "loss": 3.1892,
+      "step": 3040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.941644926541261e-05,
+      "loss": 3.3008,
+      "step": 3050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.941416083115932e-05,
+      "loss": 3.3227,
+      "step": 3060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9411872396906036e-05,
+      "loss": 3.2429,
+      "step": 3070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.940958396265276e-05,
+      "loss": 3.375,
+      "step": 3080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.940729552839947e-05,
+      "loss": 3.255,
+      "step": 3090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9405007094146185e-05,
+      "loss": 3.3024,
+      "step": 3100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9402718659892906e-05,
+      "loss": 3.3271,
+      "step": 3110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.940043022563962e-05,
+      "loss": 3.1107,
+      "step": 3120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9398141791386335e-05,
+      "loss": 3.1728,
+      "step": 3130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9395853357133055e-05,
+      "loss": 3.4514,
+      "step": 3140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.939356492287977e-05,
+      "loss": 2.8632,
+      "step": 3150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9391276488626484e-05,
+      "loss": 3.4124,
+      "step": 3160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.93889880543732e-05,
+      "loss": 3.111,
+      "step": 3170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.938669962011991e-05,
+      "loss": 3.1243,
+      "step": 3180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.938441118586663e-05,
+      "loss": 3.2455,
+      "step": 3190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.938212275161335e-05,
+      "loss": 3.1032,
+      "step": 3200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.937983431736006e-05,
+      "loss": 3.3117,
+      "step": 3210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.937754588310678e-05,
+      "loss": 3.2521,
+      "step": 3220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9375257448853496e-05,
+      "loss": 3.3084,
+      "step": 3230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.937296901460021e-05,
+      "loss": 3.1779,
+      "step": 3240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.937068058034693e-05,
+      "loss": 3.0942,
+      "step": 3250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9368392146093645e-05,
+      "loss": 2.9396,
+      "step": 3260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.936610371184036e-05,
+      "loss": 3.1771,
+      "step": 3270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.936381527758708e-05,
+      "loss": 3.116,
+      "step": 3280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9361526843333795e-05,
+      "loss": 3.1216,
+      "step": 3290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.935923840908051e-05,
+      "loss": 3.0335,
+      "step": 3300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.935694997482723e-05,
+      "loss": 3.2038,
+      "step": 3310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9354661540573944e-05,
+      "loss": 3.1194,
+      "step": 3320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.935237310632066e-05,
+      "loss": 3.0588,
+      "step": 3330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.935008467206738e-05,
+      "loss": 3.1875,
+      "step": 3340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.934779623781409e-05,
+      "loss": 3.0593,
+      "step": 3350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.934550780356081e-05,
+      "loss": 3.0694,
+      "step": 3360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.934321936930752e-05,
+      "loss": 3.0894,
+      "step": 3370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9340930935054235e-05,
+      "loss": 3.0116,
+      "step": 3380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.933864250080095e-05,
+      "loss": 3.0874,
+      "step": 3390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.933635406654767e-05,
+      "loss": 3.2471,
+      "step": 3400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9334065632294384e-05,
+      "loss": 3.1044,
+      "step": 3410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.93317771980411e-05,
+      "loss": 2.9131,
+      "step": 3420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.932948876378782e-05,
+      "loss": 3.0253,
+      "step": 3430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9327200329534534e-05,
+      "loss": 3.0284,
+      "step": 3440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.932491189528125e-05,
+      "loss": 2.9507,
+      "step": 3450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.932262346102797e-05,
+      "loss": 3.0571,
+      "step": 3460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.932033502677468e-05,
+      "loss": 3.0379,
+      "step": 3470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.93180465925214e-05,
+      "loss": 3.2292,
+      "step": 3480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.931575815826812e-05,
+      "loss": 3.0898,
+      "step": 3490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.931346972401483e-05,
+      "loss": 2.9006,
+      "step": 3500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9311181289761546e-05,
+      "loss": 2.984,
+      "step": 3510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.930889285550827e-05,
+      "loss": 3.0123,
+      "step": 3520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.930660442125498e-05,
+      "loss": 3.2169,
+      "step": 3530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9304315987001695e-05,
+      "loss": 3.252,
+      "step": 3540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9302027552748416e-05,
+      "loss": 3.0935,
+      "step": 3550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.929973911849513e-05,
+      "loss": 2.9241,
+      "step": 3560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9297450684241844e-05,
+      "loss": 3.0553,
+      "step": 3570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.929516224998856e-05,
+      "loss": 3.2508,
+      "step": 3580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.929287381573527e-05,
+      "loss": 2.9135,
+      "step": 3590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9290585381481994e-05,
+      "loss": 3.1094,
+      "step": 3600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.928829694722871e-05,
+      "loss": 3.0007,
+      "step": 3610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.928600851297542e-05,
+      "loss": 2.9614,
+      "step": 3620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.928372007872214e-05,
+      "loss": 3.1314,
+      "step": 3630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.928143164446886e-05,
+      "loss": 2.9977,
+      "step": 3640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.927914321021557e-05,
+      "loss": 2.9733,
+      "step": 3650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.927685477596229e-05,
+      "loss": 3.0223,
+      "step": 3660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9274566341709006e-05,
+      "loss": 2.9454,
+      "step": 3670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.927227790745572e-05,
+      "loss": 2.8938,
+      "step": 3680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.926998947320244e-05,
+      "loss": 2.8467,
+      "step": 3690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9267701038949155e-05,
+      "loss": 2.946,
+      "step": 3700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.926541260469587e-05,
+      "loss": 3.1829,
+      "step": 3710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.926312417044259e-05,
+      "loss": 3.0131,
+      "step": 3720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9260835736189305e-05,
+      "loss": 3.0267,
+      "step": 3730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.925854730193602e-05,
+      "loss": 3.3027,
+      "step": 3740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.925625886768274e-05,
+      "loss": 2.9908,
+      "step": 3750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.925397043342945e-05,
+      "loss": 2.8306,
+      "step": 3760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.925168199917616e-05,
+      "loss": 3.061,
+      "step": 3770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.924939356492288e-05,
+      "loss": 2.9045,
+      "step": 3780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9247105130669596e-05,
+      "loss": 2.8821,
+      "step": 3790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.924481669641631e-05,
+      "loss": 2.8715,
+      "step": 3800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.924252826216303e-05,
+      "loss": 2.9067,
+      "step": 3810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9240239827909745e-05,
+      "loss": 3.0192,
+      "step": 3820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.923795139365646e-05,
+      "loss": 3.0298,
+      "step": 3830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.923566295940318e-05,
+      "loss": 2.9034,
+      "step": 3840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9233374525149894e-05,
+      "loss": 2.9827,
+      "step": 3850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.923108609089661e-05,
+      "loss": 2.8672,
+      "step": 3860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.922879765664333e-05,
+      "loss": 2.9291,
+      "step": 3870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9226509222390044e-05,
+      "loss": 2.9763,
+      "step": 3880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.922422078813676e-05,
+      "loss": 2.8443,
+      "step": 3890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.922193235388348e-05,
+      "loss": 2.8285,
+      "step": 3900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.921964391963019e-05,
+      "loss": 2.8529,
+      "step": 3910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.921735548537691e-05,
+      "loss": 2.9645,
+      "step": 3920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.921506705112363e-05,
+      "loss": 2.9857,
+      "step": 3930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.921277861687034e-05,
+      "loss": 3.0512,
+      "step": 3940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9210490182617056e-05,
+      "loss": 2.8715,
+      "step": 3950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.920820174836377e-05,
+      "loss": 2.9225,
+      "step": 3960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9205913314110484e-05,
+      "loss": 2.7595,
+      "step": 3970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9203624879857205e-05,
+      "loss": 2.8244,
+      "step": 3980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.920133644560392e-05,
+      "loss": 2.8686,
+      "step": 3990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9199048011350634e-05,
+      "loss": 2.771,
+      "step": 4000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.919675957709735e-05,
+      "loss": 2.8594,
+      "step": 4010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.919447114284407e-05,
+      "loss": 2.9962,
+      "step": 4020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.919218270859078e-05,
+      "loss": 2.9638,
+      "step": 4030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.91898942743375e-05,
+      "loss": 3.0201,
+      "step": 4040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.918760584008422e-05,
+      "loss": 2.9256,
+      "step": 4050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.918531740583093e-05,
+      "loss": 3.0207,
+      "step": 4060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9183028971577646e-05,
+      "loss": 2.7377,
+      "step": 4070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.918074053732437e-05,
+      "loss": 2.9199,
+      "step": 4080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.917845210307108e-05,
+      "loss": 2.9998,
+      "step": 4090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9176163668817795e-05,
+      "loss": 2.8586,
+      "step": 4100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9173875234564516e-05,
+      "loss": 2.993,
+      "step": 4110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.917158680031123e-05,
+      "loss": 2.8115,
+      "step": 4120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9169298366057944e-05,
+      "loss": 2.8757,
+      "step": 4130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9167009931804665e-05,
+      "loss": 2.9125,
+      "step": 4140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.916472149755138e-05,
+      "loss": 2.8955,
+      "step": 4150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9162433063298094e-05,
+      "loss": 2.7962,
+      "step": 4160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.916014462904481e-05,
+      "loss": 2.7999,
+      "step": 4170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.915785619479152e-05,
+      "loss": 2.9908,
+      "step": 4180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.915556776053824e-05,
+      "loss": 2.8225,
+      "step": 4190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.915327932628496e-05,
+      "loss": 2.7451,
+      "step": 4200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.915099089203167e-05,
+      "loss": 3.0593,
+      "step": 4210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.914870245777839e-05,
+      "loss": 2.7044,
+      "step": 4220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9146414023525106e-05,
+      "loss": 3.036,
+      "step": 4230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.914412558927182e-05,
+      "loss": 3.0159,
+      "step": 4240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.914183715501854e-05,
+      "loss": 2.789,
+      "step": 4250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9139548720765255e-05,
+      "loss": 2.8524,
+      "step": 4260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.913726028651197e-05,
+      "loss": 2.9255,
+      "step": 4270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.913497185225869e-05,
+      "loss": 2.819,
+      "step": 4280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9132683418005404e-05,
+      "loss": 2.8614,
+      "step": 4290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.913039498375212e-05,
+      "loss": 2.8063,
+      "step": 4300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.912810654949884e-05,
+      "loss": 2.8566,
+      "step": 4310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9125818115245554e-05,
+      "loss": 2.9287,
+      "step": 4320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.912352968099227e-05,
+      "loss": 2.9403,
+      "step": 4330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.912124124673899e-05,
+      "loss": 2.7084,
+      "step": 4340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.91189528124857e-05,
+      "loss": 2.7331,
+      "step": 4350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.911666437823241e-05,
+      "loss": 2.9216,
+      "step": 4360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.911437594397913e-05,
+      "loss": 3.0126,
+      "step": 4370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9112087509725845e-05,
+      "loss": 2.9256,
+      "step": 4380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.910979907547256e-05,
+      "loss": 2.8328,
+      "step": 4390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.910751064121928e-05,
+      "loss": 2.7388,
+      "step": 4400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9105222206965994e-05,
+      "loss": 2.7339,
+      "step": 4410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.910293377271271e-05,
+      "loss": 2.8324,
+      "step": 4420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.910064533845943e-05,
+      "loss": 2.8383,
+      "step": 4430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9098356904206144e-05,
+      "loss": 2.7014,
+      "step": 4440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.909606846995286e-05,
+      "loss": 2.8201,
+      "step": 4450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.909378003569958e-05,
+      "loss": 3.0264,
+      "step": 4460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.909149160144629e-05,
+      "loss": 2.8091,
+      "step": 4470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.908920316719301e-05,
+      "loss": 2.8537,
+      "step": 4480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.908691473293973e-05,
+      "loss": 2.9466,
+      "step": 4490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.908462629868644e-05,
+      "loss": 2.9187,
+      "step": 4500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9082337864433156e-05,
+      "loss": 2.8632,
+      "step": 4510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.908004943017988e-05,
+      "loss": 2.781,
+      "step": 4520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.907776099592659e-05,
+      "loss": 2.851,
+      "step": 4530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9075472561673305e-05,
+      "loss": 2.6722,
+      "step": 4540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.907318412742002e-05,
+      "loss": 2.8292,
+      "step": 4550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9070895693166733e-05,
+      "loss": 2.9132,
+      "step": 4560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9068607258913454e-05,
+      "loss": 2.5775,
+      "step": 4570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.906631882466017e-05,
+      "loss": 2.7283,
+      "step": 4580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.906403039040688e-05,
+      "loss": 2.9837,
+      "step": 4590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9061741956153604e-05,
+      "loss": 3.0339,
+      "step": 4600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.905945352190032e-05,
+      "loss": 2.6225,
+      "step": 4610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.905716508764703e-05,
+      "loss": 2.7087,
+      "step": 4620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.905487665339375e-05,
+      "loss": 2.8347,
+      "step": 4630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.905258821914047e-05,
+      "loss": 2.7919,
+      "step": 4640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.905029978488718e-05,
+      "loss": 2.6434,
+      "step": 4650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.90480113506339e-05,
+      "loss": 3.0099,
+      "step": 4660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9045722916380616e-05,
+      "loss": 2.8333,
+      "step": 4670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.904343448212733e-05,
+      "loss": 2.7933,
+      "step": 4680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.904114604787405e-05,
+      "loss": 2.9621,
+      "step": 4690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9038857613620765e-05,
+      "loss": 2.7535,
+      "step": 4700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.903656917936748e-05,
+      "loss": 2.8269,
+      "step": 4710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.90342807451142e-05,
+      "loss": 2.7093,
+      "step": 4720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9031992310860914e-05,
+      "loss": 2.8733,
+      "step": 4730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.902970387660763e-05,
+      "loss": 2.9256,
+      "step": 4740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.902741544235434e-05,
+      "loss": 2.7271,
+      "step": 4750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.902512700810106e-05,
+      "loss": 2.8289,
+      "step": 4760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.902283857384777e-05,
+      "loss": 2.7238,
+      "step": 4770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.902055013959449e-05,
+      "loss": 2.7404,
+      "step": 4780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9018261705341206e-05,
+      "loss": 2.6856,
+      "step": 4790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.901597327108792e-05,
+      "loss": 2.8218,
+      "step": 4800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.901368483683464e-05,
+      "loss": 2.8808,
+      "step": 4810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9011396402581355e-05,
+      "loss": 2.8053,
+      "step": 4820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.900910796832807e-05,
+      "loss": 2.7797,
+      "step": 4830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.900681953407479e-05,
+      "loss": 2.6849,
+      "step": 4840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9004531099821504e-05,
+      "loss": 2.9224,
+      "step": 4850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.900224266556822e-05,
+      "loss": 2.9994,
+      "step": 4860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.899995423131494e-05,
+      "loss": 2.8157,
+      "step": 4870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8997665797061653e-05,
+      "loss": 2.7373,
+      "step": 4880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.899537736280837e-05,
+      "loss": 2.7284,
+      "step": 4890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.899308892855509e-05,
+      "loss": 2.8367,
+      "step": 4900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.89908004943018e-05,
+      "loss": 2.7574,
+      "step": 4910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.898851206004852e-05,
+      "loss": 2.9788,
+      "step": 4920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.898622362579524e-05,
+      "loss": 2.8612,
+      "step": 4930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.898393519154195e-05,
+      "loss": 2.8805,
+      "step": 4940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8981646757288666e-05,
+      "loss": 2.741,
+      "step": 4950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.897935832303538e-05,
+      "loss": 2.7458,
+      "step": 4960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8977069888782094e-05,
+      "loss": 2.7443,
+      "step": 4970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8974781454528815e-05,
+      "loss": 2.7667,
+      "step": 4980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.897249302027553e-05,
+      "loss": 2.7318,
+      "step": 4990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8970204586022243e-05,
+      "loss": 2.7639,
+      "step": 5000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8967916151768964e-05,
+      "loss": 2.8134,
+      "step": 5010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.896562771751568e-05,
+      "loss": 2.6872,
+      "step": 5020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.896333928326239e-05,
+      "loss": 2.8299,
+      "step": 5030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8961050849009113e-05,
+      "loss": 2.8018,
+      "step": 5040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.895876241475583e-05,
+      "loss": 2.7224,
+      "step": 5050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.895647398050254e-05,
+      "loss": 2.9706,
+      "step": 5060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.895418554624926e-05,
+      "loss": 2.8135,
+      "step": 5070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.895189711199598e-05,
+      "loss": 2.64,
+      "step": 5080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.894960867774269e-05,
+      "loss": 3.0843,
+      "step": 5090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.894732024348941e-05,
+      "loss": 2.7187,
+      "step": 5100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8945031809236126e-05,
+      "loss": 2.7207,
+      "step": 5110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.894274337498284e-05,
+      "loss": 2.7636,
+      "step": 5120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8940454940729554e-05,
+      "loss": 2.8889,
+      "step": 5130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8938166506476275e-05,
+      "loss": 2.5818,
+      "step": 5140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.893587807222298e-05,
+      "loss": 2.7533,
+      "step": 5150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8933589637969703e-05,
+      "loss": 2.767,
+      "step": 5160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.893130120371642e-05,
+      "loss": 2.6906,
+      "step": 5170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.892901276946313e-05,
+      "loss": 2.6821,
+      "step": 5180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.892672433520985e-05,
+      "loss": 2.7352,
+      "step": 5190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.892443590095657e-05,
+      "loss": 2.7433,
+      "step": 5200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.892214746670328e-05,
+      "loss": 2.7549,
+      "step": 5210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.891985903245e-05,
+      "loss": 2.6625,
+      "step": 5220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8917570598196716e-05,
+      "loss": 2.6029,
+      "step": 5230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.891528216394343e-05,
+      "loss": 2.5315,
+      "step": 5240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.891299372969015e-05,
+      "loss": 2.7698,
+      "step": 5250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8910705295436865e-05,
+      "loss": 2.648,
+      "step": 5260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.890841686118358e-05,
+      "loss": 2.497,
+      "step": 5270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.89061284269303e-05,
+      "loss": 2.7683,
+      "step": 5280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8903839992677014e-05,
+      "loss": 2.7788,
+      "step": 5290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.890155155842373e-05,
+      "loss": 2.6834,
+      "step": 5300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.889926312417045e-05,
+      "loss": 2.6312,
+      "step": 5310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8896974689917163e-05,
+      "loss": 2.5096,
+      "step": 5320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.889468625566388e-05,
+      "loss": 2.6317,
+      "step": 5330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.889239782141059e-05,
+      "loss": 2.7149,
+      "step": 5340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8890109387157306e-05,
+      "loss": 2.5823,
+      "step": 5350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.888782095290402e-05,
+      "loss": 2.6265,
+      "step": 5360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.888553251865074e-05,
+      "loss": 2.5827,
+      "step": 5370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8883244084397455e-05,
+      "loss": 2.5964,
+      "step": 5380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.888095565014417e-05,
+      "loss": 2.6781,
+      "step": 5390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.887866721589089e-05,
+      "loss": 2.69,
+      "step": 5400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8876378781637604e-05,
+      "loss": 2.7274,
+      "step": 5410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.887409034738432e-05,
+      "loss": 2.8764,
+      "step": 5420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.887180191313104e-05,
+      "loss": 2.8443,
+      "step": 5430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.886951347887775e-05,
+      "loss": 2.6021,
+      "step": 5440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.886722504462447e-05,
+      "loss": 2.6461,
+      "step": 5450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.886493661037119e-05,
+      "loss": 2.7554,
+      "step": 5460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.88626481761179e-05,
+      "loss": 2.6662,
+      "step": 5470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.886035974186462e-05,
+      "loss": 2.7161,
+      "step": 5480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.885807130761134e-05,
+      "loss": 2.734,
+      "step": 5490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.885578287335805e-05,
+      "loss": 2.5871,
+      "step": 5500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8853494439104766e-05,
+      "loss": 2.6411,
+      "step": 5510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.885120600485149e-05,
+      "loss": 2.6922,
+      "step": 5520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.88489175705982e-05,
+      "loss": 2.8183,
+      "step": 5530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8846629136344915e-05,
+      "loss": 2.8494,
+      "step": 5540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.884434070209163e-05,
+      "loss": 2.7624,
+      "step": 5550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.884205226783834e-05,
+      "loss": 2.4398,
+      "step": 5560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8839763833585064e-05,
+      "loss": 2.6945,
+      "step": 5570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.883747539933178e-05,
+      "loss": 2.5568,
+      "step": 5580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.883518696507849e-05,
+      "loss": 2.7731,
+      "step": 5590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8832898530825213e-05,
+      "loss": 2.6838,
+      "step": 5600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.883061009657193e-05,
+      "loss": 2.6429,
+      "step": 5610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.882832166231864e-05,
+      "loss": 2.5941,
+      "step": 5620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.882603322806536e-05,
+      "loss": 2.6691,
+      "step": 5630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.882374479381208e-05,
+      "loss": 2.7819,
+      "step": 5640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.882145635955879e-05,
+      "loss": 2.7618,
+      "step": 5650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.881916792530551e-05,
+      "loss": 2.56,
+      "step": 5660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8816879491052226e-05,
+      "loss": 2.7314,
+      "step": 5670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.881459105679894e-05,
+      "loss": 2.7079,
+      "step": 5680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.881230262254566e-05,
+      "loss": 2.5127,
+      "step": 5690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8810014188292375e-05,
+      "loss": 2.5525,
+      "step": 5700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.880772575403909e-05,
+      "loss": 2.8079,
+      "step": 5710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.880543731978581e-05,
+      "loss": 2.7778,
+      "step": 5720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8803148885532524e-05,
+      "loss": 2.7115,
+      "step": 5730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.880086045127923e-05,
+      "loss": 2.7208,
+      "step": 5740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.879857201702595e-05,
+      "loss": 2.5249,
+      "step": 5750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.879628358277267e-05,
+      "loss": 2.6811,
+      "step": 5760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.879399514851938e-05,
+      "loss": 2.6039,
+      "step": 5770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.87917067142661e-05,
+      "loss": 2.5605,
+      "step": 5780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8789418280012816e-05,
+      "loss": 2.6976,
+      "step": 5790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.878712984575953e-05,
+      "loss": 2.6381,
+      "step": 5800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.878484141150625e-05,
+      "loss": 2.8265,
+      "step": 5810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8782552977252965e-05,
+      "loss": 2.4658,
+      "step": 5820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.878026454299968e-05,
+      "loss": 2.6137,
+      "step": 5830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.87779761087464e-05,
+      "loss": 2.6636,
+      "step": 5840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8775687674493114e-05,
+      "loss": 2.6783,
+      "step": 5850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.877339924023983e-05,
+      "loss": 2.8019,
+      "step": 5860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.877111080598655e-05,
+      "loss": 2.732,
+      "step": 5870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.876882237173326e-05,
+      "loss": 2.7777,
+      "step": 5880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.876653393747998e-05,
+      "loss": 2.5736,
+      "step": 5890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.87642455032267e-05,
+      "loss": 2.8001,
+      "step": 5900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.876195706897341e-05,
+      "loss": 2.7418,
+      "step": 5910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.875966863472013e-05,
+      "loss": 2.6527,
+      "step": 5920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.875738020046685e-05,
+      "loss": 2.7263,
+      "step": 5930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8755091766213555e-05,
+      "loss": 2.4208,
+      "step": 5940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8752803331960276e-05,
+      "loss": 2.5759,
+      "step": 5950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.875051489770699e-05,
+      "loss": 2.9091,
+      "step": 5960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8748226463453704e-05,
+      "loss": 2.5897,
+      "step": 5970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8745938029200425e-05,
+      "loss": 2.8677,
+      "step": 5980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.874364959494714e-05,
+      "loss": 2.6504,
+      "step": 5990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.874136116069385e-05,
+      "loss": 2.7434,
+      "step": 6000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8739072726440574e-05,
+      "loss": 2.7148,
+      "step": 6010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.873678429218729e-05,
+      "loss": 2.6927,
+      "step": 6020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8734495857934e-05,
+      "loss": 2.54,
+      "step": 6030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.873220742368072e-05,
+      "loss": 2.6091,
+      "step": 6040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.872991898942744e-05,
+      "loss": 2.74,
+      "step": 6050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.872763055517415e-05,
+      "loss": 2.6368,
+      "step": 6060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.872534212092087e-05,
+      "loss": 2.6985,
+      "step": 6070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.872305368666759e-05,
+      "loss": 2.5432,
+      "step": 6080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.87207652524143e-05,
+      "loss": 2.5034,
+      "step": 6090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.871847681816102e-05,
+      "loss": 2.443,
+      "step": 6100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8716188383907736e-05,
+      "loss": 2.6483,
+      "step": 6110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.871389994965445e-05,
+      "loss": 2.6671,
+      "step": 6120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.871161151540117e-05,
+      "loss": 2.8266,
+      "step": 6130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.870932308114788e-05,
+      "loss": 2.6691,
+      "step": 6140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.870703464689459e-05,
+      "loss": 2.6002,
+      "step": 6150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.870474621264131e-05,
+      "loss": 2.6232,
+      "step": 6160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.870245777838803e-05,
+      "loss": 2.7423,
+      "step": 6170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.870016934413474e-05,
+      "loss": 2.6878,
+      "step": 6180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.869788090988146e-05,
+      "loss": 2.6619,
+      "step": 6190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8695592475628177e-05,
+      "loss": 2.7035,
+      "step": 6200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.869330404137489e-05,
+      "loss": 2.6629,
+      "step": 6210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.869101560712161e-05,
+      "loss": 2.5376,
+      "step": 6220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8688727172868326e-05,
+      "loss": 2.6913,
+      "step": 6230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.868643873861504e-05,
+      "loss": 2.7134,
+      "step": 6240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.868415030436176e-05,
+      "loss": 2.5514,
+      "step": 6250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8681861870108475e-05,
+      "loss": 2.7627,
+      "step": 6260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.867957343585519e-05,
+      "loss": 2.5179,
+      "step": 6270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.867728500160191e-05,
+      "loss": 2.7009,
+      "step": 6280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8674996567348624e-05,
+      "loss": 2.6736,
+      "step": 6290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.867270813309534e-05,
+      "loss": 2.6327,
+      "step": 6300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.867041969884206e-05,
+      "loss": 2.489,
+      "step": 6310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.866813126458877e-05,
+      "loss": 2.593,
+      "step": 6320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.866584283033549e-05,
+      "loss": 2.8399,
+      "step": 6330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.86635543960822e-05,
+      "loss": 2.6832,
+      "step": 6340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8661265961828916e-05,
+      "loss": 2.7637,
+      "step": 6350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.865897752757563e-05,
+      "loss": 2.4986,
+      "step": 6360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.865668909332235e-05,
+      "loss": 2.59,
+      "step": 6370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8654400659069065e-05,
+      "loss": 2.655,
+      "step": 6380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.865211222481578e-05,
+      "loss": 2.6297,
+      "step": 6390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.86498237905625e-05,
+      "loss": 2.6105,
+      "step": 6400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8647535356309214e-05,
+      "loss": 2.6415,
+      "step": 6410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.864524692205593e-05,
+      "loss": 2.6693,
+      "step": 6420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.864295848780265e-05,
+      "loss": 2.538,
+      "step": 6430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.864067005354936e-05,
+      "loss": 2.6659,
+      "step": 6440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.863838161929608e-05,
+      "loss": 2.3872,
+      "step": 6450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.86360931850428e-05,
+      "loss": 2.7324,
+      "step": 6460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.863380475078951e-05,
+      "loss": 2.728,
+      "step": 6470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8631516316536227e-05,
+      "loss": 2.7703,
+      "step": 6480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.862922788228295e-05,
+      "loss": 2.7566,
+      "step": 6490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.862693944802966e-05,
+      "loss": 2.6029,
+      "step": 6500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8624651013776376e-05,
+      "loss": 2.5409,
+      "step": 6510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.86223625795231e-05,
+      "loss": 2.8619,
+      "step": 6520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8620074145269804e-05,
+      "loss": 2.3743,
+      "step": 6530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8617785711016525e-05,
+      "loss": 2.7786,
+      "step": 6540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.861549727676324e-05,
+      "loss": 2.7577,
+      "step": 6550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.861320884250995e-05,
+      "loss": 2.6817,
+      "step": 6560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8610920408256674e-05,
+      "loss": 2.5866,
+      "step": 6570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.860863197400339e-05,
+      "loss": 2.4611,
+      "step": 6580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.86063435397501e-05,
+      "loss": 2.5759,
+      "step": 6590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.860405510549682e-05,
+      "loss": 2.4774,
+      "step": 6600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.860176667124354e-05,
+      "loss": 2.6212,
+      "step": 6610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.859947823699025e-05,
+      "loss": 2.756,
+      "step": 6620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.859718980273697e-05,
+      "loss": 2.5809,
+      "step": 6630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8594901368483687e-05,
+      "loss": 2.5641,
+      "step": 6640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.85926129342304e-05,
+      "loss": 2.5527,
+      "step": 6650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.859032449997712e-05,
+      "loss": 2.8218,
+      "step": 6660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8588036065723836e-05,
+      "loss": 2.6231,
+      "step": 6670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.858574763147055e-05,
+      "loss": 2.7367,
+      "step": 6680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.858345919721727e-05,
+      "loss": 2.5321,
+      "step": 6690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8581170762963985e-05,
+      "loss": 2.6097,
+      "step": 6700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.85788823287107e-05,
+      "loss": 2.6572,
+      "step": 6710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.857659389445742e-05,
+      "loss": 2.4732,
+      "step": 6720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.857430546020413e-05,
+      "loss": 2.7225,
+      "step": 6730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.857201702595084e-05,
+      "loss": 2.5497,
+      "step": 6740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.856972859169756e-05,
+      "loss": 2.5199,
+      "step": 6750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8567440157444276e-05,
+      "loss": 2.7116,
+      "step": 6760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.856515172319099e-05,
+      "loss": 2.6121,
+      "step": 6770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.856286328893771e-05,
+      "loss": 2.5073,
+      "step": 6780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8560574854684426e-05,
+      "loss": 2.6674,
+      "step": 6790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.855828642043114e-05,
+      "loss": 2.5039,
+      "step": 6800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.855599798617786e-05,
+      "loss": 2.6449,
+      "step": 6810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8553709551924575e-05,
+      "loss": 2.7008,
+      "step": 6820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.855142111767129e-05,
+      "loss": 2.6295,
+      "step": 6830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.854913268341801e-05,
+      "loss": 2.7345,
+      "step": 6840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8546844249164724e-05,
+      "loss": 2.7104,
+      "step": 6850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.854455581491144e-05,
+      "loss": 2.6052,
+      "step": 6860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.854226738065816e-05,
+      "loss": 2.654,
+      "step": 6870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.853997894640487e-05,
+      "loss": 2.5875,
+      "step": 6880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.853769051215159e-05,
+      "loss": 2.63,
+      "step": 6890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.853540207789831e-05,
+      "loss": 2.603,
+      "step": 6900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.853311364364502e-05,
+      "loss": 2.5952,
+      "step": 6910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8530825209391737e-05,
+      "loss": 2.5097,
+      "step": 6920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.852853677513845e-05,
+      "loss": 2.6311,
+      "step": 6930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8526248340885165e-05,
+      "loss": 2.5162,
+      "step": 6940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8523959906631886e-05,
+      "loss": 2.5743,
+      "step": 6950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.85216714723786e-05,
+      "loss": 2.5102,
+      "step": 6960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8519383038125314e-05,
+      "loss": 2.5564,
+      "step": 6970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8517094603872035e-05,
+      "loss": 2.7323,
+      "step": 6980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.851480616961875e-05,
+      "loss": 2.6003,
+      "step": 6990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.851251773536546e-05,
+      "loss": 2.6922,
+      "step": 7000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8510229301112184e-05,
+      "loss": 2.6887,
+      "step": 7010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.85079408668589e-05,
+      "loss": 2.7063,
+      "step": 7020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.850565243260561e-05,
+      "loss": 2.7458,
+      "step": 7030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.850336399835233e-05,
+      "loss": 2.6305,
+      "step": 7040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.850107556409905e-05,
+      "loss": 2.6088,
+      "step": 7050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.849878712984576e-05,
+      "loss": 2.6396,
+      "step": 7060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.849649869559248e-05,
+      "loss": 2.6442,
+      "step": 7070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8494210261339197e-05,
+      "loss": 2.5578,
+      "step": 7080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.849192182708591e-05,
+      "loss": 2.6304,
+      "step": 7090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.848963339283263e-05,
+      "loss": 2.7613,
+      "step": 7100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8487344958579346e-05,
+      "loss": 2.5997,
+      "step": 7110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.848505652432605e-05,
+      "loss": 2.5993,
+      "step": 7120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8482768090072774e-05,
+      "loss": 2.4996,
+      "step": 7130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.848047965581949e-05,
+      "loss": 2.604,
+      "step": 7140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.84781912215662e-05,
+      "loss": 2.6322,
+      "step": 7150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.847590278731292e-05,
+      "loss": 2.623,
+      "step": 7160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.847361435305964e-05,
+      "loss": 2.6787,
+      "step": 7170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.847132591880635e-05,
+      "loss": 2.5988,
+      "step": 7180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.846903748455307e-05,
+      "loss": 2.5075,
+      "step": 7190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8466749050299786e-05,
+      "loss": 2.5275,
+      "step": 7200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.84644606160465e-05,
+      "loss": 2.5091,
+      "step": 7210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.846217218179322e-05,
+      "loss": 2.7113,
+      "step": 7220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8459883747539936e-05,
+      "loss": 2.5282,
+      "step": 7230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.845759531328665e-05,
+      "loss": 2.7646,
+      "step": 7240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.845530687903337e-05,
+      "loss": 2.4846,
+      "step": 7250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8453018444780085e-05,
+      "loss": 2.5807,
+      "step": 7260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.84507300105268e-05,
+      "loss": 2.5587,
+      "step": 7270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.844844157627352e-05,
+      "loss": 2.5217,
+      "step": 7280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8446153142020234e-05,
+      "loss": 2.4994,
+      "step": 7290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.844386470776695e-05,
+      "loss": 2.6186,
+      "step": 7300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.844157627351367e-05,
+      "loss": 2.6483,
+      "step": 7310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8439287839260376e-05,
+      "loss": 2.5136,
+      "step": 7320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.84369994050071e-05,
+      "loss": 2.5561,
+      "step": 7330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.843471097075381e-05,
+      "loss": 2.6446,
+      "step": 7340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8432422536500526e-05,
+      "loss": 2.5345,
+      "step": 7350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8430134102247246e-05,
+      "loss": 2.4426,
+      "step": 7360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.842784566799396e-05,
+      "loss": 2.4672,
+      "step": 7370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8425557233740675e-05,
+      "loss": 2.4849,
+      "step": 7380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8423268799487396e-05,
+      "loss": 2.6723,
+      "step": 7390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.842098036523411e-05,
+      "loss": 2.4709,
+      "step": 7400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8418691930980824e-05,
+      "loss": 2.3804,
+      "step": 7410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8416403496727545e-05,
+      "loss": 2.3274,
+      "step": 7420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.841411506247426e-05,
+      "loss": 2.491,
+      "step": 7430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.841182662822097e-05,
+      "loss": 2.5782,
+      "step": 7440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8409538193967694e-05,
+      "loss": 2.7523,
+      "step": 7450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.840724975971441e-05,
+      "loss": 2.4707,
+      "step": 7460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.840496132546112e-05,
+      "loss": 2.479,
+      "step": 7470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8402672891207836e-05,
+      "loss": 2.5186,
+      "step": 7480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.840038445695456e-05,
+      "loss": 2.4696,
+      "step": 7490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.839809602270127e-05,
+      "loss": 2.5089,
+      "step": 7500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8395807588447986e-05,
+      "loss": 2.5794,
+      "step": 7510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.83935191541947e-05,
+      "loss": 2.4805,
+      "step": 7520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8391230719941414e-05,
+      "loss": 2.5781,
+      "step": 7530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8388942285688135e-05,
+      "loss": 2.5655,
+      "step": 7540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.838665385143485e-05,
+      "loss": 2.531,
+      "step": 7550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.838436541718156e-05,
+      "loss": 2.7172,
+      "step": 7560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8382076982928284e-05,
+      "loss": 2.6663,
+      "step": 7570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8379788548675e-05,
+      "loss": 2.5828,
+      "step": 7580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.837750011442171e-05,
+      "loss": 2.588,
+      "step": 7590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.837521168016843e-05,
+      "loss": 2.4599,
+      "step": 7600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.837292324591515e-05,
+      "loss": 2.6646,
+      "step": 7610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.837063481166186e-05,
+      "loss": 2.6683,
+      "step": 7620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.836834637740858e-05,
+      "loss": 2.6381,
+      "step": 7630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8366057943155296e-05,
+      "loss": 2.6133,
+      "step": 7640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.836376950890201e-05,
+      "loss": 2.5615,
+      "step": 7650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.836148107464873e-05,
+      "loss": 2.6379,
+      "step": 7660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8359192640395446e-05,
+      "loss": 2.5443,
+      "step": 7670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.835690420614216e-05,
+      "loss": 2.5427,
+      "step": 7680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.835461577188888e-05,
+      "loss": 2.574,
+      "step": 7690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8352327337635595e-05,
+      "loss": 2.6668,
+      "step": 7700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.835003890338231e-05,
+      "loss": 2.6152,
+      "step": 7710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.834775046912902e-05,
+      "loss": 2.5713,
+      "step": 7720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.834546203487574e-05,
+      "loss": 2.5762,
+      "step": 7730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.834317360062245e-05,
+      "loss": 2.5321,
+      "step": 7740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.834088516636917e-05,
+      "loss": 2.5608,
+      "step": 7750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8338596732115886e-05,
+      "loss": 2.6055,
+      "step": 7760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.83363082978626e-05,
+      "loss": 2.5336,
+      "step": 7770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.833401986360932e-05,
+      "loss": 2.527,
+      "step": 7780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8331731429356036e-05,
+      "loss": 2.6777,
+      "step": 7790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.832944299510275e-05,
+      "loss": 2.5895,
+      "step": 7800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.832715456084947e-05,
+      "loss": 2.4565,
+      "step": 7810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8324866126596185e-05,
+      "loss": 2.6261,
+      "step": 7820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.83225776923429e-05,
+      "loss": 2.6804,
+      "step": 7830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.832028925808962e-05,
+      "loss": 2.5074,
+      "step": 7840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8318000823836334e-05,
+      "loss": 2.4658,
+      "step": 7850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.831571238958305e-05,
+      "loss": 2.4826,
+      "step": 7860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.831342395532977e-05,
+      "loss": 2.5284,
+      "step": 7870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.831113552107648e-05,
+      "loss": 2.5864,
+      "step": 7880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.83088470868232e-05,
+      "loss": 2.4288,
+      "step": 7890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.830655865256992e-05,
+      "loss": 2.5494,
+      "step": 7900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8304270218316625e-05,
+      "loss": 2.5612,
+      "step": 7910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8301981784063346e-05,
+      "loss": 2.5314,
+      "step": 7920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.829969334981006e-05,
+      "loss": 2.6902,
+      "step": 7930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8297404915556775e-05,
+      "loss": 2.6755,
+      "step": 7940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8295116481303496e-05,
+      "loss": 2.6582,
+      "step": 7950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.829282804705021e-05,
+      "loss": 2.6417,
+      "step": 7960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8290539612796924e-05,
+      "loss": 2.5043,
+      "step": 7970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8288251178543645e-05,
+      "loss": 2.5601,
+      "step": 7980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.828596274429036e-05,
+      "loss": 2.5943,
+      "step": 7990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.828367431003707e-05,
+      "loss": 2.7184,
+      "step": 8000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8281385875783794e-05,
+      "loss": 2.4364,
+      "step": 8010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.827909744153051e-05,
+      "loss": 2.4442,
+      "step": 8020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.827680900727722e-05,
+      "loss": 2.5539,
+      "step": 8030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.827452057302394e-05,
+      "loss": 2.5392,
+      "step": 8040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.827223213877066e-05,
+      "loss": 2.5492,
+      "step": 8050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.826994370451737e-05,
+      "loss": 2.6908,
+      "step": 8060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.826765527026409e-05,
+      "loss": 2.6334,
+      "step": 8070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8265366836010806e-05,
+      "loss": 2.5766,
+      "step": 8080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.826307840175752e-05,
+      "loss": 2.5178,
+      "step": 8090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.826078996750424e-05,
+      "loss": 2.5681,
+      "step": 8100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.825850153325095e-05,
+      "loss": 2.586,
+      "step": 8110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.825621309899766e-05,
+      "loss": 2.5505,
+      "step": 8120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8253924664744384e-05,
+      "loss": 2.6139,
+      "step": 8130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.82516362304911e-05,
+      "loss": 2.6023,
+      "step": 8140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.824934779623781e-05,
+      "loss": 2.5252,
+      "step": 8150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.824705936198453e-05,
+      "loss": 2.7363,
+      "step": 8160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.824477092773125e-05,
+      "loss": 2.5805,
+      "step": 8170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.824248249347796e-05,
+      "loss": 2.46,
+      "step": 8180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.824019405922468e-05,
+      "loss": 2.6099,
+      "step": 8190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8237905624971396e-05,
+      "loss": 2.5576,
+      "step": 8200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.823561719071811e-05,
+      "loss": 2.6261,
+      "step": 8210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.823332875646483e-05,
+      "loss": 2.4407,
+      "step": 8220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8231040322211546e-05,
+      "loss": 2.6228,
+      "step": 8230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.822875188795826e-05,
+      "loss": 2.5148,
+      "step": 8240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.822646345370498e-05,
+      "loss": 2.6236,
+      "step": 8250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8224175019451695e-05,
+      "loss": 2.5435,
+      "step": 8260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.822188658519841e-05,
+      "loss": 2.4203,
+      "step": 8270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.821959815094513e-05,
+      "loss": 2.5873,
+      "step": 8280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8217309716691844e-05,
+      "loss": 2.7855,
+      "step": 8290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.821502128243856e-05,
+      "loss": 2.4898,
+      "step": 8300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.821273284818527e-05,
+      "loss": 2.5695,
+      "step": 8310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8210444413931986e-05,
+      "loss": 2.4473,
+      "step": 8320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.820815597967871e-05,
+      "loss": 2.4651,
+      "step": 8330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.820586754542542e-05,
+      "loss": 2.4806,
+      "step": 8340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8203579111172135e-05,
+      "loss": 2.6206,
+      "step": 8350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8201290676918856e-05,
+      "loss": 2.4872,
+      "step": 8360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.819900224266557e-05,
+      "loss": 2.6685,
+      "step": 8370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8196713808412285e-05,
+      "loss": 2.5091,
+      "step": 8380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8194425374159006e-05,
+      "loss": 2.5531,
+      "step": 8390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.819213693990572e-05,
+      "loss": 2.6317,
+      "step": 8400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8189848505652434e-05,
+      "loss": 2.4965,
+      "step": 8410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8187560071399155e-05,
+      "loss": 2.6153,
+      "step": 8420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.818527163714587e-05,
+      "loss": 2.5401,
+      "step": 8430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.818298320289258e-05,
+      "loss": 2.4877,
+      "step": 8440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8180694768639304e-05,
+      "loss": 2.5317,
+      "step": 8450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.817840633438602e-05,
+      "loss": 2.508,
+      "step": 8460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.817611790013273e-05,
+      "loss": 2.5201,
+      "step": 8470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.817382946587945e-05,
+      "loss": 2.5711,
+      "step": 8480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.817154103162617e-05,
+      "loss": 2.5767,
+      "step": 8490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.816925259737288e-05,
+      "loss": 2.4583,
+      "step": 8500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8166964163119595e-05,
+      "loss": 2.5907,
+      "step": 8510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.816467572886631e-05,
+      "loss": 2.7128,
+      "step": 8520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8162387294613024e-05,
+      "loss": 2.6016,
+      "step": 8530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8160098860359745e-05,
+      "loss": 2.5184,
+      "step": 8540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.815781042610646e-05,
+      "loss": 2.5818,
+      "step": 8550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.815552199185317e-05,
+      "loss": 2.5131,
+      "step": 8560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8153233557599894e-05,
+      "loss": 2.4645,
+      "step": 8570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.815094512334661e-05,
+      "loss": 2.5413,
+      "step": 8580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.814865668909332e-05,
+      "loss": 2.4652,
+      "step": 8590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.814636825484004e-05,
+      "loss": 2.6726,
+      "step": 8600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.814407982058676e-05,
+      "loss": 2.5531,
+      "step": 8610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.814179138633347e-05,
+      "loss": 2.4753,
+      "step": 8620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.813950295208019e-05,
+      "loss": 2.4839,
+      "step": 8630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8137214517826906e-05,
+      "loss": 2.5354,
+      "step": 8640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.813492608357362e-05,
+      "loss": 2.4137,
+      "step": 8650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.813263764932034e-05,
+      "loss": 2.4872,
+      "step": 8660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8130349215067055e-05,
+      "loss": 2.7044,
+      "step": 8670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.812806078081377e-05,
+      "loss": 2.5203,
+      "step": 8680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.812577234656049e-05,
+      "loss": 2.5716,
+      "step": 8690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.81234839123072e-05,
+      "loss": 2.4747,
+      "step": 8700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.812119547805391e-05,
+      "loss": 2.4187,
+      "step": 8710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.811890704380063e-05,
+      "loss": 2.5006,
+      "step": 8720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.811661860954735e-05,
+      "loss": 2.3529,
+      "step": 8730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.811433017529406e-05,
+      "loss": 2.4003,
+      "step": 8740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.811204174104078e-05,
+      "loss": 2.5774,
+      "step": 8750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8109753306787496e-05,
+      "loss": 2.5144,
+      "step": 8760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.810746487253421e-05,
+      "loss": 2.4883,
+      "step": 8770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.810517643828093e-05,
+      "loss": 2.5076,
+      "step": 8780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8102888004027645e-05,
+      "loss": 2.4752,
+      "step": 8790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.810059956977436e-05,
+      "loss": 2.5049,
+      "step": 8800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.809831113552108e-05,
+      "loss": 2.5886,
+      "step": 8810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8096022701267795e-05,
+      "loss": 2.478,
+      "step": 8820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.809373426701451e-05,
+      "loss": 2.4351,
+      "step": 8830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.809144583276123e-05,
+      "loss": 2.3795,
+      "step": 8840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8089157398507944e-05,
+      "loss": 2.5045,
+      "step": 8850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.808686896425466e-05,
+      "loss": 2.5118,
+      "step": 8860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.808458053000138e-05,
+      "loss": 2.4221,
+      "step": 8870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.808229209574809e-05,
+      "loss": 2.4182,
+      "step": 8880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.808000366149481e-05,
+      "loss": 2.5603,
+      "step": 8890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.807771522724152e-05,
+      "loss": 2.642,
+      "step": 8900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8075426792988235e-05,
+      "loss": 2.5941,
+      "step": 8910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8073138358734956e-05,
+      "loss": 2.6514,
+      "step": 8920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.807084992448167e-05,
+      "loss": 2.6656,
+      "step": 8930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8068561490228385e-05,
+      "loss": 2.6024,
+      "step": 8940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8066273055975105e-05,
+      "loss": 2.6309,
+      "step": 8950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.806398462172182e-05,
+      "loss": 2.4661,
+      "step": 8960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8061696187468534e-05,
+      "loss": 2.4152,
+      "step": 8970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8059407753215255e-05,
+      "loss": 2.4096,
+      "step": 8980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.805711931896197e-05,
+      "loss": 2.4659,
+      "step": 8990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.805483088470868e-05,
+      "loss": 2.3591,
+      "step": 9000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8052542450455404e-05,
+      "loss": 2.6971,
+      "step": 9010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.805025401620212e-05,
+      "loss": 2.5978,
+      "step": 9020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.804796558194883e-05,
+      "loss": 2.5958,
+      "step": 9030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.804567714769555e-05,
+      "loss": 2.5052,
+      "step": 9040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.804338871344227e-05,
+      "loss": 2.5074,
+      "step": 9050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.804110027918898e-05,
+      "loss": 2.442,
+      "step": 9060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.80388118449357e-05,
+      "loss": 2.4899,
+      "step": 9070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8036523410682416e-05,
+      "loss": 2.5669,
+      "step": 9080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.803423497642913e-05,
+      "loss": 2.5385,
+      "step": 9090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8031946542175845e-05,
+      "loss": 2.5921,
+      "step": 9100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.802965810792256e-05,
+      "loss": 2.5148,
+      "step": 9110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.802736967366927e-05,
+      "loss": 2.6347,
+      "step": 9120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8025081239415994e-05,
+      "loss": 2.4504,
+      "step": 9130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.802279280516271e-05,
+      "loss": 2.4635,
+      "step": 9140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.802050437090942e-05,
+      "loss": 2.3694,
+      "step": 9150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.801821593665614e-05,
+      "loss": 2.4794,
+      "step": 9160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.801592750240286e-05,
+      "loss": 2.4176,
+      "step": 9170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.801363906814957e-05,
+      "loss": 2.4976,
+      "step": 9180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.801135063389629e-05,
+      "loss": 2.6711,
+      "step": 9190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8009062199643006e-05,
+      "loss": 2.4859,
+      "step": 9200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.800677376538972e-05,
+      "loss": 2.5434,
+      "step": 9210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.800448533113644e-05,
+      "loss": 2.4095,
+      "step": 9220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8002196896883155e-05,
+      "loss": 2.4597,
+      "step": 9230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.799990846262987e-05,
+      "loss": 2.3098,
+      "step": 9240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.799762002837659e-05,
+      "loss": 2.5524,
+      "step": 9250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7995331594123305e-05,
+      "loss": 2.3824,
+      "step": 9260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.799304315987002e-05,
+      "loss": 2.6007,
+      "step": 9270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.799075472561674e-05,
+      "loss": 2.4054,
+      "step": 9280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7988466291363454e-05,
+      "loss": 2.5336,
+      "step": 9290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.798617785711017e-05,
+      "loss": 2.4023,
+      "step": 9300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.798388942285688e-05,
+      "loss": 2.3675,
+      "step": 9310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7981600988603596e-05,
+      "loss": 2.3319,
+      "step": 9320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.797931255435032e-05,
+      "loss": 2.5489,
+      "step": 9330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.797702412009703e-05,
+      "loss": 2.5419,
+      "step": 9340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7974735685843745e-05,
+      "loss": 2.3948,
+      "step": 9350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7972447251590466e-05,
+      "loss": 2.3667,
+      "step": 9360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.797015881733718e-05,
+      "loss": 2.6186,
+      "step": 9370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7967870383083894e-05,
+      "loss": 2.4423,
+      "step": 9380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7965581948830615e-05,
+      "loss": 2.331,
+      "step": 9390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.796329351457733e-05,
+      "loss": 2.4957,
+      "step": 9400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7961005080324044e-05,
+      "loss": 2.6605,
+      "step": 9410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7958716646070765e-05,
+      "loss": 2.6334,
+      "step": 9420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.795642821181748e-05,
+      "loss": 2.5243,
+      "step": 9430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.795413977756419e-05,
+      "loss": 2.503,
+      "step": 9440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7951851343310914e-05,
+      "loss": 2.5767,
+      "step": 9450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.794956290905763e-05,
+      "loss": 2.5174,
+      "step": 9460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.794727447480434e-05,
+      "loss": 2.4903,
+      "step": 9470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.794498604055106e-05,
+      "loss": 2.592,
+      "step": 9480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.794269760629777e-05,
+      "loss": 2.5382,
+      "step": 9490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7940409172044484e-05,
+      "loss": 2.5052,
+      "step": 9500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7938120737791205e-05,
+      "loss": 2.3965,
+      "step": 9510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.793583230353792e-05,
+      "loss": 2.6111,
+      "step": 9520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7933543869284634e-05,
+      "loss": 2.5139,
+      "step": 9530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7931255435031355e-05,
+      "loss": 2.3385,
+      "step": 9540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.792896700077807e-05,
+      "loss": 2.406,
+      "step": 9550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.792667856652478e-05,
+      "loss": 2.3884,
+      "step": 9560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7924390132271504e-05,
+      "loss": 2.4414,
+      "step": 9570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.792210169801822e-05,
+      "loss": 2.4814,
+      "step": 9580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.791981326376493e-05,
+      "loss": 2.5667,
+      "step": 9590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.791752482951165e-05,
+      "loss": 2.5959,
+      "step": 9600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.791523639525837e-05,
+      "loss": 2.239,
+      "step": 9610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.791294796100508e-05,
+      "loss": 2.5778,
+      "step": 9620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.79106595267518e-05,
+      "loss": 2.3503,
+      "step": 9630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7908371092498516e-05,
+      "loss": 2.5121,
+      "step": 9640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.790608265824523e-05,
+      "loss": 2.3898,
+      "step": 9650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.790379422399195e-05,
+      "loss": 2.5714,
+      "step": 9660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7901505789738665e-05,
+      "loss": 2.4222,
+      "step": 9670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.789921735548538e-05,
+      "loss": 2.5705,
+      "step": 9680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7896928921232094e-05,
+      "loss": 2.4389,
+      "step": 9690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.789464048697881e-05,
+      "loss": 2.4607,
+      "step": 9700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.789235205272553e-05,
+      "loss": 2.4833,
+      "step": 9710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.789006361847224e-05,
+      "loss": 2.6475,
+      "step": 9720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.788777518421896e-05,
+      "loss": 2.5329,
+      "step": 9730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.788548674996568e-05,
+      "loss": 2.464,
+      "step": 9740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.788319831571239e-05,
+      "loss": 2.4013,
+      "step": 9750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7880909881459106e-05,
+      "loss": 2.4784,
+      "step": 9760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.787862144720583e-05,
+      "loss": 2.4445,
+      "step": 9770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.787633301295254e-05,
+      "loss": 2.5219,
+      "step": 9780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7874044578699255e-05,
+      "loss": 2.6752,
+      "step": 9790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.787175614444597e-05,
+      "loss": 2.4733,
+      "step": 9800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.786946771019269e-05,
+      "loss": 2.3437,
+      "step": 9810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7867179275939404e-05,
+      "loss": 2.652,
+      "step": 9820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.786489084168612e-05,
+      "loss": 2.47,
+      "step": 9830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.786260240743284e-05,
+      "loss": 2.5827,
+      "step": 9840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7860313973179554e-05,
+      "loss": 2.7081,
+      "step": 9850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.785802553892627e-05,
+      "loss": 2.5675,
+      "step": 9860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.785573710467299e-05,
+      "loss": 2.5603,
+      "step": 9870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.78534486704197e-05,
+      "loss": 2.4665,
+      "step": 9880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.785116023616642e-05,
+      "loss": 2.5384,
+      "step": 9890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.784887180191313e-05,
+      "loss": 2.4818,
+      "step": 9900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7846583367659845e-05,
+      "loss": 2.4326,
+      "step": 9910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7844294933406566e-05,
+      "loss": 2.4038,
+      "step": 9920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.784200649915328e-05,
+      "loss": 2.3771,
+      "step": 9930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7839718064899994e-05,
+      "loss": 2.5238,
+      "step": 9940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7837429630646715e-05,
+      "loss": 2.6072,
+      "step": 9950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.783514119639343e-05,
+      "loss": 2.5663,
+      "step": 9960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7832852762140144e-05,
+      "loss": 2.5498,
+      "step": 9970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7830564327886864e-05,
+      "loss": 2.5463,
+      "step": 9980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.782827589363358e-05,
+      "loss": 2.6455,
+      "step": 9990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.782598745938029e-05,
+      "loss": 2.6951,
+      "step": 10000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7823699025127014e-05,
+      "loss": 2.5799,
+      "step": 10010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.782141059087373e-05,
+      "loss": 2.4259,
+      "step": 10020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.781912215662044e-05,
+      "loss": 2.2816,
+      "step": 10030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.781683372236716e-05,
+      "loss": 2.5339,
+      "step": 10040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.781454528811388e-05,
+      "loss": 2.6723,
+      "step": 10050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.781225685386059e-05,
+      "loss": 2.6393,
+      "step": 10060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.780996841960731e-05,
+      "loss": 2.4065,
+      "step": 10070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7807679985354026e-05,
+      "loss": 2.4601,
+      "step": 10080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7805391551100733e-05,
+      "loss": 2.7115,
+      "step": 10090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7803103116847454e-05,
+      "loss": 2.4182,
+      "step": 10100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.780081468259417e-05,
+      "loss": 2.4533,
+      "step": 10110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.779852624834088e-05,
+      "loss": 2.4312,
+      "step": 10120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7796237814087604e-05,
+      "loss": 2.4738,
+      "step": 10130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.779394937983432e-05,
+      "loss": 2.5474,
+      "step": 10140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.779166094558103e-05,
+      "loss": 2.328,
+      "step": 10150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.778937251132775e-05,
+      "loss": 2.4257,
+      "step": 10160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.778708407707447e-05,
+      "loss": 2.4171,
+      "step": 10170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.778479564282118e-05,
+      "loss": 2.4215,
+      "step": 10180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.77825072085679e-05,
+      "loss": 2.4224,
+      "step": 10190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7780218774314616e-05,
+      "loss": 2.5248,
+      "step": 10200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.777793034006133e-05,
+      "loss": 2.4769,
+      "step": 10210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.777564190580805e-05,
+      "loss": 2.5698,
+      "step": 10220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7773353471554765e-05,
+      "loss": 2.5372,
+      "step": 10230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.777106503730148e-05,
+      "loss": 2.486,
+      "step": 10240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.77687766030482e-05,
+      "loss": 2.4284,
+      "step": 10250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7766488168794914e-05,
+      "loss": 2.5408,
+      "step": 10260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.776419973454163e-05,
+      "loss": 2.5585,
+      "step": 10270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.776191130028834e-05,
+      "loss": 2.5371,
+      "step": 10280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.775962286603506e-05,
+      "loss": 2.5955,
+      "step": 10290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.775733443178178e-05,
+      "loss": 2.4826,
+      "step": 10300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.775504599752849e-05,
+      "loss": 2.4955,
+      "step": 10310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7752757563275206e-05,
+      "loss": 2.5477,
+      "step": 10320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.775046912902193e-05,
+      "loss": 2.3415,
+      "step": 10330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.774818069476864e-05,
+      "loss": 2.4569,
+      "step": 10340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7745892260515355e-05,
+      "loss": 2.4084,
+      "step": 10350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7743603826262076e-05,
+      "loss": 2.2895,
+      "step": 10360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.774131539200879e-05,
+      "loss": 2.4724,
+      "step": 10370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7739026957755504e-05,
+      "loss": 2.5041,
+      "step": 10380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7736738523502225e-05,
+      "loss": 2.3171,
+      "step": 10390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.773445008924894e-05,
+      "loss": 2.5456,
+      "step": 10400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7732161654995654e-05,
+      "loss": 2.5014,
+      "step": 10410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7729873220742374e-05,
+      "loss": 2.4247,
+      "step": 10420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.772758478648909e-05,
+      "loss": 2.5736,
+      "step": 10430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.77252963522358e-05,
+      "loss": 2.532,
+      "step": 10440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7723007917982524e-05,
+      "loss": 2.4522,
+      "step": 10450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.772071948372924e-05,
+      "loss": 2.4223,
+      "step": 10460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.771843104947595e-05,
+      "loss": 2.3001,
+      "step": 10470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7716142615222666e-05,
+      "loss": 2.4276,
+      "step": 10480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.771385418096938e-05,
+      "loss": 2.5513,
+      "step": 10490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7711565746716094e-05,
+      "loss": 2.3925,
+      "step": 10500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7709277312462815e-05,
+      "loss": 2.441,
+      "step": 10510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.770698887820953e-05,
+      "loss": 2.409,
+      "step": 10520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7704700443956243e-05,
+      "loss": 2.4389,
+      "step": 10530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7702412009702964e-05,
+      "loss": 2.4297,
+      "step": 10540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.770012357544968e-05,
+      "loss": 2.2659,
+      "step": 10550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.769783514119639e-05,
+      "loss": 2.3624,
+      "step": 10560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7695546706943114e-05,
+      "loss": 2.4643,
+      "step": 10570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.769325827268983e-05,
+      "loss": 2.363,
+      "step": 10580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.769096983843654e-05,
+      "loss": 2.5992,
+      "step": 10590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.768868140418326e-05,
+      "loss": 2.4871,
+      "step": 10600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.768639296992998e-05,
+      "loss": 2.4541,
+      "step": 10610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.768410453567669e-05,
+      "loss": 2.4925,
+      "step": 10620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.768181610142341e-05,
+      "loss": 2.5078,
+      "step": 10630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7679527667170126e-05,
+      "loss": 2.52,
+      "step": 10640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.767723923291684e-05,
+      "loss": 2.4908,
+      "step": 10650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.767495079866356e-05,
+      "loss": 2.4405,
+      "step": 10660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7672662364410275e-05,
+      "loss": 2.521,
+      "step": 10670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.767037393015699e-05,
+      "loss": 2.4349,
+      "step": 10680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7668085495903703e-05,
+      "loss": 2.3396,
+      "step": 10690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.766579706165042e-05,
+      "loss": 2.2763,
+      "step": 10700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.766350862739714e-05,
+      "loss": 2.421,
+      "step": 10710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.766122019314385e-05,
+      "loss": 2.4553,
+      "step": 10720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.765893175889057e-05,
+      "loss": 2.3664,
+      "step": 10730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.765664332463729e-05,
+      "loss": 2.5308,
+      "step": 10740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7654354890384e-05,
+      "loss": 2.3722,
+      "step": 10750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7652066456130716e-05,
+      "loss": 2.4598,
+      "step": 10760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.764977802187744e-05,
+      "loss": 2.4391,
+      "step": 10770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.764748958762415e-05,
+      "loss": 2.5109,
+      "step": 10780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7645201153370865e-05,
+      "loss": 2.6127,
+      "step": 10790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7642912719117586e-05,
+      "loss": 2.2723,
+      "step": 10800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.76406242848643e-05,
+      "loss": 2.4474,
+      "step": 10810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7638335850611014e-05,
+      "loss": 2.4624,
+      "step": 10820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7636047416357735e-05,
+      "loss": 2.3981,
+      "step": 10830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.763375898210445e-05,
+      "loss": 2.5574,
+      "step": 10840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7631470547851163e-05,
+      "loss": 2.4176,
+      "step": 10850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7629182113597884e-05,
+      "loss": 2.3263,
+      "step": 10860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.76268936793446e-05,
+      "loss": 2.2531,
+      "step": 10870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7624605245091306e-05,
+      "loss": 2.3908,
+      "step": 10880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.762231681083803e-05,
+      "loss": 2.6001,
+      "step": 10890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.762002837658474e-05,
+      "loss": 2.4099,
+      "step": 10900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7617739942331455e-05,
+      "loss": 2.4761,
+      "step": 10910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7615451508078176e-05,
+      "loss": 2.4877,
+      "step": 10920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.761316307382489e-05,
+      "loss": 2.3782,
+      "step": 10930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7610874639571604e-05,
+      "loss": 2.4546,
+      "step": 10940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7608586205318325e-05,
+      "loss": 2.3586,
+      "step": 10950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.760629777106504e-05,
+      "loss": 2.488,
+      "step": 10960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7604009336811753e-05,
+      "loss": 2.6488,
+      "step": 10970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7601720902558474e-05,
+      "loss": 2.3287,
+      "step": 10980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.759943246830519e-05,
+      "loss": 2.426,
+      "step": 10990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.75971440340519e-05,
+      "loss": 2.5453,
+      "step": 11000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7594855599798624e-05,
+      "loss": 2.5256,
+      "step": 11010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.759256716554534e-05,
+      "loss": 2.6035,
+      "step": 11020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.759027873129205e-05,
+      "loss": 2.3711,
+      "step": 11030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.758799029703877e-05,
+      "loss": 2.279,
+      "step": 11040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.758570186278549e-05,
+      "loss": 2.3476,
+      "step": 11050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.75834134285322e-05,
+      "loss": 2.3864,
+      "step": 11060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7581124994278915e-05,
+      "loss": 2.4579,
+      "step": 11070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.757883656002563e-05,
+      "loss": 2.251,
+      "step": 11080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.757654812577234e-05,
+      "loss": 2.3625,
+      "step": 11090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7574259691519064e-05,
+      "loss": 2.4586,
+      "step": 11100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.757197125726578e-05,
+      "loss": 2.5331,
+      "step": 11110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.756968282301249e-05,
+      "loss": 2.4914,
+      "step": 11120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7567394388759213e-05,
+      "loss": 2.4793,
+      "step": 11130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.756510595450593e-05,
+      "loss": 2.3124,
+      "step": 11140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.756281752025264e-05,
+      "loss": 2.3458,
+      "step": 11150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.756052908599936e-05,
+      "loss": 2.4674,
+      "step": 11160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.755824065174608e-05,
+      "loss": 2.428,
+      "step": 11170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.755595221749279e-05,
+      "loss": 2.5912,
+      "step": 11180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.755366378323951e-05,
+      "loss": 2.5087,
+      "step": 11190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7551375348986226e-05,
+      "loss": 2.4712,
+      "step": 11200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.754908691473294e-05,
+      "loss": 2.5694,
+      "step": 11210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.754679848047966e-05,
+      "loss": 2.6408,
+      "step": 11220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7544510046226375e-05,
+      "loss": 2.4675,
+      "step": 11230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.754222161197309e-05,
+      "loss": 2.3694,
+      "step": 11240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.753993317771981e-05,
+      "loss": 2.4658,
+      "step": 11250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7537644743466524e-05,
+      "loss": 2.4074,
+      "step": 11260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.753535630921324e-05,
+      "loss": 2.4344,
+      "step": 11270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.753306787495995e-05,
+      "loss": 2.387,
+      "step": 11280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.753077944070667e-05,
+      "loss": 2.269,
+      "step": 11290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.752849100645339e-05,
+      "loss": 2.4307,
+      "step": 11300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.75262025722001e-05,
+      "loss": 2.5995,
+      "step": 11310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7523914137946816e-05,
+      "loss": 2.4835,
+      "step": 11320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.752162570369354e-05,
+      "loss": 2.5428,
+      "step": 11330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.751933726944025e-05,
+      "loss": 2.4802,
+      "step": 11340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7517048835186965e-05,
+      "loss": 2.452,
+      "step": 11350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7514760400933686e-05,
+      "loss": 2.2569,
+      "step": 11360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.75124719666804e-05,
+      "loss": 2.5582,
+      "step": 11370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7510183532427114e-05,
+      "loss": 2.4104,
+      "step": 11380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7507895098173835e-05,
+      "loss": 2.4496,
+      "step": 11390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.750560666392055e-05,
+      "loss": 2.4586,
+      "step": 11400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7503318229667263e-05,
+      "loss": 2.4268,
+      "step": 11410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7501029795413984e-05,
+      "loss": 2.3161,
+      "step": 11420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.74987413611607e-05,
+      "loss": 2.4652,
+      "step": 11430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.749645292690741e-05,
+      "loss": 2.3014,
+      "step": 11440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7494164492654133e-05,
+      "loss": 2.4633,
+      "step": 11450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.749187605840085e-05,
+      "loss": 2.3784,
+      "step": 11460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7489587624147555e-05,
+      "loss": 2.5262,
+      "step": 11470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7487299189894276e-05,
+      "loss": 2.4124,
+      "step": 11480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.748501075564099e-05,
+      "loss": 2.4753,
+      "step": 11490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7482722321387704e-05,
+      "loss": 2.4224,
+      "step": 11500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7480433887134425e-05,
+      "loss": 2.4929,
+      "step": 11510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.747814545288114e-05,
+      "loss": 2.3438,
+      "step": 11520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.747585701862785e-05,
+      "loss": 2.4286,
+      "step": 11530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7473568584374574e-05,
+      "loss": 2.3516,
+      "step": 11540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.747128015012129e-05,
+      "loss": 2.4137,
+      "step": 11550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7468991715868e-05,
+      "loss": 2.3481,
+      "step": 11560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7466703281614723e-05,
+      "loss": 2.2856,
+      "step": 11570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.746441484736144e-05,
+      "loss": 2.4066,
+      "step": 11580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.746212641310815e-05,
+      "loss": 2.2657,
+      "step": 11590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.745983797885487e-05,
+      "loss": 2.4656,
+      "step": 11600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.745754954460159e-05,
+      "loss": 2.5749,
+      "step": 11610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.74552611103483e-05,
+      "loss": 2.4728,
+      "step": 11620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.745297267609502e-05,
+      "loss": 2.4439,
+      "step": 11630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7450684241841736e-05,
+      "loss": 2.3964,
+      "step": 11640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.744839580758845e-05,
+      "loss": 2.3424,
+      "step": 11650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.744610737333517e-05,
+      "loss": 2.5954,
+      "step": 11660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.744381893908188e-05,
+      "loss": 2.3345,
+      "step": 11670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.74415305048286e-05,
+      "loss": 2.3142,
+      "step": 11680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.743924207057531e-05,
+      "loss": 2.5515,
+      "step": 11690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.743695363632203e-05,
+      "loss": 2.4432,
+      "step": 11700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.743466520206875e-05,
+      "loss": 2.459,
+      "step": 11710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.743237676781546e-05,
+      "loss": 2.5662,
+      "step": 11720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.743008833356218e-05,
+      "loss": 2.4513,
+      "step": 11730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.74277998993089e-05,
+      "loss": 2.3803,
+      "step": 11740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.742551146505561e-05,
+      "loss": 2.4526,
+      "step": 11750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7423223030802326e-05,
+      "loss": 2.4028,
+      "step": 11760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.742093459654905e-05,
+      "loss": 2.4073,
+      "step": 11770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.741864616229576e-05,
+      "loss": 2.4882,
+      "step": 11780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7416357728042475e-05,
+      "loss": 2.2168,
+      "step": 11790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7414069293789196e-05,
+      "loss": 2.3842,
+      "step": 11800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.741178085953591e-05,
+      "loss": 2.4738,
+      "step": 11810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7409492425282624e-05,
+      "loss": 2.4862,
+      "step": 11820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7407203991029345e-05,
+      "loss": 2.2994,
+      "step": 11830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.740491555677606e-05,
+      "loss": 2.2938,
+      "step": 11840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.740262712252277e-05,
+      "loss": 2.6097,
+      "step": 11850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.740033868826949e-05,
+      "loss": 2.5511,
+      "step": 11860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.73980502540162e-05,
+      "loss": 2.5772,
+      "step": 11870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7395761819762916e-05,
+      "loss": 2.4539,
+      "step": 11880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.739347338550964e-05,
+      "loss": 2.3915,
+      "step": 11890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.739118495125635e-05,
+      "loss": 2.3369,
+      "step": 11900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7388896517003065e-05,
+      "loss": 2.5115,
+      "step": 11910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7386608082749786e-05,
+      "loss": 2.4355,
+      "step": 11920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.73843196484965e-05,
+      "loss": 2.2688,
+      "step": 11930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7382031214243214e-05,
+      "loss": 2.5251,
+      "step": 11940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7379742779989935e-05,
+      "loss": 2.4591,
+      "step": 11950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.737745434573665e-05,
+      "loss": 2.3582,
+      "step": 11960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.737516591148336e-05,
+      "loss": 2.4197,
+      "step": 11970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7372877477230084e-05,
+      "loss": 2.3609,
+      "step": 11980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.73705890429768e-05,
+      "loss": 2.3847,
+      "step": 11990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.736830060872351e-05,
+      "loss": 2.4748,
+      "step": 12000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.736601217447023e-05,
+      "loss": 2.6183,
+      "step": 12010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.736372374021695e-05,
+      "loss": 2.4806,
+      "step": 12020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.736143530596366e-05,
+      "loss": 2.3203,
+      "step": 12030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.735914687171038e-05,
+      "loss": 2.5022,
+      "step": 12040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.73568584374571e-05,
+      "loss": 2.4721,
+      "step": 12050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.735457000320381e-05,
+      "loss": 2.5584,
+      "step": 12060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7352281568950525e-05,
+      "loss": 2.5219,
+      "step": 12070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.734999313469724e-05,
+      "loss": 2.4887,
+      "step": 12080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.734770470044396e-05,
+      "loss": 2.4061,
+      "step": 12090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7345416266190674e-05,
+      "loss": 2.4451,
+      "step": 12100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.734312783193739e-05,
+      "loss": 2.5277,
+      "step": 12110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.734083939768411e-05,
+      "loss": 2.3891,
+      "step": 12120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.733855096343082e-05,
+      "loss": 2.3473,
+      "step": 12130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.733626252917754e-05,
+      "loss": 2.3845,
+      "step": 12140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.733397409492425e-05,
+      "loss": 2.2788,
+      "step": 12150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.733168566067097e-05,
+      "loss": 2.2998,
+      "step": 12160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.732939722641769e-05,
+      "loss": 2.4849,
+      "step": 12170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.73271087921644e-05,
+      "loss": 2.3921,
+      "step": 12180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.732482035791112e-05,
+      "loss": 2.4635,
+      "step": 12190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7322531923657836e-05,
+      "loss": 2.3999,
+      "step": 12200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.732024348940455e-05,
+      "loss": 2.3772,
+      "step": 12210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.731795505515127e-05,
+      "loss": 2.2757,
+      "step": 12220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7315666620897985e-05,
+      "loss": 2.4624,
+      "step": 12230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.73133781866447e-05,
+      "loss": 2.3496,
+      "step": 12240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.731108975239142e-05,
+      "loss": 2.5335,
+      "step": 12250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.730880131813813e-05,
+      "loss": 2.4936,
+      "step": 12260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.730651288388485e-05,
+      "loss": 2.3932,
+      "step": 12270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.730422444963156e-05,
+      "loss": 2.4017,
+      "step": 12280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7301936015378277e-05,
+      "loss": 2.4033,
+      "step": 12290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7299647581125e-05,
+      "loss": 2.6011,
+      "step": 12300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.729735914687171e-05,
+      "loss": 2.4405,
+      "step": 12310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7295070712618426e-05,
+      "loss": 2.5269,
+      "step": 12320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.729278227836515e-05,
+      "loss": 2.5057,
+      "step": 12330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.729049384411186e-05,
+      "loss": 2.5175,
+      "step": 12340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7288205409858575e-05,
+      "loss": 2.266,
+      "step": 12350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7285916975605296e-05,
+      "loss": 2.3182,
+      "step": 12360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.728362854135201e-05,
+      "loss": 2.4067,
+      "step": 12370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7281340107098724e-05,
+      "loss": 2.5653,
+      "step": 12380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7279051672845445e-05,
+      "loss": 2.5013,
+      "step": 12390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.727676323859216e-05,
+      "loss": 2.6026,
+      "step": 12400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.727447480433887e-05,
+      "loss": 2.4167,
+      "step": 12410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7272186370085594e-05,
+      "loss": 2.5231,
+      "step": 12420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.726989793583231e-05,
+      "loss": 2.3266,
+      "step": 12430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.726760950157902e-05,
+      "loss": 2.5778,
+      "step": 12440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.726532106732574e-05,
+      "loss": 2.6265,
+      "step": 12450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.726303263307245e-05,
+      "loss": 2.3835,
+      "step": 12460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7260744198819165e-05,
+      "loss": 2.4495,
+      "step": 12470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7258455764565886e-05,
+      "loss": 2.5442,
+      "step": 12480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.72561673303126e-05,
+      "loss": 2.5227,
+      "step": 12490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7253878896059314e-05,
+      "loss": 2.5473,
+      "step": 12500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7251590461806035e-05,
+      "loss": 2.4424,
+      "step": 12510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.724930202755275e-05,
+      "loss": 2.4274,
+      "step": 12520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.724701359329946e-05,
+      "loss": 2.4511,
+      "step": 12530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7244725159046184e-05,
+      "loss": 2.2158,
+      "step": 12540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.72424367247929e-05,
+      "loss": 2.5181,
+      "step": 12550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.724014829053961e-05,
+      "loss": 2.3674,
+      "step": 12560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.723785985628633e-05,
+      "loss": 2.6321,
+      "step": 12570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.723557142203305e-05,
+      "loss": 2.3881,
+      "step": 12580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.723328298777976e-05,
+      "loss": 2.5139,
+      "step": 12590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.723099455352648e-05,
+      "loss": 2.4597,
+      "step": 12600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7228706119273197e-05,
+      "loss": 2.4001,
+      "step": 12610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.722641768501991e-05,
+      "loss": 2.4194,
+      "step": 12620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.722412925076663e-05,
+      "loss": 2.3315,
+      "step": 12630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7221840816513346e-05,
+      "loss": 2.5334,
+      "step": 12640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.721955238226006e-05,
+      "loss": 2.3351,
+      "step": 12650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7217263948006774e-05,
+      "loss": 2.4023,
+      "step": 12660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.721497551375349e-05,
+      "loss": 2.4533,
+      "step": 12670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.721268707950021e-05,
+      "loss": 2.4789,
+      "step": 12680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.721039864524692e-05,
+      "loss": 2.3654,
+      "step": 12690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.720811021099364e-05,
+      "loss": 2.4534,
+      "step": 12700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.720582177674036e-05,
+      "loss": 2.4059,
+      "step": 12710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.720353334248707e-05,
+      "loss": 2.4726,
+      "step": 12720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7201244908233787e-05,
+      "loss": 2.5129,
+      "step": 12730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.719895647398051e-05,
+      "loss": 2.4124,
+      "step": 12740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.719666803972722e-05,
+      "loss": 2.6052,
+      "step": 12750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7194379605473936e-05,
+      "loss": 2.4838,
+      "step": 12760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7192091171220657e-05,
+      "loss": 2.4921,
+      "step": 12770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.718980273696737e-05,
+      "loss": 2.4695,
+      "step": 12780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7187514302714085e-05,
+      "loss": 2.3834,
+      "step": 12790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7185225868460806e-05,
+      "loss": 2.4955,
+      "step": 12800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.718293743420752e-05,
+      "loss": 2.5316,
+      "step": 12810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7180648999954234e-05,
+      "loss": 2.5042,
+      "step": 12820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7178360565700955e-05,
+      "loss": 2.4018,
+      "step": 12830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.717607213144767e-05,
+      "loss": 2.4403,
+      "step": 12840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7173783697194376e-05,
+      "loss": 2.4678,
+      "step": 12850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.71714952629411e-05,
+      "loss": 2.4937,
+      "step": 12860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.716920682868781e-05,
+      "loss": 2.4366,
+      "step": 12870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7166918394434526e-05,
+      "loss": 2.4178,
+      "step": 12880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7164629960181247e-05,
+      "loss": 2.3645,
+      "step": 12890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.716234152592796e-05,
+      "loss": 2.5206,
+      "step": 12900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7160053091674675e-05,
+      "loss": 2.3971,
+      "step": 12910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7157764657421396e-05,
+      "loss": 2.4581,
+      "step": 12920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.715547622316811e-05,
+      "loss": 2.3665,
+      "step": 12930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7153187788914824e-05,
+      "loss": 2.3638,
+      "step": 12940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7150899354661545e-05,
+      "loss": 2.4613,
+      "step": 12950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.714861092040826e-05,
+      "loss": 2.3943,
+      "step": 12960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.714632248615497e-05,
+      "loss": 2.511,
+      "step": 12970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7144034051901694e-05,
+      "loss": 2.3228,
+      "step": 12980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.714174561764841e-05,
+      "loss": 2.327,
+      "step": 12990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.713945718339512e-05,
+      "loss": 2.3646,
+      "step": 13000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.713716874914184e-05,
+      "loss": 2.3014,
+      "step": 13010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.713488031488856e-05,
+      "loss": 2.3736,
+      "step": 13020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.713259188063527e-05,
+      "loss": 2.2308,
+      "step": 13030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.713030344638199e-05,
+      "loss": 2.5022,
+      "step": 13040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.71280150121287e-05,
+      "loss": 2.3569,
+      "step": 13050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.712572657787542e-05,
+      "loss": 2.4698,
+      "step": 13060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7123438143622135e-05,
+      "loss": 2.4983,
+      "step": 13070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.712114970936885e-05,
+      "loss": 2.4369,
+      "step": 13080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.711886127511557e-05,
+      "loss": 2.4027,
+      "step": 13090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7116572840862284e-05,
+      "loss": 2.3425,
+      "step": 13100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7114284406609e-05,
+      "loss": 2.6171,
+      "step": 13110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.711199597235572e-05,
+      "loss": 2.3407,
+      "step": 13120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.710970753810243e-05,
+      "loss": 2.2714,
+      "step": 13130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.710741910384915e-05,
+      "loss": 2.4728,
+      "step": 13140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.710513066959587e-05,
+      "loss": 2.4691,
+      "step": 13150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.710284223534258e-05,
+      "loss": 2.3251,
+      "step": 13160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7100553801089296e-05,
+      "loss": 2.2302,
+      "step": 13170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.709826536683602e-05,
+      "loss": 2.4588,
+      "step": 13180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.709597693258273e-05,
+      "loss": 2.2778,
+      "step": 13190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7093688498329446e-05,
+      "loss": 2.3084,
+      "step": 13200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7091400064076167e-05,
+      "loss": 2.3952,
+      "step": 13210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.708911162982288e-05,
+      "loss": 2.48,
+      "step": 13220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7086823195569595e-05,
+      "loss": 2.3173,
+      "step": 13230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.708453476131631e-05,
+      "loss": 2.2694,
+      "step": 13240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.708224632706302e-05,
+      "loss": 2.4534,
+      "step": 13250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.707995789280974e-05,
+      "loss": 2.3544,
+      "step": 13260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.707766945855646e-05,
+      "loss": 2.5172,
+      "step": 13270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.707538102430317e-05,
+      "loss": 2.6517,
+      "step": 13280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7073092590049886e-05,
+      "loss": 2.4843,
+      "step": 13290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.707080415579661e-05,
+      "loss": 2.3516,
+      "step": 13300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.706851572154332e-05,
+      "loss": 2.5803,
+      "step": 13310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7066227287290036e-05,
+      "loss": 2.4354,
+      "step": 13320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7063938853036757e-05,
+      "loss": 2.4567,
+      "step": 13330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.706165041878347e-05,
+      "loss": 2.2692,
+      "step": 13340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7059361984530185e-05,
+      "loss": 2.4237,
+      "step": 13350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7057073550276906e-05,
+      "loss": 2.4367,
+      "step": 13360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.705478511602362e-05,
+      "loss": 2.3526,
+      "step": 13370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7052496681770334e-05,
+      "loss": 2.3193,
+      "step": 13380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7050208247517055e-05,
+      "loss": 2.4629,
+      "step": 13390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.704791981326377e-05,
+      "loss": 2.3296,
+      "step": 13400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.704563137901048e-05,
+      "loss": 2.4398,
+      "step": 13410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7043342944757204e-05,
+      "loss": 2.3305,
+      "step": 13420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.704105451050392e-05,
+      "loss": 2.2806,
+      "step": 13430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7038766076250626e-05,
+      "loss": 2.3806,
+      "step": 13440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7036477641997346e-05,
+      "loss": 2.4166,
+      "step": 13450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.703418920774406e-05,
+      "loss": 2.1978,
+      "step": 13460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7031900773490775e-05,
+      "loss": 2.2902,
+      "step": 13470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7029612339237496e-05,
+      "loss": 2.3657,
+      "step": 13480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.702732390498421e-05,
+      "loss": 2.3334,
+      "step": 13490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7025035470730924e-05,
+      "loss": 2.4528,
+      "step": 13500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7022747036477645e-05,
+      "loss": 2.3754,
+      "step": 13510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.702045860222436e-05,
+      "loss": 2.3537,
+      "step": 13520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.701817016797107e-05,
+      "loss": 2.4485,
+      "step": 13530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7015881733717794e-05,
+      "loss": 2.5266,
+      "step": 13540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.701359329946451e-05,
+      "loss": 2.5417,
+      "step": 13550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.701130486521122e-05,
+      "loss": 2.4745,
+      "step": 13560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.700901643095794e-05,
+      "loss": 2.3031,
+      "step": 13570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.700672799670466e-05,
+      "loss": 2.5789,
+      "step": 13580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.700443956245137e-05,
+      "loss": 2.5256,
+      "step": 13590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.700215112819809e-05,
+      "loss": 2.3836,
+      "step": 13600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6999862693944806e-05,
+      "loss": 2.4568,
+      "step": 13610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.699757425969152e-05,
+      "loss": 2.4344,
+      "step": 13620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.699528582543824e-05,
+      "loss": 2.464,
+      "step": 13630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.699299739118495e-05,
+      "loss": 2.3177,
+      "step": 13640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.699070895693167e-05,
+      "loss": 2.391,
+      "step": 13650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6988420522678384e-05,
+      "loss": 2.3471,
+      "step": 13660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.69861320884251e-05,
+      "loss": 2.4073,
+      "step": 13670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.698384365417182e-05,
+      "loss": 2.4141,
+      "step": 13680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.698155521991853e-05,
+      "loss": 2.2538,
+      "step": 13690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.697926678566525e-05,
+      "loss": 2.2921,
+      "step": 13700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.697697835141197e-05,
+      "loss": 2.4317,
+      "step": 13710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.697468991715868e-05,
+      "loss": 2.3687,
+      "step": 13720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6972401482905396e-05,
+      "loss": 2.5864,
+      "step": 13730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.697011304865212e-05,
+      "loss": 2.3882,
+      "step": 13740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.696782461439883e-05,
+      "loss": 2.3981,
+      "step": 13750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6965536180145546e-05,
+      "loss": 2.3473,
+      "step": 13760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6963247745892266e-05,
+      "loss": 2.4379,
+      "step": 13770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.696095931163898e-05,
+      "loss": 2.2774,
+      "step": 13780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6958670877385695e-05,
+      "loss": 2.2976,
+      "step": 13790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6956382443132416e-05,
+      "loss": 2.2165,
+      "step": 13800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.695409400887913e-05,
+      "loss": 2.406,
+      "step": 13810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6951805574625844e-05,
+      "loss": 2.393,
+      "step": 13820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6949517140372565e-05,
+      "loss": 2.3793,
+      "step": 13830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.694722870611927e-05,
+      "loss": 2.3294,
+      "step": 13840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6944940271865986e-05,
+      "loss": 2.3039,
+      "step": 13850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.694265183761271e-05,
+      "loss": 2.4783,
+      "step": 13860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.694036340335942e-05,
+      "loss": 2.2788,
+      "step": 13870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6938074969106135e-05,
+      "loss": 2.3901,
+      "step": 13880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6935786534852856e-05,
+      "loss": 2.3516,
+      "step": 13890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.693349810059957e-05,
+      "loss": 2.364,
+      "step": 13900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6931209666346285e-05,
+      "loss": 2.4494,
+      "step": 13910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6928921232093006e-05,
+      "loss": 2.2642,
+      "step": 13920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.692663279783972e-05,
+      "loss": 2.3744,
+      "step": 13930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6924344363586434e-05,
+      "loss": 2.3571,
+      "step": 13940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6922055929333155e-05,
+      "loss": 2.2558,
+      "step": 13950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.691976749507987e-05,
+      "loss": 2.3763,
+      "step": 13960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.691747906082658e-05,
+      "loss": 2.3461,
+      "step": 13970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6915190626573304e-05,
+      "loss": 2.4663,
+      "step": 13980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.691290219232002e-05,
+      "loss": 2.3942,
+      "step": 13990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.691061375806673e-05,
+      "loss": 2.3715,
+      "step": 14000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.690832532381345e-05,
+      "loss": 2.4367,
+      "step": 14010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.690603688956017e-05,
+      "loss": 2.523,
+      "step": 14020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.690374845530688e-05,
+      "loss": 2.3815,
+      "step": 14030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6901460021053596e-05,
+      "loss": 2.4112,
+      "step": 14040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.689917158680031e-05,
+      "loss": 2.2423,
+      "step": 14050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.689688315254703e-05,
+      "loss": 2.4133,
+      "step": 14060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6894594718293745e-05,
+      "loss": 2.3051,
+      "step": 14070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.689230628404046e-05,
+      "loss": 2.5434,
+      "step": 14080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.689001784978718e-05,
+      "loss": 2.3681,
+      "step": 14090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6887729415533894e-05,
+      "loss": 2.3268,
+      "step": 14100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.688544098128061e-05,
+      "loss": 2.3015,
+      "step": 14110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.688315254702733e-05,
+      "loss": 2.3093,
+      "step": 14120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.688086411277404e-05,
+      "loss": 2.3067,
+      "step": 14130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.687857567852076e-05,
+      "loss": 2.3219,
+      "step": 14140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.687628724426748e-05,
+      "loss": 2.4299,
+      "step": 14150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.687399881001419e-05,
+      "loss": 2.5319,
+      "step": 14160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6871710375760906e-05,
+      "loss": 2.2746,
+      "step": 14170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.686942194150763e-05,
+      "loss": 2.3274,
+      "step": 14180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.686713350725434e-05,
+      "loss": 2.4078,
+      "step": 14190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6864845073001056e-05,
+      "loss": 2.3112,
+      "step": 14200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6862556638747776e-05,
+      "loss": 2.3443,
+      "step": 14210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.686026820449449e-05,
+      "loss": 2.4104,
+      "step": 14220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.68579797702412e-05,
+      "loss": 2.3984,
+      "step": 14230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.685569133598792e-05,
+      "loss": 2.3395,
+      "step": 14240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.685340290173463e-05,
+      "loss": 2.4568,
+      "step": 14250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.685111446748135e-05,
+      "loss": 2.5401,
+      "step": 14260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.684882603322807e-05,
+      "loss": 2.387,
+      "step": 14270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.684653759897478e-05,
+      "loss": 2.3915,
+      "step": 14280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6844249164721496e-05,
+      "loss": 2.462,
+      "step": 14290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.684196073046822e-05,
+      "loss": 2.4498,
+      "step": 14300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.683967229621493e-05,
+      "loss": 2.3025,
+      "step": 14310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6837383861961645e-05,
+      "loss": 2.3295,
+      "step": 14320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6835095427708366e-05,
+      "loss": 2.463,
+      "step": 14330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.683280699345508e-05,
+      "loss": 2.3643,
+      "step": 14340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6830518559201795e-05,
+      "loss": 2.5365,
+      "step": 14350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6828230124948516e-05,
+      "loss": 2.3952,
+      "step": 14360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.682594169069523e-05,
+      "loss": 2.4698,
+      "step": 14370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6823653256441944e-05,
+      "loss": 2.3816,
+      "step": 14380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6821364822188665e-05,
+      "loss": 2.3845,
+      "step": 14390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.681907638793538e-05,
+      "loss": 2.3689,
+      "step": 14400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.681678795368209e-05,
+      "loss": 2.3599,
+      "step": 14410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6814499519428814e-05,
+      "loss": 2.3926,
+      "step": 14420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.681221108517552e-05,
+      "loss": 2.4718,
+      "step": 14430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.680992265092224e-05,
+      "loss": 2.4573,
+      "step": 14440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6807634216668956e-05,
+      "loss": 2.5606,
+      "step": 14450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.680534578241567e-05,
+      "loss": 2.4595,
+      "step": 14460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.680305734816239e-05,
+      "loss": 2.2884,
+      "step": 14470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6800768913909105e-05,
+      "loss": 2.3518,
+      "step": 14480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.679848047965582e-05,
+      "loss": 2.3527,
+      "step": 14490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6796192045402534e-05,
+      "loss": 2.5002,
+      "step": 14500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6793903611149255e-05,
+      "loss": 2.2668,
+      "step": 14510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.679161517689597e-05,
+      "loss": 2.2006,
+      "step": 14520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.678932674264268e-05,
+      "loss": 2.3714,
+      "step": 14530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6787038308389404e-05,
+      "loss": 2.539,
+      "step": 14540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.678474987413612e-05,
+      "loss": 2.3998,
+      "step": 14550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.678246143988283e-05,
+      "loss": 2.3236,
+      "step": 14560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.678017300562955e-05,
+      "loss": 2.2425,
+      "step": 14570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.677788457137627e-05,
+      "loss": 2.5189,
+      "step": 14580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.677559613712298e-05,
+      "loss": 2.3014,
+      "step": 14590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.67733077028697e-05,
+      "loss": 2.4558,
+      "step": 14600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6771019268616416e-05,
+      "loss": 2.3848,
+      "step": 14610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.676873083436313e-05,
+      "loss": 2.421,
+      "step": 14620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6766442400109845e-05,
+      "loss": 2.411,
+      "step": 14630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.676415396585656e-05,
+      "loss": 2.3328,
+      "step": 14640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.676186553160328e-05,
+      "loss": 2.4546,
+      "step": 14650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6759577097349994e-05,
+      "loss": 2.3235,
+      "step": 14660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.675728866309671e-05,
+      "loss": 2.3447,
+      "step": 14670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.675500022884343e-05,
+      "loss": 2.2483,
+      "step": 14680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.675271179459014e-05,
+      "loss": 2.3659,
+      "step": 14690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.675042336033686e-05,
+      "loss": 2.2684,
+      "step": 14700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.674813492608358e-05,
+      "loss": 2.4088,
+      "step": 14710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.674584649183029e-05,
+      "loss": 2.2825,
+      "step": 14720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6743558057577006e-05,
+      "loss": 2.4442,
+      "step": 14730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.674126962332373e-05,
+      "loss": 2.6232,
+      "step": 14740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.673898118907044e-05,
+      "loss": 2.4191,
+      "step": 14750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6736692754817155e-05,
+      "loss": 2.4357,
+      "step": 14760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6734404320563876e-05,
+      "loss": 2.2647,
+      "step": 14770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.673211588631059e-05,
+      "loss": 2.4307,
+      "step": 14780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6729827452057305e-05,
+      "loss": 2.4112,
+      "step": 14790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6727539017804026e-05,
+      "loss": 2.4224,
+      "step": 14800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.672525058355074e-05,
+      "loss": 2.3461,
+      "step": 14810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6722962149297454e-05,
+      "loss": 2.2378,
+      "step": 14820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.672067371504417e-05,
+      "loss": 2.3231,
+      "step": 14830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.671838528079088e-05,
+      "loss": 2.3036,
+      "step": 14840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6716096846537596e-05,
+      "loss": 2.2811,
+      "step": 14850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.671380841228432e-05,
+      "loss": 2.425,
+      "step": 14860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.671151997803103e-05,
+      "loss": 2.3556,
+      "step": 14870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6709231543777745e-05,
+      "loss": 2.3853,
+      "step": 14880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6706943109524466e-05,
+      "loss": 2.4541,
+      "step": 14890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.670465467527118e-05,
+      "loss": 2.4022,
+      "step": 14900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6702366241017895e-05,
+      "loss": 2.4588,
+      "step": 14910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6700077806764615e-05,
+      "loss": 2.4838,
+      "step": 14920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.669778937251133e-05,
+      "loss": 2.3258,
+      "step": 14930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6695500938258044e-05,
+      "loss": 2.3718,
+      "step": 14940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6693212504004765e-05,
+      "loss": 2.323,
+      "step": 14950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.669092406975148e-05,
+      "loss": 2.4033,
+      "step": 14960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.668863563549819e-05,
+      "loss": 2.5025,
+      "step": 14970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6686347201244914e-05,
+      "loss": 2.3335,
+      "step": 14980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.668405876699163e-05,
+      "loss": 2.425,
+      "step": 14990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.668177033273834e-05,
+      "loss": 2.4599,
+      "step": 15000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667948189848506e-05,
+      "loss": 2.3241,
+      "step": 15010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667719346423177e-05,
+      "loss": 2.4939,
+      "step": 15020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667490502997849e-05,
+      "loss": 2.3753,
+      "step": 15030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6672616595725205e-05,
+      "loss": 2.3891,
+      "step": 15040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667032816147192e-05,
+      "loss": 2.2793,
+      "step": 15050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.666803972721864e-05,
+      "loss": 2.2299,
+      "step": 15060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6665751292965355e-05,
+      "loss": 2.3184,
+      "step": 15070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.666346285871207e-05,
+      "loss": 2.1556,
+      "step": 15080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.666117442445879e-05,
+      "loss": 2.5305,
+      "step": 15090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6658885990205504e-05,
+      "loss": 2.3223,
+      "step": 15100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.665659755595222e-05,
+      "loss": 2.3775,
+      "step": 15110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.665430912169894e-05,
+      "loss": 2.4301,
+      "step": 15120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.665202068744565e-05,
+      "loss": 2.3689,
+      "step": 15130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.664973225319237e-05,
+      "loss": 2.2417,
+      "step": 15140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.664744381893909e-05,
+      "loss": 2.3305,
+      "step": 15150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.66451553846858e-05,
+      "loss": 2.3792,
+      "step": 15160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6642866950432516e-05,
+      "loss": 2.4314,
+      "step": 15170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.664057851617924e-05,
+      "loss": 2.328,
+      "step": 15180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663829008192595e-05,
+      "loss": 2.4546,
+      "step": 15190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6636001647672665e-05,
+      "loss": 2.1596,
+      "step": 15200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6633713213419386e-05,
+      "loss": 2.4049,
+      "step": 15210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6631424779166094e-05,
+      "loss": 2.309,
+      "step": 15220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.662913634491281e-05,
+      "loss": 2.3605,
+      "step": 15230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.662684791065953e-05,
+      "loss": 2.3429,
+      "step": 15240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.662455947640624e-05,
+      "loss": 2.3773,
+      "step": 15250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.662227104215296e-05,
+      "loss": 2.3893,
+      "step": 15260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.661998260789968e-05,
+      "loss": 2.4296,
+      "step": 15270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.661769417364639e-05,
+      "loss": 2.4665,
+      "step": 15280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6615405739393106e-05,
+      "loss": 2.3994,
+      "step": 15290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.661311730513983e-05,
+      "loss": 2.3696,
+      "step": 15300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.661082887088654e-05,
+      "loss": 2.4547,
+      "step": 15310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6608540436633255e-05,
+      "loss": 2.3975,
+      "step": 15320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6606252002379976e-05,
+      "loss": 2.3168,
+      "step": 15330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.660396356812669e-05,
+      "loss": 2.3848,
+      "step": 15340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6601675133873405e-05,
+      "loss": 2.4167,
+      "step": 15350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6599386699620125e-05,
+      "loss": 2.3531,
+      "step": 15360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.659709826536684e-05,
+      "loss": 2.3287,
+      "step": 15370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6594809831113554e-05,
+      "loss": 2.561,
+      "step": 15380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6592521396860275e-05,
+      "loss": 2.3686,
+      "step": 15390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.659023296260699e-05,
+      "loss": 2.5802,
+      "step": 15400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.65879445283537e-05,
+      "loss": 2.2783,
+      "step": 15410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.658565609410042e-05,
+      "loss": 2.3303,
+      "step": 15420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.658336765984713e-05,
+      "loss": 2.319,
+      "step": 15430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.658107922559385e-05,
+      "loss": 2.3461,
+      "step": 15440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6578790791340566e-05,
+      "loss": 2.2342,
+      "step": 15450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.657650235708728e-05,
+      "loss": 2.3677,
+      "step": 15460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6574213922834e-05,
+      "loss": 2.2846,
+      "step": 15470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6571925488580715e-05,
+      "loss": 2.4242,
+      "step": 15480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.656963705432743e-05,
+      "loss": 2.432,
+      "step": 15490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.656734862007415e-05,
+      "loss": 2.3391,
+      "step": 15500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6565060185820865e-05,
+      "loss": 2.5358,
+      "step": 15510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.656277175156758e-05,
+      "loss": 2.44,
+      "step": 15520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.65604833173143e-05,
+      "loss": 2.4573,
+      "step": 15530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6558194883061014e-05,
+      "loss": 2.3064,
+      "step": 15540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.655590644880773e-05,
+      "loss": 2.3635,
+      "step": 15550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.655361801455445e-05,
+      "loss": 2.481,
+      "step": 15560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.655132958030116e-05,
+      "loss": 2.3738,
+      "step": 15570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.654904114604788e-05,
+      "loss": 2.3046,
+      "step": 15580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.654675271179459e-05,
+      "loss": 2.4667,
+      "step": 15590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.654446427754131e-05,
+      "loss": 2.5539,
+      "step": 15600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6542175843288026e-05,
+      "loss": 2.3635,
+      "step": 15610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.653988740903474e-05,
+      "loss": 2.3257,
+      "step": 15620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6537598974781454e-05,
+      "loss": 2.4459,
+      "step": 15630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.653531054052817e-05,
+      "loss": 2.3885,
+      "step": 15640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.653302210627489e-05,
+      "loss": 2.3926,
+      "step": 15650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6530733672021604e-05,
+      "loss": 2.2393,
+      "step": 15660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.652844523776832e-05,
+      "loss": 2.4107,
+      "step": 15670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.652615680351504e-05,
+      "loss": 2.4162,
+      "step": 15680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.652386836926175e-05,
+      "loss": 2.2627,
+      "step": 15690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.652157993500847e-05,
+      "loss": 2.308,
+      "step": 15700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.651929150075519e-05,
+      "loss": 2.349,
+      "step": 15710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.65170030665019e-05,
+      "loss": 2.4102,
+      "step": 15720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6514714632248616e-05,
+      "loss": 2.4103,
+      "step": 15730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.651242619799534e-05,
+      "loss": 2.3012,
+      "step": 15740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.651013776374205e-05,
+      "loss": 2.3481,
+      "step": 15750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6507849329488765e-05,
+      "loss": 2.3335,
+      "step": 15760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6505560895235486e-05,
+      "loss": 2.4007,
+      "step": 15770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.65032724609822e-05,
+      "loss": 2.3978,
+      "step": 15780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6500984026728914e-05,
+      "loss": 2.2898,
+      "step": 15790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6498695592475635e-05,
+      "loss": 2.5494,
+      "step": 15800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.649640715822235e-05,
+      "loss": 2.3529,
+      "step": 15810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.649411872396906e-05,
+      "loss": 2.5162,
+      "step": 15820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.649183028971578e-05,
+      "loss": 2.3746,
+      "step": 15830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.648954185546249e-05,
+      "loss": 2.3764,
+      "step": 15840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6487253421209206e-05,
+      "loss": 2.4426,
+      "step": 15850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.648496498695593e-05,
+      "loss": 2.4078,
+      "step": 15860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.648267655270264e-05,
+      "loss": 2.4534,
+      "step": 15870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6480388118449355e-05,
+      "loss": 2.2913,
+      "step": 15880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6478099684196076e-05,
+      "loss": 2.4197,
+      "step": 15890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.647581124994279e-05,
+      "loss": 2.486,
+      "step": 15900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6473522815689504e-05,
+      "loss": 2.3556,
+      "step": 15910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6471234381436225e-05,
+      "loss": 2.41,
+      "step": 15920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.646894594718294e-05,
+      "loss": 2.4941,
+      "step": 15930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6466657512929654e-05,
+      "loss": 2.384,
+      "step": 15940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6464369078676374e-05,
+      "loss": 2.3485,
+      "step": 15950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.646208064442309e-05,
+      "loss": 2.3617,
+      "step": 15960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.64597922101698e-05,
+      "loss": 2.1509,
+      "step": 15970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6457503775916524e-05,
+      "loss": 2.3831,
+      "step": 15980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.645521534166324e-05,
+      "loss": 2.303,
+      "step": 15990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.645292690740995e-05,
+      "loss": 2.2721,
+      "step": 16000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6450638473156666e-05,
+      "loss": 2.3479,
+      "step": 16010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.644835003890338e-05,
+      "loss": 2.3042,
+      "step": 16020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.64460616046501e-05,
+      "loss": 2.5601,
+      "step": 16030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6443773170396815e-05,
+      "loss": 2.2304,
+      "step": 16040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.644148473614353e-05,
+      "loss": 2.2836,
+      "step": 16050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.643919630189025e-05,
+      "loss": 2.3329,
+      "step": 16060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6436907867636964e-05,
+      "loss": 2.3403,
+      "step": 16070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.643461943338368e-05,
+      "loss": 2.4404,
+      "step": 16080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.64323309991304e-05,
+      "loss": 2.3952,
+      "step": 16090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6430042564877114e-05,
+      "loss": 2.1968,
+      "step": 16100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.642775413062383e-05,
+      "loss": 2.4064,
+      "step": 16110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.642546569637055e-05,
+      "loss": 2.4074,
+      "step": 16120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.642317726211726e-05,
+      "loss": 2.3967,
+      "step": 16130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.642088882786398e-05,
+      "loss": 2.594,
+      "step": 16140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.64186003936107e-05,
+      "loss": 2.4373,
+      "step": 16150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.641631195935741e-05,
+      "loss": 2.5472,
+      "step": 16160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6414023525104126e-05,
+      "loss": 2.5062,
+      "step": 16170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.641173509085085e-05,
+      "loss": 2.4266,
+      "step": 16180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.640944665659756e-05,
+      "loss": 2.4457,
+      "step": 16190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6407158222344275e-05,
+      "loss": 2.372,
+      "step": 16200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.640486978809099e-05,
+      "loss": 2.2929,
+      "step": 16210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6402581353837704e-05,
+      "loss": 2.3839,
+      "step": 16220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.640029291958442e-05,
+      "loss": 2.465,
+      "step": 16230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.639800448533114e-05,
+      "loss": 2.4987,
+      "step": 16240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.639571605107785e-05,
+      "loss": 2.3799,
+      "step": 16250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.639342761682457e-05,
+      "loss": 2.4681,
+      "step": 16260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.639113918257129e-05,
+      "loss": 2.3439,
+      "step": 16270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6388850748318e-05,
+      "loss": 2.4584,
+      "step": 16280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6386562314064716e-05,
+      "loss": 2.52,
+      "step": 16290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.638427387981144e-05,
+      "loss": 2.2984,
+      "step": 16300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.638198544555815e-05,
+      "loss": 2.3908,
+      "step": 16310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6379697011304865e-05,
+      "loss": 2.3904,
+      "step": 16320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6377408577051586e-05,
+      "loss": 2.1904,
+      "step": 16330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.63751201427983e-05,
+      "loss": 2.481,
+      "step": 16340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6372831708545014e-05,
+      "loss": 2.3175,
+      "step": 16350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6370543274291735e-05,
+      "loss": 2.2616,
+      "step": 16360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.636825484003845e-05,
+      "loss": 2.3663,
+      "step": 16370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6365966405785164e-05,
+      "loss": 2.3417,
+      "step": 16380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6363677971531884e-05,
+      "loss": 2.387,
+      "step": 16390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.63613895372786e-05,
+      "loss": 2.2699,
+      "step": 16400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.635910110302531e-05,
+      "loss": 2.231,
+      "step": 16410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.635681266877203e-05,
+      "loss": 2.2993,
+      "step": 16420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.635452423451874e-05,
+      "loss": 2.408,
+      "step": 16430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.635223580026546e-05,
+      "loss": 2.4687,
+      "step": 16440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6349947366012176e-05,
+      "loss": 2.3699,
+      "step": 16450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.634765893175889e-05,
+      "loss": 2.3452,
+      "step": 16460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.634537049750561e-05,
+      "loss": 2.5306,
+      "step": 16470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6343082063252325e-05,
+      "loss": 2.4026,
+      "step": 16480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.634079362899904e-05,
+      "loss": 2.3577,
+      "step": 16490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.633850519474576e-05,
+      "loss": 2.4862,
+      "step": 16500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6336216760492474e-05,
+      "loss": 2.4687,
+      "step": 16510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.633392832623919e-05,
+      "loss": 2.3897,
+      "step": 16520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.633163989198591e-05,
+      "loss": 2.4917,
+      "step": 16530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6329351457732624e-05,
+      "loss": 2.4958,
+      "step": 16540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.632706302347934e-05,
+      "loss": 2.3769,
+      "step": 16550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.632477458922606e-05,
+      "loss": 2.2684,
+      "step": 16560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.632248615497277e-05,
+      "loss": 2.3025,
+      "step": 16570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.632019772071949e-05,
+      "loss": 2.3771,
+      "step": 16580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.631790928646621e-05,
+      "loss": 2.4185,
+      "step": 16590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.631562085221292e-05,
+      "loss": 2.5521,
+      "step": 16600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.631333241795963e-05,
+      "loss": 2.4028,
+      "step": 16610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.631104398370635e-05,
+      "loss": 2.2996,
+      "step": 16620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6308755549453064e-05,
+      "loss": 2.3649,
+      "step": 16630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.630646711519978e-05,
+      "loss": 2.2891,
+      "step": 16640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.63041786809465e-05,
+      "loss": 2.3216,
+      "step": 16650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6301890246693213e-05,
+      "loss": 2.1903,
+      "step": 16660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.629960181243993e-05,
+      "loss": 2.4027,
+      "step": 16670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.629731337818665e-05,
+      "loss": 2.4524,
+      "step": 16680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.629502494393336e-05,
+      "loss": 2.3274,
+      "step": 16690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.629273650968008e-05,
+      "loss": 2.3721,
+      "step": 16700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.62904480754268e-05,
+      "loss": 2.2037,
+      "step": 16710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.628815964117351e-05,
+      "loss": 2.3744,
+      "step": 16720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6285871206920226e-05,
+      "loss": 2.2048,
+      "step": 16730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.628358277266695e-05,
+      "loss": 2.1571,
+      "step": 16740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.628129433841366e-05,
+      "loss": 2.3747,
+      "step": 16750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6279005904160375e-05,
+      "loss": 2.263,
+      "step": 16760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6276717469907096e-05,
+      "loss": 2.5846,
+      "step": 16770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.627442903565381e-05,
+      "loss": 2.272,
+      "step": 16780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6272140601400524e-05,
+      "loss": 2.2812,
+      "step": 16790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.626985216714724e-05,
+      "loss": 2.2774,
+      "step": 16800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.626756373289395e-05,
+      "loss": 2.5418,
+      "step": 16810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.626527529864067e-05,
+      "loss": 2.206,
+      "step": 16820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.626298686438739e-05,
+      "loss": 2.3299,
+      "step": 16830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.62606984301341e-05,
+      "loss": 2.4003,
+      "step": 16840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6258409995880816e-05,
+      "loss": 2.303,
+      "step": 16850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.625612156162754e-05,
+      "loss": 2.1617,
+      "step": 16860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.625383312737425e-05,
+      "loss": 2.2616,
+      "step": 16870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6251544693120965e-05,
+      "loss": 2.4261,
+      "step": 16880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6249256258867686e-05,
+      "loss": 2.4306,
+      "step": 16890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.62469678246144e-05,
+      "loss": 2.3649,
+      "step": 16900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6244679390361114e-05,
+      "loss": 2.3116,
+      "step": 16910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6242390956107835e-05,
+      "loss": 2.2467,
+      "step": 16920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.624010252185455e-05,
+      "loss": 2.4557,
+      "step": 16930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6237814087601263e-05,
+      "loss": 2.328,
+      "step": 16940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6235525653347984e-05,
+      "loss": 2.3345,
+      "step": 16950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.62332372190947e-05,
+      "loss": 2.2343,
+      "step": 16960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.623094878484141e-05,
+      "loss": 2.3054,
+      "step": 16970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6228660350588134e-05,
+      "loss": 2.4167,
+      "step": 16980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.622637191633485e-05,
+      "loss": 2.4329,
+      "step": 16990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.622408348208156e-05,
+      "loss": 2.2721,
+      "step": 17000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6221795047828276e-05,
+      "loss": 2.3178,
+      "step": 17010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.621950661357499e-05,
+      "loss": 2.3451,
+      "step": 17020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.621721817932171e-05,
+      "loss": 2.2038,
+      "step": 17030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6214929745068425e-05,
+      "loss": 2.3536,
+      "step": 17040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.621264131081514e-05,
+      "loss": 2.259,
+      "step": 17050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.621035287656186e-05,
+      "loss": 2.4451,
+      "step": 17060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6208064442308574e-05,
+      "loss": 2.4755,
+      "step": 17070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.620577600805529e-05,
+      "loss": 2.2509,
+      "step": 17080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.620348757380201e-05,
+      "loss": 2.4435,
+      "step": 17090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6201199139548723e-05,
+      "loss": 2.3517,
+      "step": 17100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.619891070529544e-05,
+      "loss": 2.3907,
+      "step": 17110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.619662227104216e-05,
+      "loss": 2.4024,
+      "step": 17120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.619433383678887e-05,
+      "loss": 2.5077,
+      "step": 17130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.619204540253559e-05,
+      "loss": 2.5415,
+      "step": 17140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.618975696828231e-05,
+      "loss": 2.1163,
+      "step": 17150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.618746853402902e-05,
+      "loss": 2.485,
+      "step": 17160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6185180099775736e-05,
+      "loss": 2.2067,
+      "step": 17170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.618289166552246e-05,
+      "loss": 2.3718,
+      "step": 17180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.618060323126917e-05,
+      "loss": 2.3559,
+      "step": 17190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.617831479701588e-05,
+      "loss": 2.2452,
+      "step": 17200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.61760263627626e-05,
+      "loss": 2.2834,
+      "step": 17210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6173737928509313e-05,
+      "loss": 2.3021,
+      "step": 17220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.617144949425603e-05,
+      "loss": 2.4749,
+      "step": 17230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.616916106000275e-05,
+      "loss": 2.2662,
+      "step": 17240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.616687262574946e-05,
+      "loss": 2.4053,
+      "step": 17250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.616458419149618e-05,
+      "loss": 2.4304,
+      "step": 17260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.61622957572429e-05,
+      "loss": 2.3871,
+      "step": 17270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.616000732298961e-05,
+      "loss": 2.2039,
+      "step": 17280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6157718888736326e-05,
+      "loss": 2.2611,
+      "step": 17290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.615543045448305e-05,
+      "loss": 2.3669,
+      "step": 17300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.615314202022976e-05,
+      "loss": 2.336,
+      "step": 17310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6150853585976475e-05,
+      "loss": 2.3281,
+      "step": 17320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6148565151723196e-05,
+      "loss": 2.2588,
+      "step": 17330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.614627671746991e-05,
+      "loss": 2.2774,
+      "step": 17340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6143988283216624e-05,
+      "loss": 2.3469,
+      "step": 17350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6141699848963345e-05,
+      "loss": 2.3594,
+      "step": 17360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.613941141471006e-05,
+      "loss": 2.3696,
+      "step": 17370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6137122980456773e-05,
+      "loss": 2.2175,
+      "step": 17380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6134834546203494e-05,
+      "loss": 2.3206,
+      "step": 17390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.61325461119502e-05,
+      "loss": 2.3904,
+      "step": 17400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.613025767769692e-05,
+      "loss": 2.4154,
+      "step": 17410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.612796924344364e-05,
+      "loss": 2.421,
+      "step": 17420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.612568080919035e-05,
+      "loss": 2.3102,
+      "step": 17430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.612339237493707e-05,
+      "loss": 2.4184,
+      "step": 17440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6121103940683786e-05,
+      "loss": 2.3762,
+      "step": 17450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.61188155064305e-05,
+      "loss": 2.4048,
+      "step": 17460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.611652707217722e-05,
+      "loss": 2.3142,
+      "step": 17470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6114238637923935e-05,
+      "loss": 2.5533,
+      "step": 17480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.611195020367065e-05,
+      "loss": 2.2675,
+      "step": 17490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.610966176941737e-05,
+      "loss": 2.4015,
+      "step": 17500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6107373335164084e-05,
+      "loss": 2.4635,
+      "step": 17510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.61050849009108e-05,
+      "loss": 2.1911,
+      "step": 17520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.610279646665752e-05,
+      "loss": 2.3663,
+      "step": 17530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6100508032404233e-05,
+      "loss": 2.4052,
+      "step": 17540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.609821959815095e-05,
+      "loss": 2.425,
+      "step": 17550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.609593116389767e-05,
+      "loss": 2.3711,
+      "step": 17560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.609364272964438e-05,
+      "loss": 2.3581,
+      "step": 17570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.60913542953911e-05,
+      "loss": 2.2151,
+      "step": 17580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.608906586113781e-05,
+      "loss": 2.2845,
+      "step": 17590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6086777426884525e-05,
+      "loss": 2.3327,
+      "step": 17600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.608448899263124e-05,
+      "loss": 2.3474,
+      "step": 17610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.608220055837796e-05,
+      "loss": 2.3489,
+      "step": 17620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6079912124124674e-05,
+      "loss": 2.2983,
+      "step": 17630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.607762368987139e-05,
+      "loss": 2.3996,
+      "step": 17640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.607533525561811e-05,
+      "loss": 2.3051,
+      "step": 17650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.607304682136482e-05,
+      "loss": 2.1683,
+      "step": 17660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.607075838711154e-05,
+      "loss": 2.3945,
+      "step": 17670
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.606846995285826e-05,
+      "loss": 2.3088,
+      "step": 17680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.606618151860497e-05,
+      "loss": 2.3674,
+      "step": 17690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.606389308435169e-05,
+      "loss": 2.4392,
+      "step": 17700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.606160465009841e-05,
+      "loss": 2.3587,
+      "step": 17710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.605931621584512e-05,
+      "loss": 2.288,
+      "step": 17720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6057027781591836e-05,
+      "loss": 2.2625,
+      "step": 17730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.605473934733856e-05,
+      "loss": 2.2856,
+      "step": 17740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.605245091308527e-05,
+      "loss": 2.2983,
+      "step": 17750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6050162478831985e-05,
+      "loss": 2.3513,
+      "step": 17760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6047874044578706e-05,
+      "loss": 2.2521,
+      "step": 17770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.604558561032542e-05,
+      "loss": 2.2695,
+      "step": 17780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6043297176072134e-05,
+      "loss": 2.2536,
+      "step": 17790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.604100874181885e-05,
+      "loss": 2.2978,
+      "step": 17800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.603872030756556e-05,
+      "loss": 2.32,
+      "step": 17810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.603643187331228e-05,
+      "loss": 2.3636,
+      "step": 17820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6034143439059e-05,
+      "loss": 2.2627,
+      "step": 17830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.603185500480571e-05,
+      "loss": 2.3688,
+      "step": 17840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.602956657055243e-05,
+      "loss": 2.5841,
+      "step": 17850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.602727813629915e-05,
+      "loss": 2.2331,
+      "step": 17860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.602498970204586e-05,
+      "loss": 2.4572,
+      "step": 17870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.602270126779258e-05,
+      "loss": 2.3244,
+      "step": 17880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6020412833539296e-05,
+      "loss": 2.4307,
+      "step": 17890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.601812439928601e-05,
+      "loss": 2.4377,
+      "step": 17900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.601583596503273e-05,
+      "loss": 2.3428,
+      "step": 17910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6013547530779445e-05,
+      "loss": 2.2729,
+      "step": 17920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.601125909652616e-05,
+      "loss": 2.1953,
+      "step": 17930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.600897066227287e-05,
+      "loss": 2.1959,
+      "step": 17940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6006682228019594e-05,
+      "loss": 2.2828,
+      "step": 17950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.600439379376631e-05,
+      "loss": 2.3928,
+      "step": 17960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.600210535951302e-05,
+      "loss": 2.3492,
+      "step": 17970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5999816925259743e-05,
+      "loss": 2.1146,
+      "step": 17980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.599752849100645e-05,
+      "loss": 2.3273,
+      "step": 17990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.599524005675317e-05,
+      "loss": 2.1981,
+      "step": 18000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5992951622499886e-05,
+      "loss": 2.4604,
+      "step": 18010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.59906631882466e-05,
+      "loss": 2.338,
+      "step": 18020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.598837475399332e-05,
+      "loss": 2.1164,
+      "step": 18030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5986086319740035e-05,
+      "loss": 2.3731,
+      "step": 18040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.598379788548675e-05,
+      "loss": 2.2833,
+      "step": 18050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.598150945123347e-05,
+      "loss": 2.389,
+      "step": 18060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5979221016980184e-05,
+      "loss": 2.3295,
+      "step": 18070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.59769325827269e-05,
+      "loss": 2.2687,
+      "step": 18080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.597464414847362e-05,
+      "loss": 2.3204,
+      "step": 18090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.597235571422033e-05,
+      "loss": 2.1336,
+      "step": 18100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.597006727996705e-05,
+      "loss": 2.2926,
+      "step": 18110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.596777884571377e-05,
+      "loss": 2.3188,
+      "step": 18120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.596549041146048e-05,
+      "loss": 2.3169,
+      "step": 18130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.59632019772072e-05,
+      "loss": 2.3555,
+      "step": 18140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.596091354295392e-05,
+      "loss": 2.4652,
+      "step": 18150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.595862510870063e-05,
+      "loss": 2.3797,
+      "step": 18160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5956336674447346e-05,
+      "loss": 2.4016,
+      "step": 18170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.595404824019407e-05,
+      "loss": 2.3414,
+      "step": 18180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5951759805940774e-05,
+      "loss": 2.1794,
+      "step": 18190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.594947137168749e-05,
+      "loss": 2.5393,
+      "step": 18200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.594718293743421e-05,
+      "loss": 2.4237,
+      "step": 18210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.594489450318092e-05,
+      "loss": 2.5043,
+      "step": 18220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.594260606892764e-05,
+      "loss": 2.3517,
+      "step": 18230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.594031763467436e-05,
+      "loss": 2.3566,
+      "step": 18240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.593802920042107e-05,
+      "loss": 2.2744,
+      "step": 18250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5935740766167787e-05,
+      "loss": 2.3902,
+      "step": 18260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.593345233191451e-05,
+      "loss": 2.5577,
+      "step": 18270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.593116389766122e-05,
+      "loss": 2.48,
+      "step": 18280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5928875463407936e-05,
+      "loss": 2.3165,
+      "step": 18290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.592658702915466e-05,
+      "loss": 2.3504,
+      "step": 18300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.592429859490137e-05,
+      "loss": 2.5786,
+      "step": 18310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5922010160648085e-05,
+      "loss": 2.4614,
+      "step": 18320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5919721726394806e-05,
+      "loss": 2.3181,
+      "step": 18330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.591743329214152e-05,
+      "loss": 2.2768,
+      "step": 18340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5915144857888234e-05,
+      "loss": 2.1263,
+      "step": 18350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5912856423634955e-05,
+      "loss": 2.2496,
+      "step": 18360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.591056798938167e-05,
+      "loss": 2.3443,
+      "step": 18370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.590827955512838e-05,
+      "loss": 2.2266,
+      "step": 18380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.59059911208751e-05,
+      "loss": 2.5338,
+      "step": 18390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.590370268662181e-05,
+      "loss": 2.3269,
+      "step": 18400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.590141425236853e-05,
+      "loss": 2.4094,
+      "step": 18410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5899125818115247e-05,
+      "loss": 2.2199,
+      "step": 18420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.589683738386196e-05,
+      "loss": 2.4159,
+      "step": 18430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.589454894960868e-05,
+      "loss": 2.2806,
+      "step": 18440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5892260515355396e-05,
+      "loss": 2.3626,
+      "step": 18450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.588997208110211e-05,
+      "loss": 2.2269,
+      "step": 18460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.588768364684883e-05,
+      "loss": 2.4798,
+      "step": 18470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5885395212595545e-05,
+      "loss": 2.3738,
+      "step": 18480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.588310677834226e-05,
+      "loss": 2.4491,
+      "step": 18490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.588081834408898e-05,
+      "loss": 2.4643,
+      "step": 18500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5878529909835694e-05,
+      "loss": 2.3537,
+      "step": 18510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.587624147558241e-05,
+      "loss": 2.2546,
+      "step": 18520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.587395304132913e-05,
+      "loss": 2.3253,
+      "step": 18530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.587166460707584e-05,
+      "loss": 2.1583,
+      "step": 18540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.586937617282256e-05,
+      "loss": 2.3538,
+      "step": 18550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.586708773856928e-05,
+      "loss": 2.2073,
+      "step": 18560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.586479930431599e-05,
+      "loss": 2.3757,
+      "step": 18570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.58625108700627e-05,
+      "loss": 2.4133,
+      "step": 18580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.586022243580942e-05,
+      "loss": 2.2985,
+      "step": 18590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5857934001556135e-05,
+      "loss": 2.2218,
+      "step": 18600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.585564556730285e-05,
+      "loss": 2.332,
+      "step": 18610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.585335713304957e-05,
+      "loss": 2.2867,
+      "step": 18620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5851068698796284e-05,
+      "loss": 2.5229,
+      "step": 18630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5848780264543e-05,
+      "loss": 2.0596,
+      "step": 18640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.584649183028972e-05,
+      "loss": 2.3329,
+      "step": 18650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.584420339603643e-05,
+      "loss": 2.2265,
+      "step": 18660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.584191496178315e-05,
+      "loss": 2.2037,
+      "step": 18670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.583962652752987e-05,
+      "loss": 2.3796,
+      "step": 18680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.583733809327658e-05,
+      "loss": 2.3182,
+      "step": 18690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5835049659023297e-05,
+      "loss": 2.4208,
+      "step": 18700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.583276122477002e-05,
+      "loss": 2.4165,
+      "step": 18710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.583047279051673e-05,
+      "loss": 2.3324,
+      "step": 18720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5828184356263446e-05,
+      "loss": 2.2363,
+      "step": 18730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.582589592201017e-05,
+      "loss": 2.2866,
+      "step": 18740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.582360748775688e-05,
+      "loss": 2.37,
+      "step": 18750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5821319053503595e-05,
+      "loss": 2.2878,
+      "step": 18760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5819030619250316e-05,
+      "loss": 2.2953,
+      "step": 18770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.581674218499702e-05,
+      "loss": 2.4315,
+      "step": 18780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5814453750743744e-05,
+      "loss": 2.372,
+      "step": 18790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.581216531649046e-05,
+      "loss": 2.2587,
+      "step": 18800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.580987688223717e-05,
+      "loss": 2.2833,
+      "step": 18810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.580758844798389e-05,
+      "loss": 2.3189,
+      "step": 18820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.580530001373061e-05,
+      "loss": 2.3696,
+      "step": 18830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.580301157947732e-05,
+      "loss": 2.3561,
+      "step": 18840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.580072314522404e-05,
+      "loss": 2.129,
+      "step": 18850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5798434710970757e-05,
+      "loss": 2.2292,
+      "step": 18860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.579614627671747e-05,
+      "loss": 2.4677,
+      "step": 18870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.579385784246419e-05,
+      "loss": 2.4652,
+      "step": 18880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5791569408210906e-05,
+      "loss": 2.4207,
+      "step": 18890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.578928097395762e-05,
+      "loss": 2.2877,
+      "step": 18900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.578699253970434e-05,
+      "loss": 2.2329,
+      "step": 18910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5784704105451055e-05,
+      "loss": 2.3119,
+      "step": 18920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.578241567119777e-05,
+      "loss": 2.3258,
+      "step": 18930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.578012723694449e-05,
+      "loss": 2.3482,
+      "step": 18940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5777838802691204e-05,
+      "loss": 2.3033,
+      "step": 18950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.577555036843792e-05,
+      "loss": 2.4461,
+      "step": 18960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.577326193418464e-05,
+      "loss": 2.2534,
+      "step": 18970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5770973499931346e-05,
+      "loss": 2.4482,
+      "step": 18980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.576868506567806e-05,
+      "loss": 2.2505,
+      "step": 18990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.576639663142478e-05,
+      "loss": 2.2911,
+      "step": 19000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5764108197171496e-05,
+      "loss": 2.3835,
+      "step": 19010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.576181976291821e-05,
+      "loss": 2.3962,
+      "step": 19020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.575953132866493e-05,
+      "loss": 2.3376,
+      "step": 19030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5757242894411645e-05,
+      "loss": 2.324,
+      "step": 19040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.575495446015836e-05,
+      "loss": 2.2832,
+      "step": 19050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.575266602590508e-05,
+      "loss": 2.2894,
+      "step": 19060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5750377591651794e-05,
+      "loss": 2.302,
+      "step": 19070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.574808915739851e-05,
+      "loss": 2.2316,
+      "step": 19080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.574580072314523e-05,
+      "loss": 2.4414,
+      "step": 19090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.574351228889194e-05,
+      "loss": 2.2921,
+      "step": 19100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.574122385463866e-05,
+      "loss": 2.2658,
+      "step": 19110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.573893542038538e-05,
+      "loss": 2.2008,
+      "step": 19120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.573664698613209e-05,
+      "loss": 2.297,
+      "step": 19130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5734358551878807e-05,
+      "loss": 2.3842,
+      "step": 19140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.573207011762553e-05,
+      "loss": 2.0722,
+      "step": 19150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.572978168337224e-05,
+      "loss": 2.2495,
+      "step": 19160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.572749324911895e-05,
+      "loss": 2.3307,
+      "step": 19170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.572520481486567e-05,
+      "loss": 2.2726,
+      "step": 19180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5722916380612384e-05,
+      "loss": 2.1957,
+      "step": 19190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.57206279463591e-05,
+      "loss": 2.2915,
+      "step": 19200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.571833951210582e-05,
+      "loss": 2.4166,
+      "step": 19210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.571605107785253e-05,
+      "loss": 2.3094,
+      "step": 19220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.571376264359925e-05,
+      "loss": 2.4369,
+      "step": 19230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.571147420934597e-05,
+      "loss": 2.1235,
+      "step": 19240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.570918577509268e-05,
+      "loss": 2.4122,
+      "step": 19250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5706897340839396e-05,
+      "loss": 2.6411,
+      "step": 19260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.570460890658612e-05,
+      "loss": 2.2493,
+      "step": 19270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.570232047233283e-05,
+      "loss": 2.3754,
+      "step": 19280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5700032038079546e-05,
+      "loss": 2.3285,
+      "step": 19290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5697743603826267e-05,
+      "loss": 2.3002,
+      "step": 19300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.569545516957298e-05,
+      "loss": 2.2763,
+      "step": 19310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5693166735319695e-05,
+      "loss": 2.3641,
+      "step": 19320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5690878301066416e-05,
+      "loss": 2.404,
+      "step": 19330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.568858986681313e-05,
+      "loss": 2.2606,
+      "step": 19340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5686301432559844e-05,
+      "loss": 2.3274,
+      "step": 19350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5684012998306565e-05,
+      "loss": 2.3824,
+      "step": 19360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.568172456405327e-05,
+      "loss": 2.2078,
+      "step": 19370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.567943612979999e-05,
+      "loss": 2.2986,
+      "step": 19380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.567714769554671e-05,
+      "loss": 2.2778,
+      "step": 19390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.567485926129342e-05,
+      "loss": 2.3458,
+      "step": 19400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.567257082704014e-05,
+      "loss": 2.4335,
+      "step": 19410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5670282392786856e-05,
+      "loss": 2.4214,
+      "step": 19420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.566799395853357e-05,
+      "loss": 2.4537,
+      "step": 19430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.566570552428029e-05,
+      "loss": 2.416,
+      "step": 19440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5663417090027006e-05,
+      "loss": 2.3781,
+      "step": 19450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.566112865577372e-05,
+      "loss": 2.2515,
+      "step": 19460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.565884022152044e-05,
+      "loss": 2.1159,
+      "step": 19470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5656551787267155e-05,
+      "loss": 2.2833,
+      "step": 19480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.565426335301387e-05,
+      "loss": 2.1917,
+      "step": 19490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.565197491876059e-05,
+      "loss": 2.2998,
+      "step": 19500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5649686484507304e-05,
+      "loss": 2.3034,
+      "step": 19510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.564739805025402e-05,
+      "loss": 2.4424,
+      "step": 19520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.564510961600074e-05,
+      "loss": 2.32,
+      "step": 19530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.564282118174745e-05,
+      "loss": 2.3027,
+      "step": 19540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.564053274749417e-05,
+      "loss": 2.2994,
+      "step": 19550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.563824431324089e-05,
+      "loss": 2.402,
+      "step": 19560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5635955878987596e-05,
+      "loss": 2.4273,
+      "step": 19570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.563366744473431e-05,
+      "loss": 2.1845,
+      "step": 19580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.563137901048103e-05,
+      "loss": 2.3509,
+      "step": 19590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5629090576227745e-05,
+      "loss": 2.4227,
+      "step": 19600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.562680214197446e-05,
+      "loss": 2.2192,
+      "step": 19610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.562451370772118e-05,
+      "loss": 2.4397,
+      "step": 19620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5622225273467894e-05,
+      "loss": 2.3549,
+      "step": 19630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.561993683921461e-05,
+      "loss": 2.3323,
+      "step": 19640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.561764840496133e-05,
+      "loss": 2.3329,
+      "step": 19650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.561535997070804e-05,
+      "loss": 2.3269,
+      "step": 19660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.561307153645476e-05,
+      "loss": 2.3676,
+      "step": 19670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.561078310220148e-05,
+      "loss": 2.4446,
+      "step": 19680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.560849466794819e-05,
+      "loss": 2.3611,
+      "step": 19690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5606206233694906e-05,
+      "loss": 2.2033,
+      "step": 19700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.560391779944163e-05,
+      "loss": 2.1853,
+      "step": 19710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.560162936518834e-05,
+      "loss": 2.1611,
+      "step": 19720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5599340930935056e-05,
+      "loss": 2.4002,
+      "step": 19730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5597052496681776e-05,
+      "loss": 2.2953,
+      "step": 19740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.559476406242849e-05,
+      "loss": 2.3596,
+      "step": 19750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5592475628175205e-05,
+      "loss": 2.5304,
+      "step": 19760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.559018719392192e-05,
+      "loss": 2.3825,
+      "step": 19770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.558789875966863e-05,
+      "loss": 2.3109,
+      "step": 19780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5585610325415354e-05,
+      "loss": 2.3204,
+      "step": 19790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.558332189116207e-05,
+      "loss": 2.3479,
+      "step": 19800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.558103345690878e-05,
+      "loss": 2.3186,
+      "step": 19810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.55787450226555e-05,
+      "loss": 2.3685,
+      "step": 19820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.557645658840222e-05,
+      "loss": 2.2943,
+      "step": 19830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.557416815414893e-05,
+      "loss": 2.3816,
+      "step": 19840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.557187971989565e-05,
+      "loss": 2.3416,
+      "step": 19850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5569591285642366e-05,
+      "loss": 2.3142,
+      "step": 19860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.556730285138908e-05,
+      "loss": 2.2883,
+      "step": 19870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.55650144171358e-05,
+      "loss": 2.312,
+      "step": 19880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5562725982882516e-05,
+      "loss": 2.4228,
+      "step": 19890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.556043754862923e-05,
+      "loss": 2.1715,
+      "step": 19900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.555814911437595e-05,
+      "loss": 2.3988,
+      "step": 19910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5555860680122665e-05,
+      "loss": 2.5324,
+      "step": 19920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.555357224586938e-05,
+      "loss": 2.3425,
+      "step": 19930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.55512838116161e-05,
+      "loss": 2.5047,
+      "step": 19940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5548995377362814e-05,
+      "loss": 2.2874,
+      "step": 19950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.554670694310952e-05,
+      "loss": 2.3633,
+      "step": 19960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.554441850885624e-05,
+      "loss": 2.3696,
+      "step": 19970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5542130074602956e-05,
+      "loss": 2.2353,
+      "step": 19980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.553984164034967e-05,
+      "loss": 2.4472,
+      "step": 19990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.553755320609639e-05,
+      "loss": 2.4039,
+      "step": 20000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5535264771843106e-05,
+      "loss": 2.2052,
+      "step": 20010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.553297633758982e-05,
+      "loss": 2.3342,
+      "step": 20020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.553068790333654e-05,
+      "loss": 2.3392,
+      "step": 20030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5528399469083255e-05,
+      "loss": 2.3362,
+      "step": 20040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.552611103482997e-05,
+      "loss": 2.2906,
+      "step": 20050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.552382260057669e-05,
+      "loss": 2.3721,
+      "step": 20060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5521534166323404e-05,
+      "loss": 2.3866,
+      "step": 20070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.551924573207012e-05,
+      "loss": 2.2451,
+      "step": 20080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.551695729781684e-05,
+      "loss": 2.5169,
+      "step": 20090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.551466886356355e-05,
+      "loss": 2.4583,
+      "step": 20100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.551238042931027e-05,
+      "loss": 2.2339,
+      "step": 20110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.551009199505699e-05,
+      "loss": 2.2359,
+      "step": 20120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.55078035608037e-05,
+      "loss": 2.3246,
+      "step": 20130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5505515126550416e-05,
+      "loss": 2.3768,
+      "step": 20140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.550322669229714e-05,
+      "loss": 2.3263,
+      "step": 20150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5500938258043845e-05,
+      "loss": 2.3563,
+      "step": 20160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5498649823790566e-05,
+      "loss": 2.4489,
+      "step": 20170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.549636138953728e-05,
+      "loss": 2.3039,
+      "step": 20180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5494072955283994e-05,
+      "loss": 2.2202,
+      "step": 20190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5491784521030715e-05,
+      "loss": 2.3698,
+      "step": 20200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.548949608677743e-05,
+      "loss": 2.2487,
+      "step": 20210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.548720765252414e-05,
+      "loss": 2.2787,
+      "step": 20220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5484919218270864e-05,
+      "loss": 2.2818,
+      "step": 20230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.548263078401758e-05,
+      "loss": 2.2863,
+      "step": 20240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.548034234976429e-05,
+      "loss": 2.2049,
+      "step": 20250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5478053915511006e-05,
+      "loss": 2.1635,
+      "step": 20260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.547576548125773e-05,
+      "loss": 2.2524,
+      "step": 20270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.547347704700444e-05,
+      "loss": 2.3154,
+      "step": 20280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5471188612751155e-05,
+      "loss": 2.4233,
+      "step": 20290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5468900178497876e-05,
+      "loss": 2.3096,
+      "step": 20300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.546661174424459e-05,
+      "loss": 2.2627,
+      "step": 20310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5464323309991305e-05,
+      "loss": 2.4268,
+      "step": 20320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5462034875738026e-05,
+      "loss": 2.4624,
+      "step": 20330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.545974644148474e-05,
+      "loss": 2.2327,
+      "step": 20340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5457458007231454e-05,
+      "loss": 2.3966,
+      "step": 20350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.545516957297817e-05,
+      "loss": 2.3885,
+      "step": 20360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.545288113872488e-05,
+      "loss": 2.2399,
+      "step": 20370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.54505927044716e-05,
+      "loss": 2.3578,
+      "step": 20380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.544830427021832e-05,
+      "loss": 2.3703,
+      "step": 20390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.544601583596503e-05,
+      "loss": 2.4407,
+      "step": 20400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.544372740171175e-05,
+      "loss": 2.203,
+      "step": 20410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5441438967458466e-05,
+      "loss": 2.5203,
+      "step": 20420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.543915053320518e-05,
+      "loss": 2.3311,
+      "step": 20430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.54368620989519e-05,
+      "loss": 2.3078,
+      "step": 20440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5434573664698616e-05,
+      "loss": 2.3714,
+      "step": 20450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.543228523044533e-05,
+      "loss": 2.5494,
+      "step": 20460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.542999679619205e-05,
+      "loss": 2.5469,
+      "step": 20470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5427708361938765e-05,
+      "loss": 2.2872,
+      "step": 20480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.542541992768548e-05,
+      "loss": 2.1746,
+      "step": 20490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.54231314934322e-05,
+      "loss": 2.3005,
+      "step": 20500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5420843059178914e-05,
+      "loss": 2.234,
+      "step": 20510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.541855462492563e-05,
+      "loss": 2.4276,
+      "step": 20520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.541626619067235e-05,
+      "loss": 2.4703,
+      "step": 20530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.541397775641906e-05,
+      "loss": 2.2527,
+      "step": 20540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.541168932216578e-05,
+      "loss": 2.345,
+      "step": 20550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.540940088791249e-05,
+      "loss": 2.2503,
+      "step": 20560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5407112453659205e-05,
+      "loss": 2.2598,
+      "step": 20570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.540482401940592e-05,
+      "loss": 2.2499,
+      "step": 20580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.540253558515264e-05,
+      "loss": 2.2606,
+      "step": 20590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5400247150899355e-05,
+      "loss": 2.4059,
+      "step": 20600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.539795871664607e-05,
+      "loss": 2.2226,
+      "step": 20610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.539567028239279e-05,
+      "loss": 2.3535,
+      "step": 20620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5393381848139504e-05,
+      "loss": 2.3696,
+      "step": 20630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.539109341388622e-05,
+      "loss": 2.2701,
+      "step": 20640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.538880497963294e-05,
+      "loss": 2.3742,
+      "step": 20650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.538651654537965e-05,
+      "loss": 2.3911,
+      "step": 20660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.538422811112637e-05,
+      "loss": 2.4521,
+      "step": 20670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.538193967687309e-05,
+      "loss": 2.2415,
+      "step": 20680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.53796512426198e-05,
+      "loss": 2.1932,
+      "step": 20690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5377362808366516e-05,
+      "loss": 2.324,
+      "step": 20700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.537507437411324e-05,
+      "loss": 2.2722,
+      "step": 20710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.537278593985995e-05,
+      "loss": 2.2937,
+      "step": 20720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5370497505606665e-05,
+      "loss": 2.1996,
+      "step": 20730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5368209071353386e-05,
+      "loss": 2.2959,
+      "step": 20740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5365920637100094e-05,
+      "loss": 2.2833,
+      "step": 20750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5363632202846815e-05,
+      "loss": 2.2439,
+      "step": 20760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.536134376859353e-05,
+      "loss": 2.3691,
+      "step": 20770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.535905533434024e-05,
+      "loss": 2.3087,
+      "step": 20780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5356766900086964e-05,
+      "loss": 2.3394,
+      "step": 20790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.535447846583368e-05,
+      "loss": 2.2847,
+      "step": 20800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.535219003158039e-05,
+      "loss": 2.3594,
+      "step": 20810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.534990159732711e-05,
+      "loss": 2.3482,
+      "step": 20820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.534761316307383e-05,
+      "loss": 2.3275,
+      "step": 20830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.534532472882054e-05,
+      "loss": 2.5,
+      "step": 20840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.534303629456726e-05,
+      "loss": 2.3864,
+      "step": 20850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5340747860313976e-05,
+      "loss": 2.2909,
+      "step": 20860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.533845942606069e-05,
+      "loss": 2.3373,
+      "step": 20870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.533617099180741e-05,
+      "loss": 2.3343,
+      "step": 20880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5333882557554125e-05,
+      "loss": 2.2317,
+      "step": 20890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.533159412330084e-05,
+      "loss": 2.5133,
+      "step": 20900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.532930568904756e-05,
+      "loss": 2.3618,
+      "step": 20910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5327017254794275e-05,
+      "loss": 2.2439,
+      "step": 20920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.532472882054099e-05,
+      "loss": 2.4075,
+      "step": 20930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.532244038628771e-05,
+      "loss": 2.2629,
+      "step": 20940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.532015195203442e-05,
+      "loss": 2.3173,
+      "step": 20950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.531786351778113e-05,
+      "loss": 2.382,
+      "step": 20960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.531557508352785e-05,
+      "loss": 2.3389,
+      "step": 20970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5313286649274566e-05,
+      "loss": 2.3886,
+      "step": 20980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.531099821502128e-05,
+      "loss": 2.3159,
+      "step": 20990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5308709780768e-05,
+      "loss": 2.2627,
+      "step": 21000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5306421346514715e-05,
+      "loss": 2.2969,
+      "step": 21010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.530413291226143e-05,
+      "loss": 2.2948,
+      "step": 21020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.530184447800815e-05,
+      "loss": 2.3191,
+      "step": 21030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5299556043754865e-05,
+      "loss": 2.3351,
+      "step": 21040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.529726760950158e-05,
+      "loss": 2.3544,
+      "step": 21050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.52949791752483e-05,
+      "loss": 2.3897,
+      "step": 21060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5292690740995014e-05,
+      "loss": 2.4179,
+      "step": 21070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.529040230674173e-05,
+      "loss": 2.3701,
+      "step": 21080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.528811387248845e-05,
+      "loss": 2.2162,
+      "step": 21090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.528582543823516e-05,
+      "loss": 2.3018,
+      "step": 21100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.528353700398188e-05,
+      "loss": 2.3175,
+      "step": 21110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.52812485697286e-05,
+      "loss": 2.3634,
+      "step": 21120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.527896013547531e-05,
+      "loss": 2.3329,
+      "step": 21130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5276671701222026e-05,
+      "loss": 2.165,
+      "step": 21140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.527438326696874e-05,
+      "loss": 2.4515,
+      "step": 21150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5272094832715455e-05,
+      "loss": 2.3241,
+      "step": 21160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5269806398462175e-05,
+      "loss": 2.4588,
+      "step": 21170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.526751796420889e-05,
+      "loss": 2.4249,
+      "step": 21180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5265229529955604e-05,
+      "loss": 2.2971,
+      "step": 21190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5262941095702325e-05,
+      "loss": 2.3665,
+      "step": 21200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.526065266144904e-05,
+      "loss": 2.2997,
+      "step": 21210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.525836422719575e-05,
+      "loss": 2.3432,
+      "step": 21220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5256075792942474e-05,
+      "loss": 2.1502,
+      "step": 21230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.525378735868919e-05,
+      "loss": 2.2443,
+      "step": 21240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.52514989244359e-05,
+      "loss": 2.6131,
+      "step": 21250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.524921049018262e-05,
+      "loss": 2.3306,
+      "step": 21260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.524692205592934e-05,
+      "loss": 2.231,
+      "step": 21270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.524463362167605e-05,
+      "loss": 2.2551,
+      "step": 21280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.524234518742277e-05,
+      "loss": 2.223,
+      "step": 21290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5240056753169486e-05,
+      "loss": 2.2982,
+      "step": 21300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.52377683189162e-05,
+      "loss": 2.246,
+      "step": 21310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.523547988466292e-05,
+      "loss": 2.3108,
+      "step": 21320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5233191450409635e-05,
+      "loss": 2.3256,
+      "step": 21330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.523090301615635e-05,
+      "loss": 2.2823,
+      "step": 21340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5228614581903064e-05,
+      "loss": 2.1821,
+      "step": 21350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.522632614764978e-05,
+      "loss": 2.2537,
+      "step": 21360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.522403771339649e-05,
+      "loss": 2.2024,
+      "step": 21370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.522174927914321e-05,
+      "loss": 2.3644,
+      "step": 21380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.521946084488993e-05,
+      "loss": 2.4402,
+      "step": 21390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.521717241063664e-05,
+      "loss": 2.4038,
+      "step": 21400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.521488397638336e-05,
+      "loss": 2.3156,
+      "step": 21410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5212595542130076e-05,
+      "loss": 2.3538,
+      "step": 21420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.521030710787679e-05,
+      "loss": 2.3233,
+      "step": 21430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.520801867362351e-05,
+      "loss": 2.2834,
+      "step": 21440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5205730239370225e-05,
+      "loss": 2.3718,
+      "step": 21450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.520344180511694e-05,
+      "loss": 2.1681,
+      "step": 21460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.520115337086366e-05,
+      "loss": 2.4196,
+      "step": 21470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5198864936610375e-05,
+      "loss": 2.3599,
+      "step": 21480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.519657650235709e-05,
+      "loss": 2.5683,
+      "step": 21490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.519428806810381e-05,
+      "loss": 2.2748,
+      "step": 21500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5191999633850524e-05,
+      "loss": 2.3012,
+      "step": 21510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.518971119959724e-05,
+      "loss": 2.1967,
+      "step": 21520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.518742276534396e-05,
+      "loss": 2.3538,
+      "step": 21530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5185134331090666e-05,
+      "loss": 2.2923,
+      "step": 21540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.518284589683738e-05,
+      "loss": 2.1332,
+      "step": 21550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.51805574625841e-05,
+      "loss": 2.2812,
+      "step": 21560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5178269028330815e-05,
+      "loss": 2.5015,
+      "step": 21570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.517598059407753e-05,
+      "loss": 2.1773,
+      "step": 21580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.517369215982425e-05,
+      "loss": 2.1177,
+      "step": 21590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5171403725570964e-05,
+      "loss": 2.2354,
+      "step": 21600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.516911529131768e-05,
+      "loss": 2.3471,
+      "step": 21610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.51668268570644e-05,
+      "loss": 2.2099,
+      "step": 21620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5164538422811114e-05,
+      "loss": 2.2166,
+      "step": 21630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.516224998855783e-05,
+      "loss": 2.3403,
+      "step": 21640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.515996155430455e-05,
+      "loss": 2.4002,
+      "step": 21650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.515767312005126e-05,
+      "loss": 2.4118,
+      "step": 21660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.515538468579798e-05,
+      "loss": 2.078,
+      "step": 21670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.51530962515447e-05,
+      "loss": 2.3969,
+      "step": 21680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.515080781729141e-05,
+      "loss": 2.4961,
+      "step": 21690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5148519383038126e-05,
+      "loss": 2.3624,
+      "step": 21700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.514623094878485e-05,
+      "loss": 2.4318,
+      "step": 21710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.514394251453156e-05,
+      "loss": 2.3499,
+      "step": 21720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5141654080278275e-05,
+      "loss": 2.3653,
+      "step": 21730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.513936564602499e-05,
+      "loss": 2.177,
+      "step": 21740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5137077211771704e-05,
+      "loss": 2.391,
+      "step": 21750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5134788777518424e-05,
+      "loss": 2.2926,
+      "step": 21760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.513250034326514e-05,
+      "loss": 2.3313,
+      "step": 21770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.513021190901185e-05,
+      "loss": 2.3778,
+      "step": 21780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5127923474758574e-05,
+      "loss": 2.3978,
+      "step": 21790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.512563504050529e-05,
+      "loss": 2.3711,
+      "step": 21800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5123346606252e-05,
+      "loss": 2.231,
+      "step": 21810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.512105817199872e-05,
+      "loss": 2.3072,
+      "step": 21820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.511876973774544e-05,
+      "loss": 2.3983,
+      "step": 21830
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.511648130349215e-05,
+      "loss": 2.5131,
+      "step": 21840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.511419286923887e-05,
+      "loss": 2.3321,
+      "step": 21850
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5111904434985586e-05,
+      "loss": 2.2685,
+      "step": 21860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.51096160007323e-05,
+      "loss": 2.29,
+      "step": 21870
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.510732756647902e-05,
+      "loss": 2.2824,
+      "step": 21880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5105039132225735e-05,
+      "loss": 2.4528,
+      "step": 21890
+    },
+    {
+      "epoch": 1.0,
+      "eval_cer": 0.6811390598773349,
+      "eval_em": 0.00728476821192053,
+      "eval_f1": 0.00728476821192053,
+      "eval_loss": 2.216672658920288,
+      "eval_runtime": 2775.0638,
+      "eval_samples_per_second": 3.809,
+      "eval_steps_per_second": 1.904,
+      "eval_wer": 0.9927152317880795,
+      "step": 21899
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.510275069797245e-05,
+      "loss": 2.2786,
+      "step": 21900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.510046226371917e-05,
+      "loss": 2.3984,
+      "step": 21910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5098173829465885e-05,
+      "loss": 2.3169,
+      "step": 21920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.50958853952126e-05,
+      "loss": 2.1992,
+      "step": 21930
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.509359696095931e-05,
+      "loss": 2.4089,
+      "step": 21940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.509130852670603e-05,
+      "loss": 2.3878,
+      "step": 21950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.508902009245274e-05,
+      "loss": 2.2885,
+      "step": 21960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.508673165819946e-05,
+      "loss": 2.2094,
+      "step": 21970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5084443223946176e-05,
+      "loss": 2.2895,
+      "step": 21980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.508215478969289e-05,
+      "loss": 2.3139,
+      "step": 21990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.507986635543961e-05,
+      "loss": 2.3456,
+      "step": 22000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5077577921186325e-05,
+      "loss": 2.3049,
+      "step": 22010
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.507528948693304e-05,
+      "loss": 2.3241,
+      "step": 22020
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.507300105267976e-05,
+      "loss": 2.3352,
+      "step": 22030
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5070712618426474e-05,
+      "loss": 2.2437,
+      "step": 22040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.506842418417319e-05,
+      "loss": 2.3977,
+      "step": 22050
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.506613574991991e-05,
+      "loss": 2.3157,
+      "step": 22060
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5063847315666624e-05,
+      "loss": 2.2105,
+      "step": 22070
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.506155888141334e-05,
+      "loss": 2.2588,
+      "step": 22080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.505927044716006e-05,
+      "loss": 2.3755,
+      "step": 22090
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.505698201290677e-05,
+      "loss": 2.3414,
+      "step": 22100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.505469357865349e-05,
+      "loss": 2.3759,
+      "step": 22110
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.505240514440021e-05,
+      "loss": 2.2631,
+      "step": 22120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.505011671014692e-05,
+      "loss": 2.3923,
+      "step": 22130
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5047828275893636e-05,
+      "loss": 2.404,
+      "step": 22140
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.504553984164035e-05,
+      "loss": 2.1583,
+      "step": 22150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5043251407387064e-05,
+      "loss": 2.2416,
+      "step": 22160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5040962973133785e-05,
+      "loss": 2.2574,
+      "step": 22170
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.50386745388805e-05,
+      "loss": 2.1974,
+      "step": 22180
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5036386104627214e-05,
+      "loss": 2.2919,
+      "step": 22190
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5034097670373934e-05,
+      "loss": 2.1923,
+      "step": 22200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.503180923612065e-05,
+      "loss": 2.4379,
+      "step": 22210
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.502952080186736e-05,
+      "loss": 2.2519,
+      "step": 22220
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5027232367614084e-05,
+      "loss": 2.4246,
+      "step": 22230
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.50249439333608e-05,
+      "loss": 2.2347,
+      "step": 22240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.502265549910751e-05,
+      "loss": 2.1819,
+      "step": 22250
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.502036706485423e-05,
+      "loss": 2.1016,
+      "step": 22260
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.501807863060095e-05,
+      "loss": 2.3312,
+      "step": 22270
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.501579019634766e-05,
+      "loss": 2.3409,
+      "step": 22280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.501350176209438e-05,
+      "loss": 2.2873,
+      "step": 22290
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5011213327841096e-05,
+      "loss": 2.1861,
+      "step": 22300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.500892489358781e-05,
+      "loss": 2.1877,
+      "step": 22310
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.500663645933453e-05,
+      "loss": 2.1805,
+      "step": 22320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.500434802508124e-05,
+      "loss": 2.3718,
+      "step": 22330
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.500205959082795e-05,
+      "loss": 2.3247,
+      "step": 22340
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4999771156574674e-05,
+      "loss": 2.1953,
+      "step": 22350
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.499748272232139e-05,
+      "loss": 2.2575,
+      "step": 22360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.49951942880681e-05,
+      "loss": 2.4066,
+      "step": 22370
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.499290585381482e-05,
+      "loss": 2.3497,
+      "step": 22380
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.499061741956154e-05,
+      "loss": 2.3043,
+      "step": 22390
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.498832898530825e-05,
+      "loss": 2.4081,
+      "step": 22400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.498604055105497e-05,
+      "loss": 2.237,
+      "step": 22410
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4983752116801686e-05,
+      "loss": 2.4168,
+      "step": 22420
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.49814636825484e-05,
+      "loss": 2.2873,
+      "step": 22430
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.497917524829512e-05,
+      "loss": 2.2442,
+      "step": 22440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4976886814041835e-05,
+      "loss": 2.5326,
+      "step": 22450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.497459837978855e-05,
+      "loss": 2.1366,
+      "step": 22460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.497230994553527e-05,
+      "loss": 2.3103,
+      "step": 22470
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4970021511281984e-05,
+      "loss": 2.3425,
+      "step": 22480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.49677330770287e-05,
+      "loss": 2.3082,
+      "step": 22490
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.496544464277542e-05,
+      "loss": 2.2169,
+      "step": 22500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4963156208522134e-05,
+      "loss": 2.3988,
+      "step": 22510
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.496086777426885e-05,
+      "loss": 2.2196,
+      "step": 22520
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.495857934001556e-05,
+      "loss": 2.2251,
+      "step": 22530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4956290905762276e-05,
+      "loss": 2.4297,
+      "step": 22540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4954002471509e-05,
+      "loss": 2.1783,
+      "step": 22550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.495171403725571e-05,
+      "loss": 2.234,
+      "step": 22560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4949425603002425e-05,
+      "loss": 2.3437,
+      "step": 22570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4947137168749146e-05,
+      "loss": 2.3595,
+      "step": 22580
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.494484873449586e-05,
+      "loss": 2.3109,
+      "step": 22590
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4942560300242574e-05,
+      "loss": 2.1046,
+      "step": 22600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.494027186598929e-05,
+      "loss": 2.2919,
+      "step": 22610
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.493798343173601e-05,
+      "loss": 2.2704,
+      "step": 22620
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4935694997482724e-05,
+      "loss": 2.295,
+      "step": 22630
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.493340656322944e-05,
+      "loss": 2.3103,
+      "step": 22640
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.493111812897616e-05,
+      "loss": 2.3212,
+      "step": 22650
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.492882969472287e-05,
+      "loss": 2.244,
+      "step": 22660
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.492654126046959e-05,
+      "loss": 2.4235,
+      "step": 22670
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.492425282621631e-05,
+      "loss": 2.3367,
+      "step": 22680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.492196439196302e-05,
+      "loss": 2.1472,
+      "step": 22690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4919675957709736e-05,
+      "loss": 2.2936,
+      "step": 22700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.491738752345646e-05,
+      "loss": 2.321,
+      "step": 22710
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.491509908920317e-05,
+      "loss": 2.3483,
+      "step": 22720
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4912810654949885e-05,
+      "loss": 2.2685,
+      "step": 22730
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.49105222206966e-05,
+      "loss": 2.2381,
+      "step": 22740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4908233786443313e-05,
+      "loss": 2.1682,
+      "step": 22750
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4905945352190034e-05,
+      "loss": 2.3424,
+      "step": 22760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.490365691793675e-05,
+      "loss": 2.2654,
+      "step": 22770
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.490136848368346e-05,
+      "loss": 2.2483,
+      "step": 22780
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4899080049430184e-05,
+      "loss": 2.2085,
+      "step": 22790
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.48967916151769e-05,
+      "loss": 2.2406,
+      "step": 22800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.489450318092361e-05,
+      "loss": 2.1984,
+      "step": 22810
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.489221474667033e-05,
+      "loss": 2.3007,
+      "step": 22820
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.488992631241705e-05,
+      "loss": 2.2391,
+      "step": 22830
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.488763787816376e-05,
+      "loss": 2.2319,
+      "step": 22840
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.488534944391048e-05,
+      "loss": 2.3318,
+      "step": 22850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4883061009657196e-05,
+      "loss": 2.2977,
+      "step": 22860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.488077257540391e-05,
+      "loss": 2.2229,
+      "step": 22870
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.487848414115063e-05,
+      "loss": 2.2238,
+      "step": 22880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4876195706897345e-05,
+      "loss": 2.3456,
+      "step": 22890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.487390727264406e-05,
+      "loss": 2.2562,
+      "step": 22900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.487161883839078e-05,
+      "loss": 2.244,
+      "step": 22910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4869330404137494e-05,
+      "loss": 2.3936,
+      "step": 22920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.48670419698842e-05,
+      "loss": 2.3535,
+      "step": 22930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.486475353563092e-05,
+      "loss": 2.215,
+      "step": 22940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.486246510137764e-05,
+      "loss": 2.2646,
+      "step": 22950
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.486017666712435e-05,
+      "loss": 2.3649,
+      "step": 22960
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.485788823287107e-05,
+      "loss": 2.2424,
+      "step": 22970
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4855599798617786e-05,
+      "loss": 2.2997,
+      "step": 22980
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.48533113643645e-05,
+      "loss": 2.2354,
+      "step": 22990
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.485102293011122e-05,
+      "loss": 2.2004,
+      "step": 23000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4848734495857935e-05,
+      "loss": 2.3738,
+      "step": 23010
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.484644606160465e-05,
+      "loss": 2.2414,
+      "step": 23020
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.484415762735137e-05,
+      "loss": 2.2734,
+      "step": 23030
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4841869193098084e-05,
+      "loss": 2.2958,
+      "step": 23040
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.48395807588448e-05,
+      "loss": 2.5156,
+      "step": 23050
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.483729232459152e-05,
+      "loss": 2.3496,
+      "step": 23060
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4835003890338233e-05,
+      "loss": 2.3029,
+      "step": 23070
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.483271545608495e-05,
+      "loss": 2.3776,
+      "step": 23080
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.483042702183167e-05,
+      "loss": 2.3026,
+      "step": 23090
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.482813858757838e-05,
+      "loss": 2.2881,
+      "step": 23100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.48258501533251e-05,
+      "loss": 2.2495,
+      "step": 23110
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.482356171907181e-05,
+      "loss": 2.366,
+      "step": 23120
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4821273284818525e-05,
+      "loss": 2.1953,
+      "step": 23130
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4818984850565246e-05,
+      "loss": 2.1652,
+      "step": 23140
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.481669641631196e-05,
+      "loss": 2.1998,
+      "step": 23150
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4814407982058674e-05,
+      "loss": 2.4456,
+      "step": 23160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4812119547805395e-05,
+      "loss": 2.2987,
+      "step": 23170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.480983111355211e-05,
+      "loss": 2.4834,
+      "step": 23180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4807542679298823e-05,
+      "loss": 2.2657,
+      "step": 23190
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4805254245045544e-05,
+      "loss": 2.2543,
+      "step": 23200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.480296581079226e-05,
+      "loss": 2.2808,
+      "step": 23210
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.480067737653897e-05,
+      "loss": 2.2906,
+      "step": 23220
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4798388942285694e-05,
+      "loss": 2.3544,
+      "step": 23230
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.479610050803241e-05,
+      "loss": 2.1754,
+      "step": 23240
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.479381207377912e-05,
+      "loss": 2.6805,
+      "step": 23250
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.479152363952584e-05,
+      "loss": 2.3576,
+      "step": 23260
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.478923520527256e-05,
+      "loss": 2.4817,
+      "step": 23270
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.478694677101927e-05,
+      "loss": 2.2975,
+      "step": 23280
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.478465833676599e-05,
+      "loss": 2.1194,
+      "step": 23290
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4782369902512706e-05,
+      "loss": 2.2762,
+      "step": 23300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.478008146825942e-05,
+      "loss": 2.3417,
+      "step": 23310
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4777793034006134e-05,
+      "loss": 2.3347,
+      "step": 23320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.477550459975285e-05,
+      "loss": 2.1626,
+      "step": 23330
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.477321616549956e-05,
+      "loss": 2.2611,
+      "step": 23340
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4770927731246283e-05,
+      "loss": 2.2869,
+      "step": 23350
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4768639296993e-05,
+      "loss": 2.356,
+      "step": 23360
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.476635086273971e-05,
+      "loss": 2.2403,
+      "step": 23370
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.476406242848643e-05,
+      "loss": 2.3383,
+      "step": 23380
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.476177399423315e-05,
+      "loss": 2.2679,
+      "step": 23390
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.475948555997986e-05,
+      "loss": 2.2827,
+      "step": 23400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.475719712572658e-05,
+      "loss": 2.2793,
+      "step": 23410
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4754908691473296e-05,
+      "loss": 2.1258,
+      "step": 23420
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.475262025722001e-05,
+      "loss": 2.1554,
+      "step": 23430
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.475033182296673e-05,
+      "loss": 2.2401,
+      "step": 23440
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4748043388713445e-05,
+      "loss": 2.397,
+      "step": 23450
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.474575495446016e-05,
+      "loss": 2.4403,
+      "step": 23460
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.474346652020688e-05,
+      "loss": 2.1008,
+      "step": 23470
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4741178085953594e-05,
+      "loss": 2.1997,
+      "step": 23480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.473888965170031e-05,
+      "loss": 2.2415,
+      "step": 23490
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.473660121744703e-05,
+      "loss": 2.299,
+      "step": 23500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4734312783193743e-05,
+      "loss": 2.3859,
+      "step": 23510
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.473202434894046e-05,
+      "loss": 2.2103,
+      "step": 23520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.472973591468717e-05,
+      "loss": 2.3727,
+      "step": 23530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4727447480433886e-05,
+      "loss": 2.2905,
+      "step": 23540
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.472515904618061e-05,
+      "loss": 2.3087,
+      "step": 23550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.472287061192732e-05,
+      "loss": 2.2841,
+      "step": 23560
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4720582177674035e-05,
+      "loss": 2.2952,
+      "step": 23570
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4718293743420756e-05,
+      "loss": 2.3625,
+      "step": 23580
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.471600530916747e-05,
+      "loss": 2.2248,
+      "step": 23590
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4713716874914184e-05,
+      "loss": 2.3198,
+      "step": 23600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4711428440660905e-05,
+      "loss": 2.2969,
+      "step": 23610
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.470914000640762e-05,
+      "loss": 2.379,
+      "step": 23620
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.470685157215433e-05,
+      "loss": 2.3187,
+      "step": 23630
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4704563137901054e-05,
+      "loss": 2.2606,
+      "step": 23640
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.470227470364777e-05,
+      "loss": 2.2952,
+      "step": 23650
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.469998626939448e-05,
+      "loss": 2.2486,
+      "step": 23660
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4697697835141203e-05,
+      "loss": 2.3264,
+      "step": 23670
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.469540940088792e-05,
+      "loss": 2.2853,
+      "step": 23680
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.469312096663463e-05,
+      "loss": 2.1832,
+      "step": 23690
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.469083253238135e-05,
+      "loss": 2.273,
+      "step": 23700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.468854409812807e-05,
+      "loss": 2.2548,
+      "step": 23710
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4686255663874774e-05,
+      "loss": 2.2789,
+      "step": 23720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4683967229621495e-05,
+      "loss": 2.2469,
+      "step": 23730
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.468167879536821e-05,
+      "loss": 2.286,
+      "step": 23740
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.467939036111492e-05,
+      "loss": 2.4641,
+      "step": 23750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4677101926861644e-05,
+      "loss": 2.186,
+      "step": 23760
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.467481349260836e-05,
+      "loss": 2.405,
+      "step": 23770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.467252505835507e-05,
+      "loss": 2.2874,
+      "step": 23780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4670236624101793e-05,
+      "loss": 2.5123,
+      "step": 23790
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.466794818984851e-05,
+      "loss": 2.2953,
+      "step": 23800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.466565975559522e-05,
+      "loss": 2.3845,
+      "step": 23810
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.466337132134194e-05,
+      "loss": 2.3188,
+      "step": 23820
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.466108288708866e-05,
+      "loss": 2.3553,
+      "step": 23830
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.465879445283537e-05,
+      "loss": 2.3429,
+      "step": 23840
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.465650601858209e-05,
+      "loss": 2.4726,
+      "step": 23850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4654217584328806e-05,
+      "loss": 2.1972,
+      "step": 23860
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.465192915007552e-05,
+      "loss": 2.1696,
+      "step": 23870
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.464964071582224e-05,
+      "loss": 2.1854,
+      "step": 23880
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4647352281568955e-05,
+      "loss": 2.2547,
+      "step": 23890
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.464506384731567e-05,
+      "loss": 2.2351,
+      "step": 23900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.464277541306238e-05,
+      "loss": 2.213,
+      "step": 23910
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.46404869788091e-05,
+      "loss": 2.2166,
+      "step": 23920
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.463819854455581e-05,
+      "loss": 2.3071,
+      "step": 23930
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.463591011030253e-05,
+      "loss": 2.3706,
+      "step": 23940
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.463362167604925e-05,
+      "loss": 2.2887,
+      "step": 23950
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.463133324179596e-05,
+      "loss": 2.1549,
+      "step": 23960
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.462904480754268e-05,
+      "loss": 2.4724,
+      "step": 23970
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4626756373289396e-05,
+      "loss": 2.2536,
+      "step": 23980
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.462446793903611e-05,
+      "loss": 2.2249,
+      "step": 23990
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.462217950478283e-05,
+      "loss": 2.1397,
+      "step": 24000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4619891070529545e-05,
+      "loss": 2.2628,
+      "step": 24010
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.461760263627626e-05,
+      "loss": 2.2585,
+      "step": 24020
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.461531420202298e-05,
+      "loss": 2.2245,
+      "step": 24030
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4613025767769694e-05,
+      "loss": 2.2137,
+      "step": 24040
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.461073733351641e-05,
+      "loss": 2.2693,
+      "step": 24050
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.460844889926313e-05,
+      "loss": 2.3853,
+      "step": 24060
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.460616046500984e-05,
+      "loss": 2.3593,
+      "step": 24070
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.460387203075656e-05,
+      "loss": 2.2257,
+      "step": 24080
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.460158359650328e-05,
+      "loss": 2.373,
+      "step": 24090
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.459929516224999e-05,
+      "loss": 2.1701,
+      "step": 24100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.459700672799671e-05,
+      "loss": 2.2545,
+      "step": 24110
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.459471829374342e-05,
+      "loss": 2.2352,
+      "step": 24120
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4592429859490135e-05,
+      "loss": 2.2635,
+      "step": 24130
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4590141425236856e-05,
+      "loss": 2.3063,
+      "step": 24140
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.458785299098357e-05,
+      "loss": 2.3205,
+      "step": 24150
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4585564556730284e-05,
+      "loss": 2.2775,
+      "step": 24160
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4583276122477005e-05,
+      "loss": 2.2944,
+      "step": 24170
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.458098768822372e-05,
+      "loss": 2.3028,
+      "step": 24180
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.457869925397043e-05,
+      "loss": 2.3336,
+      "step": 24190
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4576410819717154e-05,
+      "loss": 2.2674,
+      "step": 24200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.457412238546387e-05,
+      "loss": 2.2639,
+      "step": 24210
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.457183395121058e-05,
+      "loss": 2.3059,
+      "step": 24220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.45695455169573e-05,
+      "loss": 2.2512,
+      "step": 24230
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.456725708270402e-05,
+      "loss": 2.2663,
+      "step": 24240
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.456496864845073e-05,
+      "loss": 2.2695,
+      "step": 24250
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.456268021419745e-05,
+      "loss": 2.3058,
+      "step": 24260
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.456039177994417e-05,
+      "loss": 2.1553,
+      "step": 24270
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.455810334569088e-05,
+      "loss": 2.3837,
+      "step": 24280
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.45558149114376e-05,
+      "loss": 2.3801,
+      "step": 24290
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4553526477184316e-05,
+      "loss": 2.3198,
+      "step": 24300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.455123804293102e-05,
+      "loss": 2.2181,
+      "step": 24310
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4548949608677744e-05,
+      "loss": 2.1765,
+      "step": 24320
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.454666117442446e-05,
+      "loss": 2.2311,
+      "step": 24330
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.454437274017117e-05,
+      "loss": 2.3063,
+      "step": 24340
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.454208430591789e-05,
+      "loss": 2.1993,
+      "step": 24350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.453979587166461e-05,
+      "loss": 2.1779,
+      "step": 24360
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.453750743741132e-05,
+      "loss": 2.2112,
+      "step": 24370
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.453521900315804e-05,
+      "loss": 2.1919,
+      "step": 24380
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.453293056890476e-05,
+      "loss": 2.3088,
+      "step": 24390
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.453064213465147e-05,
+      "loss": 2.2676,
+      "step": 24400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.452835370039819e-05,
+      "loss": 2.3745,
+      "step": 24410
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4526065266144906e-05,
+      "loss": 2.1507,
+      "step": 24420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.452377683189162e-05,
+      "loss": 2.304,
+      "step": 24430
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.452148839763834e-05,
+      "loss": 2.1808,
+      "step": 24440
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4519199963385055e-05,
+      "loss": 2.2421,
+      "step": 24450
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.451691152913177e-05,
+      "loss": 2.3523,
+      "step": 24460
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.451462309487849e-05,
+      "loss": 2.2287,
+      "step": 24470
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4512334660625204e-05,
+      "loss": 2.1491,
+      "step": 24480
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.451004622637192e-05,
+      "loss": 2.2709,
+      "step": 24490
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.450775779211864e-05,
+      "loss": 2.124,
+      "step": 24500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4505469357865347e-05,
+      "loss": 2.2316,
+      "step": 24510
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.450318092361207e-05,
+      "loss": 2.1693,
+      "step": 24520
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.450089248935878e-05,
+      "loss": 2.2246,
+      "step": 24530
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4498604055105496e-05,
+      "loss": 2.1091,
+      "step": 24540
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.449631562085222e-05,
+      "loss": 2.2332,
+      "step": 24550
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.449402718659893e-05,
+      "loss": 2.2496,
+      "step": 24560
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4491738752345645e-05,
+      "loss": 2.2261,
+      "step": 24570
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4489450318092366e-05,
+      "loss": 2.2377,
+      "step": 24580
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.448716188383908e-05,
+      "loss": 2.3601,
+      "step": 24590
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4484873449585794e-05,
+      "loss": 2.1195,
+      "step": 24600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4482585015332515e-05,
+      "loss": 2.2967,
+      "step": 24610
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.448029658107923e-05,
+      "loss": 2.2698,
+      "step": 24620
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.447800814682594e-05,
+      "loss": 2.2255,
+      "step": 24630
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4475719712572664e-05,
+      "loss": 2.159,
+      "step": 24640
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.447343127831938e-05,
+      "loss": 2.2749,
+      "step": 24650
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.447114284406609e-05,
+      "loss": 2.1301,
+      "step": 24660
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.446885440981281e-05,
+      "loss": 2.2038,
+      "step": 24670
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.446656597555953e-05,
+      "loss": 2.4247,
+      "step": 24680
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.446427754130624e-05,
+      "loss": 2.2751,
+      "step": 24690
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4461989107052956e-05,
+      "loss": 2.2593,
+      "step": 24700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.445970067279967e-05,
+      "loss": 2.448,
+      "step": 24710
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4457412238546384e-05,
+      "loss": 2.4001,
+      "step": 24720
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4455123804293105e-05,
+      "loss": 2.2066,
+      "step": 24730
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.445283537003982e-05,
+      "loss": 2.1857,
+      "step": 24740
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.445054693578653e-05,
+      "loss": 2.3528,
+      "step": 24750
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4448258501533254e-05,
+      "loss": 2.3385,
+      "step": 24760
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.444597006727997e-05,
+      "loss": 2.1986,
+      "step": 24770
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.444368163302668e-05,
+      "loss": 2.4069,
+      "step": 24780
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.44413931987734e-05,
+      "loss": 2.1739,
+      "step": 24790
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.443910476452012e-05,
+      "loss": 2.2184,
+      "step": 24800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.443681633026683e-05,
+      "loss": 2.302,
+      "step": 24810
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.443452789601355e-05,
+      "loss": 2.4128,
+      "step": 24820
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4432239461760267e-05,
+      "loss": 2.4487,
+      "step": 24830
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.442995102750698e-05,
+      "loss": 2.3491,
+      "step": 24840
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.44276625932537e-05,
+      "loss": 2.211,
+      "step": 24850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4425374159000416e-05,
+      "loss": 2.3704,
+      "step": 24860
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.442308572474713e-05,
+      "loss": 2.0797,
+      "step": 24870
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.442079729049385e-05,
+      "loss": 2.2076,
+      "step": 24880
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4418508856240565e-05,
+      "loss": 2.2729,
+      "step": 24890
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.441622042198728e-05,
+      "loss": 2.1714,
+      "step": 24900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.441393198773399e-05,
+      "loss": 2.45,
+      "step": 24910
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.441164355348071e-05,
+      "loss": 2.2123,
+      "step": 24920
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.440935511922743e-05,
+      "loss": 2.249,
+      "step": 24930
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.440706668497414e-05,
+      "loss": 2.2784,
+      "step": 24940
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4404778250720857e-05,
+      "loss": 2.155,
+      "step": 24950
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.440248981646757e-05,
+      "loss": 2.21,
+      "step": 24960
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.440020138221429e-05,
+      "loss": 2.3263,
+      "step": 24970
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4397912947961006e-05,
+      "loss": 2.2638,
+      "step": 24980
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.439562451370772e-05,
+      "loss": 2.243,
+      "step": 24990
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.439333607945444e-05,
+      "loss": 2.3748,
+      "step": 25000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4391047645201155e-05,
+      "loss": 2.3651,
+      "step": 25010
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.438875921094787e-05,
+      "loss": 2.2125,
+      "step": 25020
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.438647077669459e-05,
+      "loss": 2.4181,
+      "step": 25030
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4384182342441304e-05,
+      "loss": 2.312,
+      "step": 25040
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.438189390818802e-05,
+      "loss": 2.2971,
+      "step": 25050
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.437960547393474e-05,
+      "loss": 2.239,
+      "step": 25060
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.437731703968145e-05,
+      "loss": 2.1821,
+      "step": 25070
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.437502860542817e-05,
+      "loss": 2.3857,
+      "step": 25080
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.437274017117489e-05,
+      "loss": 2.255,
+      "step": 25090
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4370451736921596e-05,
+      "loss": 2.1913,
+      "step": 25100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4368163302668317e-05,
+      "loss": 2.3231,
+      "step": 25110
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.436587486841503e-05,
+      "loss": 2.2103,
+      "step": 25120
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4363586434161745e-05,
+      "loss": 2.2058,
+      "step": 25130
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4361297999908466e-05,
+      "loss": 2.2103,
+      "step": 25140
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.435900956565518e-05,
+      "loss": 2.2999,
+      "step": 25150
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4356721131401894e-05,
+      "loss": 2.2136,
+      "step": 25160
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4354432697148615e-05,
+      "loss": 2.4168,
+      "step": 25170
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.435214426289533e-05,
+      "loss": 2.3901,
+      "step": 25180
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.434985582864204e-05,
+      "loss": 2.3331,
+      "step": 25190
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4347567394388764e-05,
+      "loss": 2.1937,
+      "step": 25200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.434527896013548e-05,
+      "loss": 2.1437,
+      "step": 25210
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.434299052588219e-05,
+      "loss": 2.2642,
+      "step": 25220
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.434070209162891e-05,
+      "loss": 2.2769,
+      "step": 25230
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.433841365737563e-05,
+      "loss": 2.2554,
+      "step": 25240
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.433612522312234e-05,
+      "loss": 2.259,
+      "step": 25250
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.433383678886906e-05,
+      "loss": 2.1573,
+      "step": 25260
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4331548354615777e-05,
+      "loss": 2.0877,
+      "step": 25270
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.432925992036249e-05,
+      "loss": 2.2957,
+      "step": 25280
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.432697148610921e-05,
+      "loss": 2.1028,
+      "step": 25290
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.432468305185592e-05,
+      "loss": 2.2047,
+      "step": 25300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.432239461760263e-05,
+      "loss": 2.2976,
+      "step": 25310
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4320106183349354e-05,
+      "loss": 2.2646,
+      "step": 25320
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.431781774909607e-05,
+      "loss": 2.3561,
+      "step": 25330
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.431552931484278e-05,
+      "loss": 2.3223,
+      "step": 25340
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.43132408805895e-05,
+      "loss": 2.1644,
+      "step": 25350
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.431095244633622e-05,
+      "loss": 2.2785,
+      "step": 25360
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.430866401208293e-05,
+      "loss": 2.2536,
+      "step": 25370
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.430637557782965e-05,
+      "loss": 2.23,
+      "step": 25380
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4304087143576366e-05,
+      "loss": 2.262,
+      "step": 25390
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.430179870932308e-05,
+      "loss": 2.1855,
+      "step": 25400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.42995102750698e-05,
+      "loss": 2.2378,
+      "step": 25410
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4297221840816516e-05,
+      "loss": 2.4013,
+      "step": 25420
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.429493340656323e-05,
+      "loss": 2.1159,
+      "step": 25430
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.429264497230995e-05,
+      "loss": 2.2293,
+      "step": 25440
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4290356538056665e-05,
+      "loss": 2.2844,
+      "step": 25450
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.428806810380338e-05,
+      "loss": 2.5541,
+      "step": 25460
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.42857796695501e-05,
+      "loss": 2.2027,
+      "step": 25470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4283491235296814e-05,
+      "loss": 2.3265,
+      "step": 25480
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.428120280104353e-05,
+      "loss": 2.1812,
+      "step": 25490
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.427891436679024e-05,
+      "loss": 2.2683,
+      "step": 25500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4276625932536956e-05,
+      "loss": 2.2711,
+      "step": 25510
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.427433749828368e-05,
+      "loss": 2.3269,
+      "step": 25520
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.427204906403039e-05,
+      "loss": 2.3256,
+      "step": 25530
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4269760629777106e-05,
+      "loss": 2.257,
+      "step": 25540
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4267472195523826e-05,
+      "loss": 2.2607,
+      "step": 25550
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.426518376127054e-05,
+      "loss": 2.2594,
+      "step": 25560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4262895327017255e-05,
+      "loss": 2.4334,
+      "step": 25570
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4260606892763976e-05,
+      "loss": 2.2299,
+      "step": 25580
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.425831845851069e-05,
+      "loss": 2.2076,
+      "step": 25590
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4256030024257404e-05,
+      "loss": 2.3395,
+      "step": 25600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4253741590004125e-05,
+      "loss": 2.1569,
+      "step": 25610
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.425145315575084e-05,
+      "loss": 2.3235,
+      "step": 25620
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.424916472149755e-05,
+      "loss": 2.2791,
+      "step": 25630
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4246876287244274e-05,
+      "loss": 2.2436,
+      "step": 25640
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.424458785299099e-05,
+      "loss": 2.2923,
+      "step": 25650
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.42422994187377e-05,
+      "loss": 2.2488,
+      "step": 25660
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.424001098448442e-05,
+      "loss": 2.3037,
+      "step": 25670
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.423772255023114e-05,
+      "loss": 2.2168,
+      "step": 25680
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4235434115977845e-05,
+      "loss": 2.0943,
+      "step": 25690
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4233145681724566e-05,
+      "loss": 2.2102,
+      "step": 25700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.423085724747128e-05,
+      "loss": 2.3707,
+      "step": 25710
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4228568813217994e-05,
+      "loss": 2.3201,
+      "step": 25720
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4226280378964715e-05,
+      "loss": 2.3987,
+      "step": 25730
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.422399194471143e-05,
+      "loss": 2.279,
+      "step": 25740
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.422170351045814e-05,
+      "loss": 2.2009,
+      "step": 25750
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4219415076204864e-05,
+      "loss": 2.3064,
+      "step": 25760
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.421712664195158e-05,
+      "loss": 2.1598,
+      "step": 25770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.421483820769829e-05,
+      "loss": 2.2582,
+      "step": 25780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.421254977344501e-05,
+      "loss": 2.1256,
+      "step": 25790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.421026133919173e-05,
+      "loss": 2.2932,
+      "step": 25800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.420797290493844e-05,
+      "loss": 2.3943,
+      "step": 25810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.420568447068516e-05,
+      "loss": 2.2726,
+      "step": 25820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4203396036431876e-05,
+      "loss": 2.2157,
+      "step": 25830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.420110760217859e-05,
+      "loss": 2.2888,
+      "step": 25840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.419881916792531e-05,
+      "loss": 2.4838,
+      "step": 25850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4196530733672026e-05,
+      "loss": 2.1625,
+      "step": 25860
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.419424229941874e-05,
+      "loss": 2.2363,
+      "step": 25870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.419195386516546e-05,
+      "loss": 2.4828,
+      "step": 25880
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.418966543091217e-05,
+      "loss": 2.4822,
+      "step": 25890
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.418737699665889e-05,
+      "loss": 2.302,
+      "step": 25900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.41850885624056e-05,
+      "loss": 2.1256,
+      "step": 25910
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.418280012815232e-05,
+      "loss": 2.1914,
+      "step": 25920
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.418051169389904e-05,
+      "loss": 2.079,
+      "step": 25930
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.417822325964575e-05,
+      "loss": 2.2081,
+      "step": 25940
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4175934825392466e-05,
+      "loss": 2.1153,
+      "step": 25950
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.417364639113919e-05,
+      "loss": 2.274,
+      "step": 25960
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.41713579568859e-05,
+      "loss": 2.2522,
+      "step": 25970
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4169069522632616e-05,
+      "loss": 2.2895,
+      "step": 25980
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4166781088379336e-05,
+      "loss": 2.4238,
+      "step": 25990
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.416449265412605e-05,
+      "loss": 2.1795,
+      "step": 26000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4162204219872765e-05,
+      "loss": 2.237,
+      "step": 26010
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4159915785619486e-05,
+      "loss": 2.1216,
+      "step": 26020
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.41576273513662e-05,
+      "loss": 2.1635,
+      "step": 26030
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4155338917112914e-05,
+      "loss": 2.122,
+      "step": 26040
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.415305048285963e-05,
+      "loss": 2.3703,
+      "step": 26050
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.415076204860635e-05,
+      "loss": 2.2444,
+      "step": 26060
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.414847361435306e-05,
+      "loss": 2.2513,
+      "step": 26070
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.414618518009978e-05,
+      "loss": 2.2434,
+      "step": 26080
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.414389674584649e-05,
+      "loss": 2.326,
+      "step": 26090
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4141608311593205e-05,
+      "loss": 2.1341,
+      "step": 26100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4139319877339926e-05,
+      "loss": 2.2537,
+      "step": 26110
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.413703144308664e-05,
+      "loss": 2.4402,
+      "step": 26120
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4134743008833355e-05,
+      "loss": 2.3245,
+      "step": 26130
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4132454574580076e-05,
+      "loss": 2.3338,
+      "step": 26140
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.413016614032679e-05,
+      "loss": 2.2362,
+      "step": 26150
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4127877706073504e-05,
+      "loss": 2.321,
+      "step": 26160
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4125589271820225e-05,
+      "loss": 2.208,
+      "step": 26170
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.412330083756694e-05,
+      "loss": 2.131,
+      "step": 26180
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.412101240331365e-05,
+      "loss": 2.1384,
+      "step": 26190
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4118723969060374e-05,
+      "loss": 2.3083,
+      "step": 26200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.411643553480709e-05,
+      "loss": 2.2815,
+      "step": 26210
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.41141471005538e-05,
+      "loss": 2.2491,
+      "step": 26220
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.411185866630052e-05,
+      "loss": 2.4467,
+      "step": 26230
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.410957023204724e-05,
+      "loss": 2.09,
+      "step": 26240
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.410728179779395e-05,
+      "loss": 2.202,
+      "step": 26250
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.410499336354067e-05,
+      "loss": 2.3052,
+      "step": 26260
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4102704929287386e-05,
+      "loss": 2.277,
+      "step": 26270
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.41004164950341e-05,
+      "loss": 2.2987,
+      "step": 26280
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4098128060780815e-05,
+      "loss": 2.1745,
+      "step": 26290
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.409583962652753e-05,
+      "loss": 2.3993,
+      "step": 26300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.409355119227424e-05,
+      "loss": 2.2328,
+      "step": 26310
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4091262758020964e-05,
+      "loss": 2.2647,
+      "step": 26320
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.408897432376768e-05,
+      "loss": 2.1995,
+      "step": 26330
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.408668588951439e-05,
+      "loss": 2.2455,
+      "step": 26340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.408439745526111e-05,
+      "loss": 2.3337,
+      "step": 26350
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.408210902100783e-05,
+      "loss": 2.2002,
+      "step": 26360
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.407982058675454e-05,
+      "loss": 2.2063,
+      "step": 26370
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.407753215250126e-05,
+      "loss": 2.3306,
+      "step": 26380
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4075243718247976e-05,
+      "loss": 2.1077,
+      "step": 26390
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.407295528399469e-05,
+      "loss": 2.3221,
+      "step": 26400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.407066684974141e-05,
+      "loss": 2.1849,
+      "step": 26410
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4068378415488126e-05,
+      "loss": 2.2153,
+      "step": 26420
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.406608998123484e-05,
+      "loss": 2.1272,
+      "step": 26430
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.406380154698156e-05,
+      "loss": 2.0843,
+      "step": 26440
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4061513112728275e-05,
+      "loss": 2.248,
+      "step": 26450
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.405922467847499e-05,
+      "loss": 2.4257,
+      "step": 26460
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.405693624422171e-05,
+      "loss": 2.2721,
+      "step": 26470
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.405464780996842e-05,
+      "loss": 2.1948,
+      "step": 26480
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.405235937571514e-05,
+      "loss": 2.2881,
+      "step": 26490
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.405007094146185e-05,
+      "loss": 2.3021,
+      "step": 26500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4047782507208566e-05,
+      "loss": 2.2531,
+      "step": 26510
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.404549407295529e-05,
+      "loss": 2.3317,
+      "step": 26520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4043205638702e-05,
+      "loss": 2.2281,
+      "step": 26530
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4040917204448715e-05,
+      "loss": 2.3392,
+      "step": 26540
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4038628770195436e-05,
+      "loss": 2.3488,
+      "step": 26550
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.403634033594215e-05,
+      "loss": 2.1875,
+      "step": 26560
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4034051901688865e-05,
+      "loss": 2.2004,
+      "step": 26570
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4031763467435586e-05,
+      "loss": 2.2766,
+      "step": 26580
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.40294750331823e-05,
+      "loss": 2.3332,
+      "step": 26590
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4027186598929014e-05,
+      "loss": 2.3131,
+      "step": 26600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4024898164675735e-05,
+      "loss": 2.1778,
+      "step": 26610
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.402260973042245e-05,
+      "loss": 2.4183,
+      "step": 26620
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.402032129616916e-05,
+      "loss": 2.2488,
+      "step": 26630
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4018032861915884e-05,
+      "loss": 2.129,
+      "step": 26640
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.40157444276626e-05,
+      "loss": 2.291,
+      "step": 26650
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.401345599340931e-05,
+      "loss": 2.258,
+      "step": 26660
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.401116755915603e-05,
+      "loss": 2.3021,
+      "step": 26670
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.400887912490274e-05,
+      "loss": 2.2252,
+      "step": 26680
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4006590690649455e-05,
+      "loss": 2.3878,
+      "step": 26690
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4004302256396175e-05,
+      "loss": 2.107,
+      "step": 26700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.400201382214289e-05,
+      "loss": 2.3986,
+      "step": 26710
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3999725387889604e-05,
+      "loss": 2.2451,
+      "step": 26720
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3997436953636325e-05,
+      "loss": 2.1133,
+      "step": 26730
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.399514851938304e-05,
+      "loss": 2.2691,
+      "step": 26740
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.399286008512975e-05,
+      "loss": 2.1614,
+      "step": 26750
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3990571650876474e-05,
+      "loss": 2.1668,
+      "step": 26760
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.398828321662319e-05,
+      "loss": 2.167,
+      "step": 26770
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.39859947823699e-05,
+      "loss": 2.3436,
+      "step": 26780
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.398370634811662e-05,
+      "loss": 2.1062,
+      "step": 26790
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.398141791386334e-05,
+      "loss": 2.1801,
+      "step": 26800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.397912947961005e-05,
+      "loss": 2.2421,
+      "step": 26810
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.397684104535677e-05,
+      "loss": 2.3968,
+      "step": 26820
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3974552611103486e-05,
+      "loss": 2.1781,
+      "step": 26830
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.39722641768502e-05,
+      "loss": 2.2151,
+      "step": 26840
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.396997574259692e-05,
+      "loss": 2.1584,
+      "step": 26850
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3967687308343635e-05,
+      "loss": 2.0782,
+      "step": 26860
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.396539887409035e-05,
+      "loss": 2.3362,
+      "step": 26870
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3963110439837064e-05,
+      "loss": 2.2685,
+      "step": 26880
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.396082200558378e-05,
+      "loss": 2.2134,
+      "step": 26890
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.39585335713305e-05,
+      "loss": 2.2323,
+      "step": 26900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.395624513707721e-05,
+      "loss": 2.125,
+      "step": 26910
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.395395670282393e-05,
+      "loss": 2.4238,
+      "step": 26920
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.395166826857065e-05,
+      "loss": 2.3486,
+      "step": 26930
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.394937983431736e-05,
+      "loss": 2.2849,
+      "step": 26940
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3947091400064076e-05,
+      "loss": 2.2468,
+      "step": 26950
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.39448029658108e-05,
+      "loss": 2.2602,
+      "step": 26960
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.394251453155751e-05,
+      "loss": 2.2907,
+      "step": 26970
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3940226097304225e-05,
+      "loss": 2.2525,
+      "step": 26980
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3937937663050946e-05,
+      "loss": 2.3345,
+      "step": 26990
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.393564922879766e-05,
+      "loss": 2.2611,
+      "step": 27000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3933360794544375e-05,
+      "loss": 2.2594,
+      "step": 27010
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3931072360291096e-05,
+      "loss": 2.1907,
+      "step": 27020
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.392878392603781e-05,
+      "loss": 2.2139,
+      "step": 27030
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3926495491784524e-05,
+      "loss": 2.2012,
+      "step": 27040
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3924207057531245e-05,
+      "loss": 2.1965,
+      "step": 27050
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.392191862327796e-05,
+      "loss": 2.3018,
+      "step": 27060
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.391963018902467e-05,
+      "loss": 2.1636,
+      "step": 27070
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.391734175477139e-05,
+      "loss": 2.136,
+      "step": 27080
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.39150533205181e-05,
+      "loss": 2.4316,
+      "step": 27090
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3912764886264815e-05,
+      "loss": 2.3352,
+      "step": 27100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3910476452011536e-05,
+      "loss": 2.2044,
+      "step": 27110
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.390818801775825e-05,
+      "loss": 2.113,
+      "step": 27120
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3905899583504965e-05,
+      "loss": 2.262,
+      "step": 27130
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3903611149251685e-05,
+      "loss": 2.2298,
+      "step": 27140
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.39013227149984e-05,
+      "loss": 2.2533,
+      "step": 27150
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3899034280745114e-05,
+      "loss": 2.4151,
+      "step": 27160
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3896745846491835e-05,
+      "loss": 2.229,
+      "step": 27170
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.389445741223855e-05,
+      "loss": 2.3317,
+      "step": 27180
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.389216897798526e-05,
+      "loss": 2.257,
+      "step": 27190
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3889880543731984e-05,
+      "loss": 2.2413,
+      "step": 27200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.38875921094787e-05,
+      "loss": 2.2385,
+      "step": 27210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.388530367522541e-05,
+      "loss": 2.5024,
+      "step": 27220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.388301524097213e-05,
+      "loss": 2.3374,
+      "step": 27230
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.388072680671885e-05,
+      "loss": 2.2077,
+      "step": 27240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.387843837246556e-05,
+      "loss": 2.2619,
+      "step": 27250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.387614993821228e-05,
+      "loss": 2.2026,
+      "step": 27260
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.387386150395899e-05,
+      "loss": 2.3982,
+      "step": 27270
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3871573069705704e-05,
+      "loss": 2.3381,
+      "step": 27280
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3869284635452425e-05,
+      "loss": 2.3795,
+      "step": 27290
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.386699620119914e-05,
+      "loss": 2.2448,
+      "step": 27300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.386470776694585e-05,
+      "loss": 2.3164,
+      "step": 27310
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3862419332692574e-05,
+      "loss": 2.197,
+      "step": 27320
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.386013089843929e-05,
+      "loss": 2.3321,
+      "step": 27330
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3857842464186e-05,
+      "loss": 2.147,
+      "step": 27340
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.385555402993272e-05,
+      "loss": 2.2972,
+      "step": 27350
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.385326559567944e-05,
+      "loss": 2.1299,
+      "step": 27360
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.385097716142615e-05,
+      "loss": 2.1349,
+      "step": 27370
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.384868872717287e-05,
+      "loss": 2.3583,
+      "step": 27380
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3846400292919586e-05,
+      "loss": 2.2612,
+      "step": 27390
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.38441118586663e-05,
+      "loss": 2.3364,
+      "step": 27400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.384182342441302e-05,
+      "loss": 2.1991,
+      "step": 27410
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3839534990159735e-05,
+      "loss": 2.3132,
+      "step": 27420
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.383724655590645e-05,
+      "loss": 2.2604,
+      "step": 27430
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.383495812165317e-05,
+      "loss": 2.1945,
+      "step": 27440
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3832669687399885e-05,
+      "loss": 2.1017,
+      "step": 27450
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.38303812531466e-05,
+      "loss": 2.199,
+      "step": 27460
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.382809281889331e-05,
+      "loss": 2.2346,
+      "step": 27470
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.382580438464003e-05,
+      "loss": 2.2823,
+      "step": 27480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.382351595038675e-05,
+      "loss": 2.1686,
+      "step": 27490
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.382122751613346e-05,
+      "loss": 2.3466,
+      "step": 27500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3818939081880176e-05,
+      "loss": 2.3026,
+      "step": 27510
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.38166506476269e-05,
+      "loss": 2.3596,
+      "step": 27520
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.381436221337361e-05,
+      "loss": 2.3329,
+      "step": 27530
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3812073779120325e-05,
+      "loss": 2.2289,
+      "step": 27540
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3809785344867046e-05,
+      "loss": 2.1802,
+      "step": 27550
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.380749691061376e-05,
+      "loss": 2.45,
+      "step": 27560
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3805208476360474e-05,
+      "loss": 2.0827,
+      "step": 27570
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3802920042107195e-05,
+      "loss": 2.2093,
+      "step": 27580
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.380063160785391e-05,
+      "loss": 2.4665,
+      "step": 27590
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3798343173600624e-05,
+      "loss": 2.1467,
+      "step": 27600
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3796054739347345e-05,
+      "loss": 2.149,
+      "step": 27610
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.379376630509406e-05,
+      "loss": 2.1548,
+      "step": 27620
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.379147787084077e-05,
+      "loss": 2.2776,
+      "step": 27630
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3789189436587494e-05,
+      "loss": 2.3913,
+      "step": 27640
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.378690100233421e-05,
+      "loss": 2.1884,
+      "step": 27650
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.378461256808092e-05,
+      "loss": 2.1749,
+      "step": 27660
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3782324133827636e-05,
+      "loss": 2.176,
+      "step": 27670
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.378003569957435e-05,
+      "loss": 2.4215,
+      "step": 27680
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3777747265321064e-05,
+      "loss": 2.2409,
+      "step": 27690
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3775458831067785e-05,
+      "loss": 2.2433,
+      "step": 27700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.37731703968145e-05,
+      "loss": 2.2805,
+      "step": 27710
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3770881962561214e-05,
+      "loss": 2.2067,
+      "step": 27720
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3768593528307935e-05,
+      "loss": 2.2702,
+      "step": 27730
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.376630509405465e-05,
+      "loss": 2.4218,
+      "step": 27740
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.376401665980136e-05,
+      "loss": 2.3797,
+      "step": 27750
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3761728225548084e-05,
+      "loss": 2.188,
+      "step": 27760
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.37594397912948e-05,
+      "loss": 2.3258,
+      "step": 27770
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.375715135704151e-05,
+      "loss": 2.2971,
+      "step": 27780
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.375486292278823e-05,
+      "loss": 2.1436,
+      "step": 27790
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.375257448853495e-05,
+      "loss": 2.3227,
+      "step": 27800
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.375028605428166e-05,
+      "loss": 2.2492,
+      "step": 27810
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.374799762002838e-05,
+      "loss": 2.2638,
+      "step": 27820
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3745709185775096e-05,
+      "loss": 2.3175,
+      "step": 27830
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.374342075152181e-05,
+      "loss": 2.2288,
+      "step": 27840
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.374113231726853e-05,
+      "loss": 2.333,
+      "step": 27850
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3738843883015245e-05,
+      "loss": 2.2562,
+      "step": 27860
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.373655544876196e-05,
+      "loss": 2.28,
+      "step": 27870
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3734267014508674e-05,
+      "loss": 2.2299,
+      "step": 27880
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.373197858025539e-05,
+      "loss": 2.0271,
+      "step": 27890
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.372969014600211e-05,
+      "loss": 2.1865,
+      "step": 27900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.372740171174882e-05,
+      "loss": 2.3976,
+      "step": 27910
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.372511327749554e-05,
+      "loss": 2.3263,
+      "step": 27920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.372282484324226e-05,
+      "loss": 2.114,
+      "step": 27930
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.372053640898897e-05,
+      "loss": 2.3048,
+      "step": 27940
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3718247974735686e-05,
+      "loss": 2.2542,
+      "step": 27950
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.371595954048241e-05,
+      "loss": 2.135,
+      "step": 27960
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.371367110622912e-05,
+      "loss": 2.2208,
+      "step": 27970
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3711382671975835e-05,
+      "loss": 2.266,
+      "step": 27980
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3709094237722556e-05,
+      "loss": 2.2769,
+      "step": 27990
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.370680580346927e-05,
+      "loss": 2.1441,
+      "step": 28000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3704517369215984e-05,
+      "loss": 2.2935,
+      "step": 28010
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3702228934962705e-05,
+      "loss": 2.401,
+      "step": 28020
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.369994050070942e-05,
+      "loss": 2.2079,
+      "step": 28030
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3697652066456134e-05,
+      "loss": 2.1913,
+      "step": 28040
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3695363632202855e-05,
+      "loss": 2.2576,
+      "step": 28050
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.369307519794956e-05,
+      "loss": 2.2689,
+      "step": 28060
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3690786763696276e-05,
+      "loss": 2.1698,
+      "step": 28070
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3688498329443e-05,
+      "loss": 2.1253,
+      "step": 28080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.368620989518971e-05,
+      "loss": 2.2868,
+      "step": 28090
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3683921460936425e-05,
+      "loss": 2.0501,
+      "step": 28100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3681633026683146e-05,
+      "loss": 2.3896,
+      "step": 28110
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.367934459242986e-05,
+      "loss": 2.1717,
+      "step": 28120
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3677056158176574e-05,
+      "loss": 2.2957,
+      "step": 28130
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3674767723923295e-05,
+      "loss": 2.2337,
+      "step": 28140
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.367247928967001e-05,
+      "loss": 2.3352,
+      "step": 28150
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3670190855416724e-05,
+      "loss": 2.2068,
+      "step": 28160
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3667902421163444e-05,
+      "loss": 2.3666,
+      "step": 28170
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.366561398691016e-05,
+      "loss": 2.306,
+      "step": 28180
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.366332555265687e-05,
+      "loss": 2.2115,
+      "step": 28190
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3661037118403594e-05,
+      "loss": 2.1288,
+      "step": 28200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.365874868415031e-05,
+      "loss": 2.0608,
+      "step": 28210
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.365646024989702e-05,
+      "loss": 2.3001,
+      "step": 28220
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.365417181564374e-05,
+      "loss": 2.3103,
+      "step": 28230
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.365188338139046e-05,
+      "loss": 2.239,
+      "step": 28240
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.364959494713717e-05,
+      "loss": 2.1624,
+      "step": 28250
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3647306512883885e-05,
+      "loss": 2.2734,
+      "step": 28260
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.36450180786306e-05,
+      "loss": 2.1851,
+      "step": 28270
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.364272964437732e-05,
+      "loss": 2.3625,
+      "step": 28280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3640441210124034e-05,
+      "loss": 2.3119,
+      "step": 28290
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.363815277587075e-05,
+      "loss": 2.3616,
+      "step": 28300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.363586434161747e-05,
+      "loss": 2.0218,
+      "step": 28310
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3633575907364184e-05,
+      "loss": 2.2025,
+      "step": 28320
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.36312874731109e-05,
+      "loss": 2.2558,
+      "step": 28330
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.362899903885762e-05,
+      "loss": 2.2127,
+      "step": 28340
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.362671060460433e-05,
+      "loss": 2.3427,
+      "step": 28350
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.362442217035105e-05,
+      "loss": 2.1714,
+      "step": 28360
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.362213373609777e-05,
+      "loss": 2.1931,
+      "step": 28370
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.361984530184448e-05,
+      "loss": 2.3745,
+      "step": 28380
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3617556867591196e-05,
+      "loss": 2.1743,
+      "step": 28390
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.361526843333791e-05,
+      "loss": 2.3113,
+      "step": 28400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.361297999908463e-05,
+      "loss": 2.3448,
+      "step": 28410
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3610691564831345e-05,
+      "loss": 2.1585,
+      "step": 28420
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.360840313057806e-05,
+      "loss": 2.1739,
+      "step": 28430
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.360611469632478e-05,
+      "loss": 2.2331,
+      "step": 28440
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3603826262071494e-05,
+      "loss": 2.2834,
+      "step": 28450
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.360153782781821e-05,
+      "loss": 2.2659,
+      "step": 28460
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.359924939356492e-05,
+      "loss": 2.173,
+      "step": 28470
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.359696095931164e-05,
+      "loss": 2.3407,
+      "step": 28480
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.359467252505836e-05,
+      "loss": 2.1362,
+      "step": 28490
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.359238409080507e-05,
+      "loss": 2.2518,
+      "step": 28500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3590095656551786e-05,
+      "loss": 2.2939,
+      "step": 28510
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.358780722229851e-05,
+      "loss": 2.243,
+      "step": 28520
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.358551878804522e-05,
+      "loss": 2.2809,
+      "step": 28530
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3583230353791935e-05,
+      "loss": 2.1699,
+      "step": 28540
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3580941919538656e-05,
+      "loss": 2.1813,
+      "step": 28550
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.357865348528537e-05,
+      "loss": 2.2395,
+      "step": 28560
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3576365051032084e-05,
+      "loss": 2.1388,
+      "step": 28570
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3574076616778805e-05,
+      "loss": 2.3555,
+      "step": 28580
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.357178818252552e-05,
+      "loss": 2.2796,
+      "step": 28590
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3569499748272234e-05,
+      "loss": 2.355,
+      "step": 28600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3567211314018954e-05,
+      "loss": 2.1346,
+      "step": 28610
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.356492287976567e-05,
+      "loss": 2.1793,
+      "step": 28620
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.356263444551238e-05,
+      "loss": 2.2621,
+      "step": 28630
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3560346011259104e-05,
+      "loss": 2.1665,
+      "step": 28640
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.355805757700582e-05,
+      "loss": 2.3427,
+      "step": 28650
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3555769142752525e-05,
+      "loss": 2.2908,
+      "step": 28660
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3553480708499246e-05,
+      "loss": 2.1994,
+      "step": 28670
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.355119227424596e-05,
+      "loss": 2.181,
+      "step": 28680
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3548903839992674e-05,
+      "loss": 2.2747,
+      "step": 28690
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3546615405739395e-05,
+      "loss": 2.2688,
+      "step": 28700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.354432697148611e-05,
+      "loss": 2.1805,
+      "step": 28710
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3542038537232823e-05,
+      "loss": 2.3188,
+      "step": 28720
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3539750102979544e-05,
+      "loss": 2.2117,
+      "step": 28730
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.353746166872626e-05,
+      "loss": 2.194,
+      "step": 28740
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.353517323447297e-05,
+      "loss": 2.222,
+      "step": 28750
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3532884800219694e-05,
+      "loss": 2.0096,
+      "step": 28760
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.353059636596641e-05,
+      "loss": 2.2394,
+      "step": 28770
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.352830793171312e-05,
+      "loss": 2.2936,
+      "step": 28780
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.352601949745984e-05,
+      "loss": 2.2695,
+      "step": 28790
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.352373106320656e-05,
+      "loss": 2.2188,
+      "step": 28800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.352144262895327e-05,
+      "loss": 2.1946,
+      "step": 28810
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.351915419469999e-05,
+      "loss": 2.2908,
+      "step": 28820
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3516865760446706e-05,
+      "loss": 2.197,
+      "step": 28830
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.351457732619342e-05,
+      "loss": 2.385,
+      "step": 28840
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3512288891940134e-05,
+      "loss": 2.284,
+      "step": 28850
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.351000045768685e-05,
+      "loss": 2.1954,
+      "step": 28860
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.350771202343357e-05,
+      "loss": 2.1071,
+      "step": 28870
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3505423589180283e-05,
+      "loss": 2.2875,
+      "step": 28880
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3503135154927e-05,
+      "loss": 2.2441,
+      "step": 28890
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.350084672067372e-05,
+      "loss": 2.1706,
+      "step": 28900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.349855828642043e-05,
+      "loss": 2.3903,
+      "step": 28910
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.349626985216715e-05,
+      "loss": 2.2887,
+      "step": 28920
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.349398141791387e-05,
+      "loss": 2.2433,
+      "step": 28930
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.349169298366058e-05,
+      "loss": 2.3196,
+      "step": 28940
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3489404549407296e-05,
+      "loss": 2.2339,
+      "step": 28950
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.348711611515402e-05,
+      "loss": 2.179,
+      "step": 28960
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.348482768090073e-05,
+      "loss": 2.2136,
+      "step": 28970
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3482539246647445e-05,
+      "loss": 2.2694,
+      "step": 28980
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3480250812394166e-05,
+      "loss": 2.2348,
+      "step": 28990
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.347796237814088e-05,
+      "loss": 2.3096,
+      "step": 29000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3475673943887594e-05,
+      "loss": 2.2403,
+      "step": 29010
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3473385509634315e-05,
+      "loss": 2.2729,
+      "step": 29020
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.347109707538103e-05,
+      "loss": 2.1817,
+      "step": 29030
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3468808641127744e-05,
+      "loss": 2.182,
+      "step": 29040
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.346652020687446e-05,
+      "loss": 2.2399,
+      "step": 29050
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.346423177262117e-05,
+      "loss": 2.1898,
+      "step": 29060
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3461943338367886e-05,
+      "loss": 2.3003,
+      "step": 29070
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.345965490411461e-05,
+      "loss": 2.3884,
+      "step": 29080
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.345736646986132e-05,
+      "loss": 2.1533,
+      "step": 29090
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3455078035608035e-05,
+      "loss": 2.2165,
+      "step": 29100
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3452789601354756e-05,
+      "loss": 2.0636,
+      "step": 29110
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.345050116710147e-05,
+      "loss": 2.2275,
+      "step": 29120
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3448212732848184e-05,
+      "loss": 2.3564,
+      "step": 29130
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3445924298594905e-05,
+      "loss": 2.2038,
+      "step": 29140
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.344363586434162e-05,
+      "loss": 2.1875,
+      "step": 29150
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3441347430088333e-05,
+      "loss": 2.4906,
+      "step": 29160
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3439058995835054e-05,
+      "loss": 2.1997,
+      "step": 29170
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.343677056158177e-05,
+      "loss": 2.3372,
+      "step": 29180
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.343448212732848e-05,
+      "loss": 2.2557,
+      "step": 29190
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3432193693075204e-05,
+      "loss": 2.3428,
+      "step": 29200
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.342990525882192e-05,
+      "loss": 2.0956,
+      "step": 29210
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.342761682456863e-05,
+      "loss": 2.347,
+      "step": 29220
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.342532839031535e-05,
+      "loss": 2.3811,
+      "step": 29230
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.342303995606207e-05,
+      "loss": 2.2781,
+      "step": 29240
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.342075152180878e-05,
+      "loss": 2.2374,
+      "step": 29250
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3418463087555495e-05,
+      "loss": 2.1602,
+      "step": 29260
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.341617465330221e-05,
+      "loss": 2.1937,
+      "step": 29270
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.341388621904893e-05,
+      "loss": 2.3525,
+      "step": 29280
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3411597784795644e-05,
+      "loss": 2.1304,
+      "step": 29290
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.340930935054236e-05,
+      "loss": 2.0873,
+      "step": 29300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.340702091628908e-05,
+      "loss": 2.2035,
+      "step": 29310
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3404732482035793e-05,
+      "loss": 2.3351,
+      "step": 29320
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.340244404778251e-05,
+      "loss": 2.1116,
+      "step": 29330
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.340015561352923e-05,
+      "loss": 2.2325,
+      "step": 29340
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.339786717927594e-05,
+      "loss": 2.2103,
+      "step": 29350
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.339557874502266e-05,
+      "loss": 2.2155,
+      "step": 29360
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.339329031076938e-05,
+      "loss": 2.2223,
+      "step": 29370
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.339100187651609e-05,
+      "loss": 2.2957,
+      "step": 29380
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3388713442262806e-05,
+      "loss": 2.1328,
+      "step": 29390
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.338642500800953e-05,
+      "loss": 2.3111,
+      "step": 29400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.338413657375624e-05,
+      "loss": 2.2328,
+      "step": 29410
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3381848139502955e-05,
+      "loss": 2.3519,
+      "step": 29420
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3379559705249676e-05,
+      "loss": 2.2333,
+      "step": 29430
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.337727127099639e-05,
+      "loss": 2.3311,
+      "step": 29440
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.33749828367431e-05,
+      "loss": 2.1258,
+      "step": 29450
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.337269440248982e-05,
+      "loss": 2.1338,
+      "step": 29460
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.337040596823653e-05,
+      "loss": 2.1954,
+      "step": 29470
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.336811753398325e-05,
+      "loss": 2.5336,
+      "step": 29480
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.336582909972997e-05,
+      "loss": 2.3028,
+      "step": 29490
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.336354066547668e-05,
+      "loss": 2.3005,
+      "step": 29500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3361252231223396e-05,
+      "loss": 2.2478,
+      "step": 29510
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.335896379697012e-05,
+      "loss": 2.4291,
+      "step": 29520
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.335667536271683e-05,
+      "loss": 2.392,
+      "step": 29530
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3354386928463545e-05,
+      "loss": 2.1811,
+      "step": 29540
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3352098494210266e-05,
+      "loss": 2.1566,
+      "step": 29550
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.334981005995698e-05,
+      "loss": 2.3832,
+      "step": 29560
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3347521625703694e-05,
+      "loss": 2.2504,
+      "step": 29570
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3345233191450415e-05,
+      "loss": 2.4139,
+      "step": 29580
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.334294475719713e-05,
+      "loss": 2.1897,
+      "step": 29590
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3340656322943843e-05,
+      "loss": 2.2243,
+      "step": 29600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3338367888690564e-05,
+      "loss": 2.1772,
+      "step": 29610
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.333607945443728e-05,
+      "loss": 2.2222,
+      "step": 29620
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.333379102018399e-05,
+      "loss": 2.1948,
+      "step": 29630
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.333150258593071e-05,
+      "loss": 2.1693,
+      "step": 29640
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.332921415167742e-05,
+      "loss": 2.162,
+      "step": 29650
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3326925717424135e-05,
+      "loss": 2.2364,
+      "step": 29660
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3324637283170856e-05,
+      "loss": 2.3002,
+      "step": 29670
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.332234884891757e-05,
+      "loss": 2.2576,
+      "step": 29680
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3320060414664284e-05,
+      "loss": 2.2445,
+      "step": 29690
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3317771980411005e-05,
+      "loss": 2.1332,
+      "step": 29700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.331548354615772e-05,
+      "loss": 2.1975,
+      "step": 29710
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.331319511190443e-05,
+      "loss": 2.2195,
+      "step": 29720
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3310906677651154e-05,
+      "loss": 2.2482,
+      "step": 29730
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.330861824339787e-05,
+      "loss": 2.1857,
+      "step": 29740
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.330632980914458e-05,
+      "loss": 2.2443,
+      "step": 29750
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3304041374891303e-05,
+      "loss": 2.1629,
+      "step": 29760
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.330175294063802e-05,
+      "loss": 2.196,
+      "step": 29770
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.329946450638473e-05,
+      "loss": 2.174,
+      "step": 29780
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.329717607213145e-05,
+      "loss": 2.3651,
+      "step": 29790
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.329488763787817e-05,
+      "loss": 2.2519,
+      "step": 29800
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.329259920362488e-05,
+      "loss": 2.2075,
+      "step": 29810
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.32903107693716e-05,
+      "loss": 2.2942,
+      "step": 29820
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3288022335118316e-05,
+      "loss": 2.3036,
+      "step": 29830
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.328573390086503e-05,
+      "loss": 2.184,
+      "step": 29840
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3283445466611744e-05,
+      "loss": 2.0728,
+      "step": 29850
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.328115703235846e-05,
+      "loss": 2.0687,
+      "step": 29860
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.327886859810518e-05,
+      "loss": 2.3478,
+      "step": 29870
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.327658016385189e-05,
+      "loss": 2.0224,
+      "step": 29880
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.327429172959861e-05,
+      "loss": 2.1782,
+      "step": 29890
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.327200329534533e-05,
+      "loss": 2.2575,
+      "step": 29900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.326971486109204e-05,
+      "loss": 2.1917,
+      "step": 29910
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.326742642683876e-05,
+      "loss": 2.2868,
+      "step": 29920
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.326513799258548e-05,
+      "loss": 2.1408,
+      "step": 29930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.326284955833219e-05,
+      "loss": 2.0978,
+      "step": 29940
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3260561124078906e-05,
+      "loss": 2.102,
+      "step": 29950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.325827268982563e-05,
+      "loss": 2.0969,
+      "step": 29960
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.325598425557234e-05,
+      "loss": 2.1811,
+      "step": 29970
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3253695821319055e-05,
+      "loss": 2.1494,
+      "step": 29980
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3251407387065776e-05,
+      "loss": 2.1645,
+      "step": 29990
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.324911895281249e-05,
+      "loss": 2.4111,
+      "step": 30000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3246830518559204e-05,
+      "loss": 2.1847,
+      "step": 30010
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3244542084305925e-05,
+      "loss": 2.154,
+      "step": 30020
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.324225365005264e-05,
+      "loss": 2.1531,
+      "step": 30030
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3239965215799347e-05,
+      "loss": 2.2329,
+      "step": 30040
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.323767678154607e-05,
+      "loss": 2.3304,
+      "step": 30050
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.323538834729278e-05,
+      "loss": 2.2206,
+      "step": 30060
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3233099913039496e-05,
+      "loss": 2.0979,
+      "step": 30070
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.323081147878622e-05,
+      "loss": 2.2306,
+      "step": 30080
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.322852304453293e-05,
+      "loss": 2.2456,
+      "step": 30090
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3226234610279645e-05,
+      "loss": 2.248,
+      "step": 30100
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3223946176026366e-05,
+      "loss": 2.3657,
+      "step": 30110
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.322165774177308e-05,
+      "loss": 2.3088,
+      "step": 30120
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3219369307519794e-05,
+      "loss": 2.0992,
+      "step": 30130
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3217080873266515e-05,
+      "loss": 2.1818,
+      "step": 30140
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.321479243901323e-05,
+      "loss": 2.174,
+      "step": 30150
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.321250400475994e-05,
+      "loss": 2.2604,
+      "step": 30160
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3210215570506664e-05,
+      "loss": 2.0914,
+      "step": 30170
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.320792713625338e-05,
+      "loss": 2.1598,
+      "step": 30180
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.320563870200009e-05,
+      "loss": 2.2641,
+      "step": 30190
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3203350267746813e-05,
+      "loss": 2.2908,
+      "step": 30200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.320106183349353e-05,
+      "loss": 2.1448,
+      "step": 30210
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.319877339924024e-05,
+      "loss": 2.205,
+      "step": 30220
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.319648496498696e-05,
+      "loss": 2.3599,
+      "step": 30230
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.319419653073367e-05,
+      "loss": 2.1949,
+      "step": 30240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.319190809648039e-05,
+      "loss": 2.2394,
+      "step": 30250
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3189619662227105e-05,
+      "loss": 2.227,
+      "step": 30260
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.318733122797382e-05,
+      "loss": 2.2041,
+      "step": 30270
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.318504279372054e-05,
+      "loss": 2.2031,
+      "step": 30280
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3182754359467254e-05,
+      "loss": 2.2333,
+      "step": 30290
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.318046592521397e-05,
+      "loss": 2.2222,
+      "step": 30300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.317817749096069e-05,
+      "loss": 2.2026,
+      "step": 30310
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.31758890567074e-05,
+      "loss": 2.2933,
+      "step": 30320
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.317360062245412e-05,
+      "loss": 2.2006,
+      "step": 30330
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.317131218820084e-05,
+      "loss": 2.212,
+      "step": 30340
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.316902375394755e-05,
+      "loss": 2.1924,
+      "step": 30350
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.316673531969427e-05,
+      "loss": 2.1161,
+      "step": 30360
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.316444688544099e-05,
+      "loss": 2.2663,
+      "step": 30370
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.31621584511877e-05,
+      "loss": 2.1717,
+      "step": 30380
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3159870016934416e-05,
+      "loss": 2.3292,
+      "step": 30390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.315758158268114e-05,
+      "loss": 2.1534,
+      "step": 30400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.315529314842785e-05,
+      "loss": 2.1418,
+      "step": 30410
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3153004714174565e-05,
+      "loss": 2.2721,
+      "step": 30420
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.315071627992128e-05,
+      "loss": 2.1498,
+      "step": 30430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.314842784566799e-05,
+      "loss": 2.2679,
+      "step": 30440
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.314613941141471e-05,
+      "loss": 2.1634,
+      "step": 30450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.314385097716143e-05,
+      "loss": 2.0886,
+      "step": 30460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.314156254290814e-05,
+      "loss": 1.9656,
+      "step": 30470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3139274108654857e-05,
+      "loss": 2.1539,
+      "step": 30480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.313698567440158e-05,
+      "loss": 2.4574,
+      "step": 30490
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.313469724014829e-05,
+      "loss": 2.3302,
+      "step": 30500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3132408805895006e-05,
+      "loss": 2.1459,
+      "step": 30510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.313012037164173e-05,
+      "loss": 2.1973,
+      "step": 30520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.312783193738844e-05,
+      "loss": 2.088,
+      "step": 30530
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3125543503135155e-05,
+      "loss": 2.2457,
+      "step": 30540
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3123255068881876e-05,
+      "loss": 2.1393,
+      "step": 30550
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.312096663462859e-05,
+      "loss": 2.2307,
+      "step": 30560
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3118678200375304e-05,
+      "loss": 2.3452,
+      "step": 30570
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3116389766122025e-05,
+      "loss": 2.3315,
+      "step": 30580
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.311410133186874e-05,
+      "loss": 2.3237,
+      "step": 30590
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.311181289761545e-05,
+      "loss": 2.3163,
+      "step": 30600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3109524463362174e-05,
+      "loss": 2.1789,
+      "step": 30610
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.310723602910889e-05,
+      "loss": 2.214,
+      "step": 30620
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.31049475948556e-05,
+      "loss": 2.2299,
+      "step": 30630
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3102659160602317e-05,
+      "loss": 2.218,
+      "step": 30640
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.310037072634903e-05,
+      "loss": 2.2392,
+      "step": 30650
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.309808229209575e-05,
+      "loss": 2.1551,
+      "step": 30660
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3095793857842466e-05,
+      "loss": 2.0402,
+      "step": 30670
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.309350542358918e-05,
+      "loss": 2.219,
+      "step": 30680
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.30912169893359e-05,
+      "loss": 2.3105,
+      "step": 30690
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3088928555082615e-05,
+      "loss": 2.3078,
+      "step": 30700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.308664012082933e-05,
+      "loss": 2.2992,
+      "step": 30710
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.308435168657604e-05,
+      "loss": 2.0893,
+      "step": 30720
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3082063252322764e-05,
+      "loss": 2.1845,
+      "step": 30730
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.307977481806948e-05,
+      "loss": 2.3656,
+      "step": 30740
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.307748638381619e-05,
+      "loss": 2.1835,
+      "step": 30750
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.307519794956291e-05,
+      "loss": 2.1121,
+      "step": 30760
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.307290951530963e-05,
+      "loss": 2.2853,
+      "step": 30770
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.307062108105634e-05,
+      "loss": 2.2598,
+      "step": 30780
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.306833264680306e-05,
+      "loss": 2.2574,
+      "step": 30790
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3066044212549777e-05,
+      "loss": 2.143,
+      "step": 30800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.306375577829649e-05,
+      "loss": 2.1987,
+      "step": 30810
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.306146734404321e-05,
+      "loss": 2.2918,
+      "step": 30820
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.305917890978992e-05,
+      "loss": 2.1506,
+      "step": 30830
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.305689047553664e-05,
+      "loss": 2.3496,
+      "step": 30840
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3054602041283354e-05,
+      "loss": 2.2199,
+      "step": 30850
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.305231360703007e-05,
+      "loss": 2.1885,
+      "step": 30860
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.305002517277679e-05,
+      "loss": 2.1256,
+      "step": 30870
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.30477367385235e-05,
+      "loss": 2.2452,
+      "step": 30880
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.304544830427022e-05,
+      "loss": 2.198,
+      "step": 30890
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.304315987001694e-05,
+      "loss": 2.1818,
+      "step": 30900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.304087143576365e-05,
+      "loss": 2.3153,
+      "step": 30910
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3038583001510367e-05,
+      "loss": 2.311,
+      "step": 30920
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.303629456725709e-05,
+      "loss": 2.1966,
+      "step": 30930
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.30340061330038e-05,
+      "loss": 2.289,
+      "step": 30940
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3031717698750516e-05,
+      "loss": 2.2648,
+      "step": 30950
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.302942926449724e-05,
+      "loss": 2.1904,
+      "step": 30960
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.302714083024395e-05,
+      "loss": 2.3601,
+      "step": 30970
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3024852395990665e-05,
+      "loss": 2.2747,
+      "step": 30980
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.3022563961737386e-05,
+      "loss": 2.2631,
+      "step": 30990
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.30202755274841e-05,
+      "loss": 2.2069,
+      "step": 31000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.3017987093230814e-05,
+      "loss": 2.3265,
+      "step": 31010
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.3015698658977535e-05,
+      "loss": 2.4389,
+      "step": 31020
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.301341022472424e-05,
+      "loss": 2.2436,
+      "step": 31030
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.3011121790470956e-05,
+      "loss": 2.1312,
+      "step": 31040
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.300883335621768e-05,
+      "loss": 2.3308,
+      "step": 31050
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.300654492196439e-05,
+      "loss": 2.114,
+      "step": 31060
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.3004256487711106e-05,
+      "loss": 2.1551,
+      "step": 31070
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.3001968053457827e-05,
+      "loss": 2.1645,
+      "step": 31080
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.299967961920454e-05,
+      "loss": 2.3614,
+      "step": 31090
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2997391184951255e-05,
+      "loss": 2.2709,
+      "step": 31100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2995102750697976e-05,
+      "loss": 2.3112,
+      "step": 31110
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.299281431644469e-05,
+      "loss": 2.4847,
+      "step": 31120
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2990525882191404e-05,
+      "loss": 2.1995,
+      "step": 31130
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2988237447938125e-05,
+      "loss": 2.3218,
+      "step": 31140
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.298594901368484e-05,
+      "loss": 1.9954,
+      "step": 31150
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.298366057943155e-05,
+      "loss": 2.1424,
+      "step": 31160
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2981372145178274e-05,
+      "loss": 2.0122,
+      "step": 31170
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.297908371092499e-05,
+      "loss": 2.2239,
+      "step": 31180
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.29767952766717e-05,
+      "loss": 2.2821,
+      "step": 31190
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.297450684241842e-05,
+      "loss": 2.274,
+      "step": 31200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.297221840816514e-05,
+      "loss": 2.2991,
+      "step": 31210
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.296992997391185e-05,
+      "loss": 2.1071,
+      "step": 31220
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2967641539658566e-05,
+      "loss": 2.3295,
+      "step": 31230
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.296535310540528e-05,
+      "loss": 2.2285,
+      "step": 31240
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2963064671152e-05,
+      "loss": 1.9909,
+      "step": 31250
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2960776236898715e-05,
+      "loss": 2.1087,
+      "step": 31260
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.295848780264543e-05,
+      "loss": 2.1625,
+      "step": 31270
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.295619936839215e-05,
+      "loss": 2.1927,
+      "step": 31280
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2953910934138864e-05,
+      "loss": 2.1889,
+      "step": 31290
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.295162249988558e-05,
+      "loss": 2.2267,
+      "step": 31300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.29493340656323e-05,
+      "loss": 2.2689,
+      "step": 31310
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.294704563137901e-05,
+      "loss": 2.2412,
+      "step": 31320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.294475719712573e-05,
+      "loss": 2.4487,
+      "step": 31330
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.294246876287245e-05,
+      "loss": 2.3758,
+      "step": 31340
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.294018032861916e-05,
+      "loss": 2.1497,
+      "step": 31350
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2937891894365876e-05,
+      "loss": 2.1315,
+      "step": 31360
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.29356034601126e-05,
+      "loss": 2.3703,
+      "step": 31370
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.293331502585931e-05,
+      "loss": 2.2968,
+      "step": 31380
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2931026591606026e-05,
+      "loss": 2.3343,
+      "step": 31390
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2928738157352747e-05,
+      "loss": 2.2466,
+      "step": 31400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.292644972309946e-05,
+      "loss": 2.387,
+      "step": 31410
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.292416128884617e-05,
+      "loss": 2.2064,
+      "step": 31420
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.292187285459289e-05,
+      "loss": 2.2675,
+      "step": 31430
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.29195844203396e-05,
+      "loss": 2.1657,
+      "step": 31440
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.291729598608632e-05,
+      "loss": 2.1825,
+      "step": 31450
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.291500755183304e-05,
+      "loss": 2.33,
+      "step": 31460
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.291271911757975e-05,
+      "loss": 2.2375,
+      "step": 31470
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2910430683326466e-05,
+      "loss": 2.3336,
+      "step": 31480
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.290814224907319e-05,
+      "loss": 2.1744,
+      "step": 31490
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.29058538148199e-05,
+      "loss": 2.1692,
+      "step": 31500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2903565380566616e-05,
+      "loss": 2.2882,
+      "step": 31510
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2901276946313337e-05,
+      "loss": 2.0783,
+      "step": 31520
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.289898851206005e-05,
+      "loss": 2.2574,
+      "step": 31530
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2896700077806765e-05,
+      "loss": 2.2084,
+      "step": 31540
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2894411643553486e-05,
+      "loss": 2.2735,
+      "step": 31550
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.28921232093002e-05,
+      "loss": 2.1513,
+      "step": 31560
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2889834775046914e-05,
+      "loss": 2.118,
+      "step": 31570
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2887546340793635e-05,
+      "loss": 2.2275,
+      "step": 31580
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.288525790654035e-05,
+      "loss": 2.0872,
+      "step": 31590
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.288296947228706e-05,
+      "loss": 2.2174,
+      "step": 31600
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2880681038033784e-05,
+      "loss": 2.1212,
+      "step": 31610
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.287839260378049e-05,
+      "loss": 2.3599,
+      "step": 31620
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.287610416952721e-05,
+      "loss": 2.2492,
+      "step": 31630
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2873815735273926e-05,
+      "loss": 2.2715,
+      "step": 31640
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.287152730102064e-05,
+      "loss": 2.2337,
+      "step": 31650
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.286923886676736e-05,
+      "loss": 2.173,
+      "step": 31660
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2866950432514076e-05,
+      "loss": 2.2317,
+      "step": 31670
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.286466199826079e-05,
+      "loss": 2.2065,
+      "step": 31680
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.286237356400751e-05,
+      "loss": 2.3149,
+      "step": 31690
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2860085129754225e-05,
+      "loss": 2.3307,
+      "step": 31700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.285779669550094e-05,
+      "loss": 2.31,
+      "step": 31710
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.285550826124766e-05,
+      "loss": 2.145,
+      "step": 31720
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2853219826994374e-05,
+      "loss": 2.2352,
+      "step": 31730
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.285093139274109e-05,
+      "loss": 2.2666,
+      "step": 31740
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.284864295848781e-05,
+      "loss": 2.1685,
+      "step": 31750
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.284635452423452e-05,
+      "loss": 2.1901,
+      "step": 31760
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.284406608998124e-05,
+      "loss": 2.2585,
+      "step": 31770
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.284177765572796e-05,
+      "loss": 2.0612,
+      "step": 31780
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.283948922147467e-05,
+      "loss": 2.1315,
+      "step": 31790
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2837200787221386e-05,
+      "loss": 2.166,
+      "step": 31800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.283491235296811e-05,
+      "loss": 2.2068,
+      "step": 31810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2832623918714815e-05,
+      "loss": 2.3342,
+      "step": 31820
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.283033548446153e-05,
+      "loss": 2.1124,
+      "step": 31830
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.282804705020825e-05,
+      "loss": 2.3784,
+      "step": 31840
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2825758615954964e-05,
+      "loss": 2.3891,
+      "step": 31850
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.282347018170168e-05,
+      "loss": 2.11,
+      "step": 31860
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.28211817474484e-05,
+      "loss": 2.2112,
+      "step": 31870
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.281889331319511e-05,
+      "loss": 2.2235,
+      "step": 31880
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.281660487894183e-05,
+      "loss": 2.289,
+      "step": 31890
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.281431644468855e-05,
+      "loss": 2.3127,
+      "step": 31900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.281202801043526e-05,
+      "loss": 2.2106,
+      "step": 31910
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2809739576181976e-05,
+      "loss": 2.1248,
+      "step": 31920
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.28074511419287e-05,
+      "loss": 2.3596,
+      "step": 31930
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.280516270767541e-05,
+      "loss": 2.3839,
+      "step": 31940
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2802874273422126e-05,
+      "loss": 2.2054,
+      "step": 31950
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2800585839168846e-05,
+      "loss": 2.1694,
+      "step": 31960
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.279829740491556e-05,
+      "loss": 2.4445,
+      "step": 31970
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2796008970662275e-05,
+      "loss": 2.2234,
+      "step": 31980
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2793720536408996e-05,
+      "loss": 2.2469,
+      "step": 31990
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.279143210215571e-05,
+      "loss": 2.2219,
+      "step": 32000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.278914366790242e-05,
+      "loss": 2.0641,
+      "step": 32010
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.278685523364914e-05,
+      "loss": 2.2913,
+      "step": 32020
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.278456679939585e-05,
+      "loss": 2.2551,
+      "step": 32030
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2782278365142566e-05,
+      "loss": 2.1322,
+      "step": 32040
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.277998993088929e-05,
+      "loss": 2.2193,
+      "step": 32050
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2777701496636e-05,
+      "loss": 2.2741,
+      "step": 32060
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2775413062382716e-05,
+      "loss": 2.1849,
+      "step": 32070
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2773124628129436e-05,
+      "loss": 2.2053,
+      "step": 32080
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.277083619387615e-05,
+      "loss": 2.1136,
+      "step": 32090
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2768547759622865e-05,
+      "loss": 2.3134,
+      "step": 32100
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2766259325369586e-05,
+      "loss": 2.2523,
+      "step": 32110
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.27639708911163e-05,
+      "loss": 2.2155,
+      "step": 32120
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2761682456863014e-05,
+      "loss": 2.2054,
+      "step": 32130
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2759394022609735e-05,
+      "loss": 2.3381,
+      "step": 32140
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.275710558835645e-05,
+      "loss": 2.2805,
+      "step": 32150
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.275481715410316e-05,
+      "loss": 2.1963,
+      "step": 32160
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2752528719849884e-05,
+      "loss": 2.2045,
+      "step": 32170
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.27502402855966e-05,
+      "loss": 2.1517,
+      "step": 32180
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.274795185134331e-05,
+      "loss": 2.1496,
+      "step": 32190
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.274566341709003e-05,
+      "loss": 2.3078,
+      "step": 32200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.274337498283674e-05,
+      "loss": 2.0796,
+      "step": 32210
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.274108654858346e-05,
+      "loss": 2.3359,
+      "step": 32220
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2738798114330176e-05,
+      "loss": 2.1233,
+      "step": 32230
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.273650968007689e-05,
+      "loss": 2.2771,
+      "step": 32240
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.273422124582361e-05,
+      "loss": 2.1216,
+      "step": 32250
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2731932811570325e-05,
+      "loss": 2.2559,
+      "step": 32260
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.272964437731704e-05,
+      "loss": 2.2291,
+      "step": 32270
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.272735594306376e-05,
+      "loss": 2.1884,
+      "step": 32280
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2725067508810474e-05,
+      "loss": 2.0338,
+      "step": 32290
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.272277907455719e-05,
+      "loss": 2.162,
+      "step": 32300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.272049064030391e-05,
+      "loss": 2.153,
+      "step": 32310
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.271820220605062e-05,
+      "loss": 2.2952,
+      "step": 32320
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.271591377179734e-05,
+      "loss": 2.3006,
+      "step": 32330
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.271362533754406e-05,
+      "loss": 2.1166,
+      "step": 32340
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.271133690329077e-05,
+      "loss": 2.2669,
+      "step": 32350
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2709048469037486e-05,
+      "loss": 2.1376,
+      "step": 32360
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.270676003478421e-05,
+      "loss": 2.1991,
+      "step": 32370
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.270447160053092e-05,
+      "loss": 2.0621,
+      "step": 32380
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2702183166277636e-05,
+      "loss": 2.4651,
+      "step": 32390
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2699894732024356e-05,
+      "loss": 2.3465,
+      "step": 32400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2697606297771064e-05,
+      "loss": 2.3222,
+      "step": 32410
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.269531786351778e-05,
+      "loss": 2.0481,
+      "step": 32420
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.26930294292645e-05,
+      "loss": 2.2245,
+      "step": 32430
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.269074099501121e-05,
+      "loss": 2.1254,
+      "step": 32440
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.268845256075793e-05,
+      "loss": 2.2384,
+      "step": 32450
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.268616412650465e-05,
+      "loss": 2.1082,
+      "step": 32460
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.268387569225136e-05,
+      "loss": 2.1515,
+      "step": 32470
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2681587257998076e-05,
+      "loss": 2.1837,
+      "step": 32480
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.26792988237448e-05,
+      "loss": 2.254,
+      "step": 32490
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.267701038949151e-05,
+      "loss": 2.3806,
+      "step": 32500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2674721955238225e-05,
+      "loss": 2.1795,
+      "step": 32510
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2672433520984946e-05,
+      "loss": 2.1577,
+      "step": 32520
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.267014508673166e-05,
+      "loss": 2.1453,
+      "step": 32530
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2667856652478375e-05,
+      "loss": 2.1211,
+      "step": 32540
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2665568218225096e-05,
+      "loss": 2.3648,
+      "step": 32550
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.266327978397181e-05,
+      "loss": 2.2016,
+      "step": 32560
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2660991349718524e-05,
+      "loss": 2.2246,
+      "step": 32570
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2658702915465245e-05,
+      "loss": 2.2008,
+      "step": 32580
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.265641448121196e-05,
+      "loss": 2.0195,
+      "step": 32590
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.265412604695867e-05,
+      "loss": 2.3262,
+      "step": 32600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.265183761270539e-05,
+      "loss": 2.2448,
+      "step": 32610
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.26495491784521e-05,
+      "loss": 2.1386,
+      "step": 32620
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.264726074419882e-05,
+      "loss": 2.1245,
+      "step": 32630
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2644972309945536e-05,
+      "loss": 2.3334,
+      "step": 32640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.264268387569225e-05,
+      "loss": 2.2253,
+      "step": 32650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.264039544143897e-05,
+      "loss": 2.2033,
+      "step": 32660
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2638107007185685e-05,
+      "loss": 2.2252,
+      "step": 32670
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.26358185729324e-05,
+      "loss": 2.2026,
+      "step": 32680
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.263353013867912e-05,
+      "loss": 2.0728,
+      "step": 32690
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2631241704425835e-05,
+      "loss": 2.2078,
+      "step": 32700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.262895327017255e-05,
+      "loss": 2.3617,
+      "step": 32710
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.262666483591927e-05,
+      "loss": 2.1821,
+      "step": 32720
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2624376401665984e-05,
+      "loss": 2.247,
+      "step": 32730
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.26220879674127e-05,
+      "loss": 2.1951,
+      "step": 32740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.261979953315942e-05,
+      "loss": 2.1331,
+      "step": 32750
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.261751109890613e-05,
+      "loss": 2.1596,
+      "step": 32760
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.261522266465285e-05,
+      "loss": 2.1194,
+      "step": 32770
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.261293423039957e-05,
+      "loss": 2.3136,
+      "step": 32780
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.261064579614628e-05,
+      "loss": 2.2408,
+      "step": 32790
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.260835736189299e-05,
+      "loss": 2.1984,
+      "step": 32800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.260606892763971e-05,
+      "loss": 2.2737,
+      "step": 32810
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2603780493386425e-05,
+      "loss": 2.0658,
+      "step": 32820
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.260149205913314e-05,
+      "loss": 2.1589,
+      "step": 32830
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.259920362487986e-05,
+      "loss": 2.0915,
+      "step": 32840
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2596915190626574e-05,
+      "loss": 2.3036,
+      "step": 32850
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.259462675637329e-05,
+      "loss": 2.3611,
+      "step": 32860
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.259233832212001e-05,
+      "loss": 2.3285,
+      "step": 32870
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.259004988786672e-05,
+      "loss": 2.2063,
+      "step": 32880
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.258776145361344e-05,
+      "loss": 2.2708,
+      "step": 32890
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.258547301936016e-05,
+      "loss": 2.1761,
+      "step": 32900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.258318458510687e-05,
+      "loss": 2.1592,
+      "step": 32910
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2580896150853586e-05,
+      "loss": 2.2499,
+      "step": 32920
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.257860771660031e-05,
+      "loss": 2.2562,
+      "step": 32930
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.257631928234702e-05,
+      "loss": 2.2092,
+      "step": 32940
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2574030848093735e-05,
+      "loss": 2.1096,
+      "step": 32950
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2571742413840456e-05,
+      "loss": 2.2354,
+      "step": 32960
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.256945397958717e-05,
+      "loss": 2.2108,
+      "step": 32970
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2567165545333885e-05,
+      "loss": 2.287,
+      "step": 32980
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2564877111080606e-05,
+      "loss": 2.2402,
+      "step": 32990
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.256258867682731e-05,
+      "loss": 2.3406,
+      "step": 33000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2560300242574034e-05,
+      "loss": 2.3176,
+      "step": 33010
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.255801180832075e-05,
+      "loss": 2.161,
+      "step": 33020
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.255572337406746e-05,
+      "loss": 2.1451,
+      "step": 33030
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.255343493981418e-05,
+      "loss": 2.235,
+      "step": 33040
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.25511465055609e-05,
+      "loss": 2.1922,
+      "step": 33050
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.254885807130761e-05,
+      "loss": 2.2576,
+      "step": 33060
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2546569637054325e-05,
+      "loss": 2.1162,
+      "step": 33070
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2544281202801046e-05,
+      "loss": 2.3137,
+      "step": 33080
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.254199276854776e-05,
+      "loss": 2.2528,
+      "step": 33090
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2539704334294475e-05,
+      "loss": 2.1501,
+      "step": 33100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2537415900041195e-05,
+      "loss": 2.0808,
+      "step": 33110
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.253512746578791e-05,
+      "loss": 2.0351,
+      "step": 33120
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2532839031534624e-05,
+      "loss": 2.2041,
+      "step": 33130
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2530550597281345e-05,
+      "loss": 2.2334,
+      "step": 33140
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.252826216302806e-05,
+      "loss": 2.347,
+      "step": 33150
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.252597372877477e-05,
+      "loss": 2.1335,
+      "step": 33160
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2523685294521494e-05,
+      "loss": 2.1263,
+      "step": 33170
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.252139686026821e-05,
+      "loss": 2.3133,
+      "step": 33180
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.251910842601492e-05,
+      "loss": 2.4409,
+      "step": 33190
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2516819991761636e-05,
+      "loss": 2.1563,
+      "step": 33200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.251453155750835e-05,
+      "loss": 2.1349,
+      "step": 33210
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.251224312325507e-05,
+      "loss": 2.1547,
+      "step": 33220
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2509954689001785e-05,
+      "loss": 2.0367,
+      "step": 33230
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.25076662547485e-05,
+      "loss": 2.245,
+      "step": 33240
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.250537782049522e-05,
+      "loss": 2.1288,
+      "step": 33250
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2503089386241935e-05,
+      "loss": 2.2669,
+      "step": 33260
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.250080095198865e-05,
+      "loss": 2.0192,
+      "step": 33270
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.249851251773537e-05,
+      "loss": 2.2243,
+      "step": 33280
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2496224083482084e-05,
+      "loss": 2.3135,
+      "step": 33290
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.24939356492288e-05,
+      "loss": 2.1103,
+      "step": 33300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.249164721497552e-05,
+      "loss": 2.1592,
+      "step": 33310
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.248935878072223e-05,
+      "loss": 2.1977,
+      "step": 33320
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.248707034646895e-05,
+      "loss": 2.2873,
+      "step": 33330
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.248478191221567e-05,
+      "loss": 2.2804,
+      "step": 33340
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.248249347796238e-05,
+      "loss": 2.268,
+      "step": 33350
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2480205043709096e-05,
+      "loss": 2.0027,
+      "step": 33360
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.247791660945582e-05,
+      "loss": 2.2616,
+      "step": 33370
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.247562817520253e-05,
+      "loss": 2.1679,
+      "step": 33380
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2473339740949245e-05,
+      "loss": 2.0082,
+      "step": 33390
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.247105130669596e-05,
+      "loss": 2.1035,
+      "step": 33400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2468762872442674e-05,
+      "loss": 2.2817,
+      "step": 33410
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.246647443818939e-05,
+      "loss": 2.2879,
+      "step": 33420
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.246418600393611e-05,
+      "loss": 2.1779,
+      "step": 33430
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.246189756968282e-05,
+      "loss": 2.3953,
+      "step": 33440
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.245960913542954e-05,
+      "loss": 2.1585,
+      "step": 33450
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.245732070117626e-05,
+      "loss": 2.2779,
+      "step": 33460
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.245503226692297e-05,
+      "loss": 2.228,
+      "step": 33470
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2452743832669686e-05,
+      "loss": 2.1003,
+      "step": 33480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.245045539841641e-05,
+      "loss": 2.2574,
+      "step": 33490
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.244816696416312e-05,
+      "loss": 2.0659,
+      "step": 33500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2445878529909835e-05,
+      "loss": 2.2704,
+      "step": 33510
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2443590095656556e-05,
+      "loss": 2.2114,
+      "step": 33520
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.244130166140327e-05,
+      "loss": 2.3033,
+      "step": 33530
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2439013227149985e-05,
+      "loss": 2.0605,
+      "step": 33540
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2436724792896705e-05,
+      "loss": 2.0271,
+      "step": 33550
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.243443635864342e-05,
+      "loss": 2.305,
+      "step": 33560
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2432147924390134e-05,
+      "loss": 2.2433,
+      "step": 33570
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2429859490136855e-05,
+      "loss": 2.2109,
+      "step": 33580
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.242757105588356e-05,
+      "loss": 2.1544,
+      "step": 33590
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.242528262163028e-05,
+      "loss": 2.1839,
+      "step": 33600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2422994187377e-05,
+      "loss": 2.2283,
+      "step": 33610
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.242070575312371e-05,
+      "loss": 2.2768,
+      "step": 33620
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.241841731887043e-05,
+      "loss": 2.0847,
+      "step": 33630
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2416128884617146e-05,
+      "loss": 2.0689,
+      "step": 33640
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.241384045036386e-05,
+      "loss": 2.2655,
+      "step": 33650
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.241155201611058e-05,
+      "loss": 2.2705,
+      "step": 33660
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2409263581857295e-05,
+      "loss": 2.2235,
+      "step": 33670
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.240697514760401e-05,
+      "loss": 2.251,
+      "step": 33680
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.240468671335073e-05,
+      "loss": 2.1739,
+      "step": 33690
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2402398279097445e-05,
+      "loss": 2.2423,
+      "step": 33700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.240010984484416e-05,
+      "loss": 2.1197,
+      "step": 33710
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.239782141059088e-05,
+      "loss": 2.2901,
+      "step": 33720
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2395532976337594e-05,
+      "loss": 2.186,
+      "step": 33730
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.239324454208431e-05,
+      "loss": 2.0998,
+      "step": 33740
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.239095610783103e-05,
+      "loss": 2.1911,
+      "step": 33750
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.238866767357774e-05,
+      "loss": 2.308,
+      "step": 33760
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.238637923932446e-05,
+      "loss": 2.2337,
+      "step": 33770
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.238409080507118e-05,
+      "loss": 2.1828,
+      "step": 33780
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2381802370817885e-05,
+      "loss": 2.1256,
+      "step": 33790
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.23795139365646e-05,
+      "loss": 2.1925,
+      "step": 33800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.237722550231132e-05,
+      "loss": 2.3036,
+      "step": 33810
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2374937068058034e-05,
+      "loss": 2.2225,
+      "step": 33820
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.237264863380475e-05,
+      "loss": 2.1024,
+      "step": 33830
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.237036019955147e-05,
+      "loss": 2.2119,
+      "step": 33840
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2368071765298184e-05,
+      "loss": 2.2511,
+      "step": 33850
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.23657833310449e-05,
+      "loss": 2.2003,
+      "step": 33860
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.236349489679162e-05,
+      "loss": 2.0514,
+      "step": 33870
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.236120646253833e-05,
+      "loss": 2.2364,
+      "step": 33880
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.235891802828505e-05,
+      "loss": 2.2835,
+      "step": 33890
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.235662959403177e-05,
+      "loss": 2.043,
+      "step": 33900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.235434115977848e-05,
+      "loss": 2.3057,
+      "step": 33910
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2352052725525196e-05,
+      "loss": 2.2361,
+      "step": 33920
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.234976429127192e-05,
+      "loss": 2.3479,
+      "step": 33930
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.234747585701863e-05,
+      "loss": 2.2349,
+      "step": 33940
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2345187422765345e-05,
+      "loss": 2.2076,
+      "step": 33950
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2342898988512066e-05,
+      "loss": 2.1177,
+      "step": 33960
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.234061055425878e-05,
+      "loss": 2.2607,
+      "step": 33970
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2338322120005494e-05,
+      "loss": 2.271,
+      "step": 33980
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.233603368575221e-05,
+      "loss": 2.2826,
+      "step": 33990
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.233374525149892e-05,
+      "loss": 2.2464,
+      "step": 34000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2331456817245644e-05,
+      "loss": 2.1265,
+      "step": 34010
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.232916838299236e-05,
+      "loss": 2.1062,
+      "step": 34020
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.232687994873907e-05,
+      "loss": 2.1647,
+      "step": 34030
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.232459151448579e-05,
+      "loss": 2.1633,
+      "step": 34040
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.232230308023251e-05,
+      "loss": 2.2624,
+      "step": 34050
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.232001464597922e-05,
+      "loss": 2.3734,
+      "step": 34060
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.231772621172594e-05,
+      "loss": 2.2075,
+      "step": 34070
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2315437777472656e-05,
+      "loss": 2.2135,
+      "step": 34080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.231314934321937e-05,
+      "loss": 2.2792,
+      "step": 34090
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.231086090896609e-05,
+      "loss": 2.1923,
+      "step": 34100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2308572474712805e-05,
+      "loss": 2.2663,
+      "step": 34110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.230628404045952e-05,
+      "loss": 2.2341,
+      "step": 34120
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.230399560620624e-05,
+      "loss": 2.1135,
+      "step": 34130
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2301707171952955e-05,
+      "loss": 2.1812,
+      "step": 34140
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.229941873769967e-05,
+      "loss": 2.2677,
+      "step": 34150
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.229713030344639e-05,
+      "loss": 2.1136,
+      "step": 34160
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2294841869193104e-05,
+      "loss": 2.4226,
+      "step": 34170
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.229255343493982e-05,
+      "loss": 2.2432,
+      "step": 34180
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.229026500068653e-05,
+      "loss": 2.1768,
+      "step": 34190
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2287976566433246e-05,
+      "loss": 2.2269,
+      "step": 34200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.228568813217996e-05,
+      "loss": 2.222,
+      "step": 34210
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.228339969792668e-05,
+      "loss": 2.1754,
+      "step": 34220
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2281111263673395e-05,
+      "loss": 2.3036,
+      "step": 34230
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.227882282942011e-05,
+      "loss": 2.2271,
+      "step": 34240
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.227653439516683e-05,
+      "loss": 2.2327,
+      "step": 34250
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2274245960913544e-05,
+      "loss": 2.2574,
+      "step": 34260
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.227195752666026e-05,
+      "loss": 2.2664,
+      "step": 34270
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.226966909240698e-05,
+      "loss": 2.2366,
+      "step": 34280
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2267380658153694e-05,
+      "loss": 2.1192,
+      "step": 34290
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.226509222390041e-05,
+      "loss": 2.1495,
+      "step": 34300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.226280378964713e-05,
+      "loss": 2.2084,
+      "step": 34310
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.226051535539384e-05,
+      "loss": 2.1518,
+      "step": 34320
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.225822692114056e-05,
+      "loss": 2.4809,
+      "step": 34330
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.225593848688728e-05,
+      "loss": 2.2692,
+      "step": 34340
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.225365005263399e-05,
+      "loss": 2.1326,
+      "step": 34350
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2251361618380706e-05,
+      "loss": 2.1786,
+      "step": 34360
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.224907318412743e-05,
+      "loss": 2.1346,
+      "step": 34370
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2246784749874134e-05,
+      "loss": 2.1845,
+      "step": 34380
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.224449631562085e-05,
+      "loss": 2.4012,
+      "step": 34390
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.224220788136757e-05,
+      "loss": 2.2577,
+      "step": 34400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2239919447114284e-05,
+      "loss": 2.1266,
+      "step": 34410
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2237631012861e-05,
+      "loss": 2.18,
+      "step": 34420
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.223534257860772e-05,
+      "loss": 2.1683,
+      "step": 34430
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.223305414435443e-05,
+      "loss": 2.1835,
+      "step": 34440
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.223076571010115e-05,
+      "loss": 2.0724,
+      "step": 34450
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.222847727584787e-05,
+      "loss": 2.1576,
+      "step": 34460
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.222618884159458e-05,
+      "loss": 2.2414,
+      "step": 34470
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2223900407341296e-05,
+      "loss": 2.2214,
+      "step": 34480
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.222161197308802e-05,
+      "loss": 2.2895,
+      "step": 34490
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.221932353883473e-05,
+      "loss": 2.1561,
+      "step": 34500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2217035104581445e-05,
+      "loss": 2.1956,
+      "step": 34510
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2214746670328166e-05,
+      "loss": 2.3574,
+      "step": 34520
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.221245823607488e-05,
+      "loss": 2.2993,
+      "step": 34530
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2210169801821594e-05,
+      "loss": 2.1817,
+      "step": 34540
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2207881367568315e-05,
+      "loss": 2.2136,
+      "step": 34550
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.220559293331503e-05,
+      "loss": 2.1727,
+      "step": 34560
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2203304499061744e-05,
+      "loss": 2.0491,
+      "step": 34570
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.220101606480846e-05,
+      "loss": 2.1265,
+      "step": 34580
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.219872763055517e-05,
+      "loss": 2.2473,
+      "step": 34590
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.219643919630189e-05,
+      "loss": 2.1427,
+      "step": 34600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.219415076204861e-05,
+      "loss": 2.3436,
+      "step": 34610
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.219186232779532e-05,
+      "loss": 2.1976,
+      "step": 34620
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.218957389354204e-05,
+      "loss": 2.0964,
+      "step": 34630
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2187285459288756e-05,
+      "loss": 2.1698,
+      "step": 34640
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.218499702503547e-05,
+      "loss": 2.3433,
+      "step": 34650
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.218270859078219e-05,
+      "loss": 2.0955,
+      "step": 34660
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2180420156528905e-05,
+      "loss": 2.2234,
+      "step": 34670
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.217813172227562e-05,
+      "loss": 2.2557,
+      "step": 34680
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.217584328802234e-05,
+      "loss": 2.3399,
+      "step": 34690
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2173554853769054e-05,
+      "loss": 2.0831,
+      "step": 34700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.217126641951577e-05,
+      "loss": 2.2014,
+      "step": 34710
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.216897798526249e-05,
+      "loss": 2.1051,
+      "step": 34720
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2166689551009204e-05,
+      "loss": 2.2036,
+      "step": 34730
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.216440111675592e-05,
+      "loss": 2.1197,
+      "step": 34740
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.216211268250264e-05,
+      "loss": 2.357,
+      "step": 34750
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.215982424824935e-05,
+      "loss": 2.2071,
+      "step": 34760
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.215753581399607e-05,
+      "loss": 2.2252,
+      "step": 34770
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.215524737974278e-05,
+      "loss": 2.0906,
+      "step": 34780
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2152958945489495e-05,
+      "loss": 2.2578,
+      "step": 34790
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.215067051123621e-05,
+      "loss": 1.9847,
+      "step": 34800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.214838207698293e-05,
+      "loss": 2.2064,
+      "step": 34810
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2146093642729644e-05,
+      "loss": 2.3614,
+      "step": 34820
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.214380520847636e-05,
+      "loss": 2.3928,
+      "step": 34830
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.214151677422308e-05,
+      "loss": 2.265,
+      "step": 34840
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2139228339969794e-05,
+      "loss": 2.1797,
+      "step": 34850
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.213693990571651e-05,
+      "loss": 2.2691,
+      "step": 34860
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.213465147146323e-05,
+      "loss": 2.347,
+      "step": 34870
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.213236303720994e-05,
+      "loss": 2.0507,
+      "step": 34880
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.213007460295666e-05,
+      "loss": 2.1732,
+      "step": 34890
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.212778616870338e-05,
+      "loss": 2.0792,
+      "step": 34900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.212549773445009e-05,
+      "loss": 2.2117,
+      "step": 34910
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2123209300196806e-05,
+      "loss": 2.2029,
+      "step": 34920
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.212092086594353e-05,
+      "loss": 2.1016,
+      "step": 34930
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.211863243169024e-05,
+      "loss": 2.1843,
+      "step": 34940
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2116343997436955e-05,
+      "loss": 2.24,
+      "step": 34950
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2114055563183676e-05,
+      "loss": 2.0722,
+      "step": 34960
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.211176712893039e-05,
+      "loss": 2.1436,
+      "step": 34970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2109478694677104e-05,
+      "loss": 2.1719,
+      "step": 34980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.210719026042382e-05,
+      "loss": 2.1263,
+      "step": 34990
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.210490182617053e-05,
+      "loss": 2.2023,
+      "step": 35000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2102613391917254e-05,
+      "loss": 2.2307,
+      "step": 35010
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.210032495766397e-05,
+      "loss": 2.2277,
+      "step": 35020
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.209803652341068e-05,
+      "loss": 2.1839,
+      "step": 35030
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.20957480891574e-05,
+      "loss": 2.2606,
+      "step": 35040
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.209345965490412e-05,
+      "loss": 2.0376,
+      "step": 35050
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.209117122065083e-05,
+      "loss": 2.1165,
+      "step": 35060
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.208888278639755e-05,
+      "loss": 2.3497,
+      "step": 35070
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2086594352144266e-05,
+      "loss": 2.3684,
+      "step": 35080
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.208430591789098e-05,
+      "loss": 2.1159,
+      "step": 35090
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.20820174836377e-05,
+      "loss": 2.2071,
+      "step": 35100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2079729049384415e-05,
+      "loss": 2.2192,
+      "step": 35110
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.207744061513113e-05,
+      "loss": 2.303,
+      "step": 35120
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.207515218087785e-05,
+      "loss": 2.2169,
+      "step": 35130
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2072863746624564e-05,
+      "loss": 2.209,
+      "step": 35140
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.207057531237128e-05,
+      "loss": 2.2105,
+      "step": 35150
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2068286878118e-05,
+      "loss": 2.392,
+      "step": 35160
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.206599844386471e-05,
+      "loss": 2.2361,
+      "step": 35170
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.206371000961142e-05,
+      "loss": 2.148,
+      "step": 35180
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.206142157535814e-05,
+      "loss": 2.2382,
+      "step": 35190
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2059133141104856e-05,
+      "loss": 2.2803,
+      "step": 35200
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.205684470685157e-05,
+      "loss": 2.0674,
+      "step": 35210
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.205455627259829e-05,
+      "loss": 2.23,
+      "step": 35220
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2052267838345005e-05,
+      "loss": 2.1598,
+      "step": 35230
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.204997940409172e-05,
+      "loss": 2.2057,
+      "step": 35240
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.204769096983844e-05,
+      "loss": 2.2194,
+      "step": 35250
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2045402535585154e-05,
+      "loss": 2.1509,
+      "step": 35260
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.204311410133187e-05,
+      "loss": 2.2268,
+      "step": 35270
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.204082566707859e-05,
+      "loss": 2.303,
+      "step": 35280
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2038537232825303e-05,
+      "loss": 2.1539,
+      "step": 35290
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.203624879857202e-05,
+      "loss": 2.3462,
+      "step": 35300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.203396036431874e-05,
+      "loss": 2.2761,
+      "step": 35310
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.203167193006545e-05,
+      "loss": 2.194,
+      "step": 35320
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.202938349581217e-05,
+      "loss": 2.2376,
+      "step": 35330
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.202709506155889e-05,
+      "loss": 2.2305,
+      "step": 35340
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.20248066273056e-05,
+      "loss": 2.3042,
+      "step": 35350
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2022518193052316e-05,
+      "loss": 2.0659,
+      "step": 35360
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.202022975879903e-05,
+      "loss": 2.0786,
+      "step": 35370
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.2017941324545744e-05,
+      "loss": 2.2156,
+      "step": 35380
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.2015652890292465e-05,
+      "loss": 2.1824,
+      "step": 35390
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.201336445603918e-05,
+      "loss": 2.2179,
+      "step": 35400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.2011076021785893e-05,
+      "loss": 2.1806,
+      "step": 35410
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.200878758753261e-05,
+      "loss": 2.192,
+      "step": 35420
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.200649915327933e-05,
+      "loss": 2.2702,
+      "step": 35430
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.200421071902604e-05,
+      "loss": 2.1344,
+      "step": 35440
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.200192228477276e-05,
+      "loss": 2.1847,
+      "step": 35450
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.199963385051948e-05,
+      "loss": 2.2211,
+      "step": 35460
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.199734541626619e-05,
+      "loss": 2.2877,
+      "step": 35470
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1995056982012906e-05,
+      "loss": 2.2058,
+      "step": 35480
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.199276854775963e-05,
+      "loss": 2.2641,
+      "step": 35490
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.199048011350634e-05,
+      "loss": 2.1562,
+      "step": 35500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1988191679253055e-05,
+      "loss": 2.0076,
+      "step": 35510
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1985903244999776e-05,
+      "loss": 2.1856,
+      "step": 35520
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.198361481074649e-05,
+      "loss": 2.3047,
+      "step": 35530
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1981326376493204e-05,
+      "loss": 2.4964,
+      "step": 35540
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1979037942239925e-05,
+      "loss": 2.1926,
+      "step": 35550
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.197674950798664e-05,
+      "loss": 2.1895,
+      "step": 35560
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1974461073733353e-05,
+      "loss": 2.2181,
+      "step": 35570
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.197217263948007e-05,
+      "loss": 2.2334,
+      "step": 35580
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.196988420522678e-05,
+      "loss": 2.2817,
+      "step": 35590
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.19675957709735e-05,
+      "loss": 2.2428,
+      "step": 35600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.196530733672022e-05,
+      "loss": 1.9813,
+      "step": 35610
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.196301890246693e-05,
+      "loss": 2.4459,
+      "step": 35620
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.196073046821365e-05,
+      "loss": 2.0868,
+      "step": 35630
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1958442033960366e-05,
+      "loss": 2.1707,
+      "step": 35640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.195615359970708e-05,
+      "loss": 2.2882,
+      "step": 35650
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.19538651654538e-05,
+      "loss": 2.267,
+      "step": 35660
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1951576731200515e-05,
+      "loss": 2.1949,
+      "step": 35670
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.194928829694723e-05,
+      "loss": 2.2133,
+      "step": 35680
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.194699986269395e-05,
+      "loss": 2.2678,
+      "step": 35690
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1944711428440664e-05,
+      "loss": 2.2512,
+      "step": 35700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.194242299418738e-05,
+      "loss": 2.2948,
+      "step": 35710
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.19401345599341e-05,
+      "loss": 2.1776,
+      "step": 35720
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1937846125680813e-05,
+      "loss": 2.172,
+      "step": 35730
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.193555769142753e-05,
+      "loss": 2.1015,
+      "step": 35740
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.193326925717425e-05,
+      "loss": 2.1385,
+      "step": 35750
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.193098082292096e-05,
+      "loss": 2.1631,
+      "step": 35760
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.192869238866767e-05,
+      "loss": 2.1801,
+      "step": 35770
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.192640395441439e-05,
+      "loss": 2.1689,
+      "step": 35780
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1924115520161105e-05,
+      "loss": 2.4101,
+      "step": 35790
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.192182708590782e-05,
+      "loss": 2.1634,
+      "step": 35800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.191953865165454e-05,
+      "loss": 2.196,
+      "step": 35810
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1917250217401254e-05,
+      "loss": 2.0348,
+      "step": 35820
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.191496178314797e-05,
+      "loss": 2.0357,
+      "step": 35830
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.191267334889469e-05,
+      "loss": 2.1229,
+      "step": 35840
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.19103849146414e-05,
+      "loss": 2.1913,
+      "step": 35850
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.190809648038812e-05,
+      "loss": 2.1357,
+      "step": 35860
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.190580804613484e-05,
+      "loss": 1.9988,
+      "step": 35870
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.190351961188155e-05,
+      "loss": 2.1352,
+      "step": 35880
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.190123117762827e-05,
+      "loss": 2.013,
+      "step": 35890
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.189894274337499e-05,
+      "loss": 2.3225,
+      "step": 35900
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.18966543091217e-05,
+      "loss": 2.0352,
+      "step": 35910
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1894365874868416e-05,
+      "loss": 2.2108,
+      "step": 35920
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.189207744061514e-05,
+      "loss": 2.2225,
+      "step": 35930
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.188978900636185e-05,
+      "loss": 2.2175,
+      "step": 35940
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1887500572108565e-05,
+      "loss": 2.1259,
+      "step": 35950
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1885212137855286e-05,
+      "loss": 2.2423,
+      "step": 35960
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.188292370360199e-05,
+      "loss": 2.2987,
+      "step": 35970
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1880635269348714e-05,
+      "loss": 2.1954,
+      "step": 35980
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.187834683509543e-05,
+      "loss": 2.2845,
+      "step": 35990
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.187605840084214e-05,
+      "loss": 2.1676,
+      "step": 36000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1873769966588863e-05,
+      "loss": 2.2662,
+      "step": 36010
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.187148153233558e-05,
+      "loss": 2.204,
+      "step": 36020
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.186919309808229e-05,
+      "loss": 2.3134,
+      "step": 36030
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.186690466382901e-05,
+      "loss": 2.1039,
+      "step": 36040
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.186461622957573e-05,
+      "loss": 2.2618,
+      "step": 36050
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.186232779532244e-05,
+      "loss": 2.1634,
+      "step": 36060
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.186003936106916e-05,
+      "loss": 2.1834,
+      "step": 36070
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1857750926815876e-05,
+      "loss": 2.2815,
+      "step": 36080
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.185546249256259e-05,
+      "loss": 2.2876,
+      "step": 36090
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.185317405830931e-05,
+      "loss": 2.3654,
+      "step": 36100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1850885624056025e-05,
+      "loss": 2.3615,
+      "step": 36110
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.184859718980274e-05,
+      "loss": 2.2244,
+      "step": 36120
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.184630875554946e-05,
+      "loss": 2.1728,
+      "step": 36130
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1844020321296174e-05,
+      "loss": 2.1475,
+      "step": 36140
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.184173188704289e-05,
+      "loss": 2.1333,
+      "step": 36150
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.18394434527896e-05,
+      "loss": 2.2034,
+      "step": 36160
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.183715501853632e-05,
+      "loss": 2.1358,
+      "step": 36170
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.183486658428303e-05,
+      "loss": 2.1884,
+      "step": 36180
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.183257815002975e-05,
+      "loss": 2.1662,
+      "step": 36190
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1830289715776466e-05,
+      "loss": 2.1639,
+      "step": 36200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.182800128152318e-05,
+      "loss": 2.0841,
+      "step": 36210
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.18257128472699e-05,
+      "loss": 2.1795,
+      "step": 36220
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1823424413016615e-05,
+      "loss": 2.14,
+      "step": 36230
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.182113597876333e-05,
+      "loss": 2.1729,
+      "step": 36240
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.181884754451005e-05,
+      "loss": 2.1571,
+      "step": 36250
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1816559110256764e-05,
+      "loss": 2.1301,
+      "step": 36260
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.181427067600348e-05,
+      "loss": 2.2749,
+      "step": 36270
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.18119822417502e-05,
+      "loss": 2.138,
+      "step": 36280
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.180969380749691e-05,
+      "loss": 2.3211,
+      "step": 36290
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.180740537324363e-05,
+      "loss": 2.1484,
+      "step": 36300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.180511693899035e-05,
+      "loss": 2.22,
+      "step": 36310
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.180282850473706e-05,
+      "loss": 2.0964,
+      "step": 36320
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.180054007048378e-05,
+      "loss": 2.2091,
+      "step": 36330
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.17982516362305e-05,
+      "loss": 2.3755,
+      "step": 36340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.179596320197721e-05,
+      "loss": 2.1484,
+      "step": 36350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1793674767723926e-05,
+      "loss": 2.1213,
+      "step": 36360
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.179138633347064e-05,
+      "loss": 2.1509,
+      "step": 36370
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1789097899217354e-05,
+      "loss": 2.2193,
+      "step": 36380
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1786809464964075e-05,
+      "loss": 2.2811,
+      "step": 36390
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.178452103071079e-05,
+      "loss": 2.1019,
+      "step": 36400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.17822325964575e-05,
+      "loss": 2.2255,
+      "step": 36410
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1779944162204224e-05,
+      "loss": 2.2865,
+      "step": 36420
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.177765572795094e-05,
+      "loss": 2.2252,
+      "step": 36430
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.177536729369765e-05,
+      "loss": 2.2945,
+      "step": 36440
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.177307885944437e-05,
+      "loss": 2.2925,
+      "step": 36450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.177079042519109e-05,
+      "loss": 2.1438,
+      "step": 36460
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.17685019909378e-05,
+      "loss": 2.3323,
+      "step": 36470
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.176621355668452e-05,
+      "loss": 2.3067,
+      "step": 36480
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.176392512243124e-05,
+      "loss": 2.1549,
+      "step": 36490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.176163668817795e-05,
+      "loss": 2.3096,
+      "step": 36500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1759348253924665e-05,
+      "loss": 2.2461,
+      "step": 36510
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1757059819671386e-05,
+      "loss": 2.2352,
+      "step": 36520
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.17547713854181e-05,
+      "loss": 2.2697,
+      "step": 36530
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1752482951164814e-05,
+      "loss": 2.1806,
+      "step": 36540
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1750194516911535e-05,
+      "loss": 2.3018,
+      "step": 36550
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.174790608265824e-05,
+      "loss": 2.2201,
+      "step": 36560
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.174561764840496e-05,
+      "loss": 2.0023,
+      "step": 36570
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.174332921415168e-05,
+      "loss": 2.3606,
+      "step": 36580
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.174104077989839e-05,
+      "loss": 2.0941,
+      "step": 36590
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.173875234564511e-05,
+      "loss": 2.1995,
+      "step": 36600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1736463911391827e-05,
+      "loss": 2.2288,
+      "step": 36610
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.173417547713854e-05,
+      "loss": 2.1888,
+      "step": 36620
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.173188704288526e-05,
+      "loss": 2.156,
+      "step": 36630
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1729598608631976e-05,
+      "loss": 2.1314,
+      "step": 36640
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.172731017437869e-05,
+      "loss": 2.2787,
+      "step": 36650
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.172502174012541e-05,
+      "loss": 1.9101,
+      "step": 36660
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1722733305872125e-05,
+      "loss": 2.1467,
+      "step": 36670
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.172044487161884e-05,
+      "loss": 2.1305,
+      "step": 36680
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.171815643736556e-05,
+      "loss": 2.182,
+      "step": 36690
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1715868003112274e-05,
+      "loss": 2.3326,
+      "step": 36700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.171357956885899e-05,
+      "loss": 2.2136,
+      "step": 36710
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.171129113460571e-05,
+      "loss": 2.2463,
+      "step": 36720
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.170900270035242e-05,
+      "loss": 2.1664,
+      "step": 36730
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.170671426609914e-05,
+      "loss": 2.271,
+      "step": 36740
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.170442583184586e-05,
+      "loss": 2.2753,
+      "step": 36750
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1702137397592566e-05,
+      "loss": 2.1586,
+      "step": 36760
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.169984896333928e-05,
+      "loss": 2.1588,
+      "step": 36770
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1697560529086e-05,
+      "loss": 2.1353,
+      "step": 36780
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1695272094832715e-05,
+      "loss": 2.0176,
+      "step": 36790
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.169298366057943e-05,
+      "loss": 2.0487,
+      "step": 36800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.169069522632615e-05,
+      "loss": 2.1775,
+      "step": 36810
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1688406792072864e-05,
+      "loss": 2.1976,
+      "step": 36820
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.168611835781958e-05,
+      "loss": 2.2163,
+      "step": 36830
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.16838299235663e-05,
+      "loss": 2.1888,
+      "step": 36840
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.168154148931301e-05,
+      "loss": 2.2368,
+      "step": 36850
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.167925305505973e-05,
+      "loss": 2.2871,
+      "step": 36860
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.167696462080645e-05,
+      "loss": 2.2263,
+      "step": 36870
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.167467618655316e-05,
+      "loss": 2.3149,
+      "step": 36880
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1672387752299877e-05,
+      "loss": 2.2872,
+      "step": 36890
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.16700993180466e-05,
+      "loss": 2.1909,
+      "step": 36900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.166781088379331e-05,
+      "loss": 2.3224,
+      "step": 36910
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1665522449540026e-05,
+      "loss": 2.2413,
+      "step": 36920
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.166323401528675e-05,
+      "loss": 2.1985,
+      "step": 36930
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.166094558103346e-05,
+      "loss": 2.1627,
+      "step": 36940
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1658657146780175e-05,
+      "loss": 2.1385,
+      "step": 36950
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.165636871252689e-05,
+      "loss": 2.2795,
+      "step": 36960
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.16540802782736e-05,
+      "loss": 2.2094,
+      "step": 36970
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1651791844020324e-05,
+      "loss": 2.1953,
+      "step": 36980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.164950340976704e-05,
+      "loss": 2.2623,
+      "step": 36990
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.164721497551375e-05,
+      "loss": 2.2628,
+      "step": 37000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.164492654126047e-05,
+      "loss": 2.2957,
+      "step": 37010
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.164263810700719e-05,
+      "loss": 2.2331,
+      "step": 37020
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.16403496727539e-05,
+      "loss": 2.1442,
+      "step": 37030
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.163806123850062e-05,
+      "loss": 2.1199,
+      "step": 37040
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1635772804247337e-05,
+      "loss": 2.1521,
+      "step": 37050
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.163348436999405e-05,
+      "loss": 2.1245,
+      "step": 37060
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.163119593574077e-05,
+      "loss": 2.1625,
+      "step": 37070
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1628907501487486e-05,
+      "loss": 2.1908,
+      "step": 37080
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.16266190672342e-05,
+      "loss": 2.1789,
+      "step": 37090
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.162433063298092e-05,
+      "loss": 2.0488,
+      "step": 37100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1622042198727635e-05,
+      "loss": 2.1799,
+      "step": 37110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.161975376447435e-05,
+      "loss": 2.2801,
+      "step": 37120
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.161746533022107e-05,
+      "loss": 2.2477,
+      "step": 37130
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1615176895967784e-05,
+      "loss": 2.1636,
+      "step": 37140
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.161288846171449e-05,
+      "loss": 2.3028,
+      "step": 37150
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.161060002746121e-05,
+      "loss": 2.271,
+      "step": 37160
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1608311593207927e-05,
+      "loss": 2.3294,
+      "step": 37170
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.160602315895464e-05,
+      "loss": 2.2149,
+      "step": 37180
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.160373472470136e-05,
+      "loss": 2.1625,
+      "step": 37190
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1601446290448076e-05,
+      "loss": 2.3488,
+      "step": 37200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.159915785619479e-05,
+      "loss": 2.157,
+      "step": 37210
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.159686942194151e-05,
+      "loss": 2.0363,
+      "step": 37220
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1594580987688225e-05,
+      "loss": 2.1384,
+      "step": 37230
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.159229255343494e-05,
+      "loss": 2.1258,
+      "step": 37240
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.159000411918166e-05,
+      "loss": 2.2632,
+      "step": 37250
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1587715684928374e-05,
+      "loss": 2.07,
+      "step": 37260
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.158542725067509e-05,
+      "loss": 2.2473,
+      "step": 37270
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.158313881642181e-05,
+      "loss": 2.1957,
+      "step": 37280
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.158085038216852e-05,
+      "loss": 2.2194,
+      "step": 37290
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.157856194791524e-05,
+      "loss": 2.1726,
+      "step": 37300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.157627351366196e-05,
+      "loss": 2.0769,
+      "step": 37310
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.157398507940867e-05,
+      "loss": 2.2937,
+      "step": 37320
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1571696645155387e-05,
+      "loss": 2.1365,
+      "step": 37330
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.156940821090211e-05,
+      "loss": 2.067,
+      "step": 37340
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1567119776648815e-05,
+      "loss": 2.1474,
+      "step": 37350
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1564831342395536e-05,
+      "loss": 2.1031,
+      "step": 37360
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.156254290814225e-05,
+      "loss": 2.1677,
+      "step": 37370
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1560254473888964e-05,
+      "loss": 2.2256,
+      "step": 37380
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1557966039635685e-05,
+      "loss": 2.3279,
+      "step": 37390
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.15556776053824e-05,
+      "loss": 2.1411,
+      "step": 37400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.155338917112911e-05,
+      "loss": 2.3298,
+      "step": 37410
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1551100736875834e-05,
+      "loss": 2.2089,
+      "step": 37420
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.154881230262255e-05,
+      "loss": 2.1381,
+      "step": 37430
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.154652386836926e-05,
+      "loss": 2.1928,
+      "step": 37440
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.154423543411598e-05,
+      "loss": 2.2086,
+      "step": 37450
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.15419469998627e-05,
+      "loss": 2.1065,
+      "step": 37460
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.153965856560941e-05,
+      "loss": 2.2628,
+      "step": 37470
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.153737013135613e-05,
+      "loss": 2.2599,
+      "step": 37480
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1535081697102847e-05,
+      "loss": 2.1146,
+      "step": 37490
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.153279326284956e-05,
+      "loss": 2.2617,
+      "step": 37500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.153050482859628e-05,
+      "loss": 2.0535,
+      "step": 37510
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1528216394342996e-05,
+      "loss": 2.3426,
+      "step": 37520
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.152592796008971e-05,
+      "loss": 2.3054,
+      "step": 37530
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.152363952583643e-05,
+      "loss": 2.1889,
+      "step": 37540
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.152135109158314e-05,
+      "loss": 2.2644,
+      "step": 37550
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.151906265732985e-05,
+      "loss": 2.2223,
+      "step": 37560
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.151677422307657e-05,
+      "loss": 2.2571,
+      "step": 37570
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.151448578882329e-05,
+      "loss": 2.2414,
+      "step": 37580
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.151219735457e-05,
+      "loss": 2.383,
+      "step": 37590
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.150990892031672e-05,
+      "loss": 2.1785,
+      "step": 37600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1507620486063436e-05,
+      "loss": 2.0195,
+      "step": 37610
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.150533205181015e-05,
+      "loss": 2.2087,
+      "step": 37620
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.150304361755687e-05,
+      "loss": 2.0984,
+      "step": 37630
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1500755183303586e-05,
+      "loss": 2.2044,
+      "step": 37640
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.14984667490503e-05,
+      "loss": 2.2106,
+      "step": 37650
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.149617831479702e-05,
+      "loss": 2.1247,
+      "step": 37660
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1493889880543735e-05,
+      "loss": 2.118,
+      "step": 37670
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.149160144629045e-05,
+      "loss": 2.2166,
+      "step": 37680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.148931301203717e-05,
+      "loss": 2.2971,
+      "step": 37690
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1487024577783884e-05,
+      "loss": 2.2142,
+      "step": 37700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.14847361435306e-05,
+      "loss": 2.214,
+      "step": 37710
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.148244770927732e-05,
+      "loss": 2.2088,
+      "step": 37720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.148015927502403e-05,
+      "loss": 2.0875,
+      "step": 37730
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.147787084077074e-05,
+      "loss": 2.2364,
+      "step": 37740
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.147558240651746e-05,
+      "loss": 2.1708,
+      "step": 37750
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1473293972264176e-05,
+      "loss": 2.1383,
+      "step": 37760
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.147100553801089e-05,
+      "loss": 2.0741,
+      "step": 37770
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.146871710375761e-05,
+      "loss": 2.2901,
+      "step": 37780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1466428669504325e-05,
+      "loss": 2.1857,
+      "step": 37790
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.146414023525104e-05,
+      "loss": 2.0843,
+      "step": 37800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.146185180099776e-05,
+      "loss": 2.1526,
+      "step": 37810
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1459563366744474e-05,
+      "loss": 2.0827,
+      "step": 37820
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.145727493249119e-05,
+      "loss": 2.1325,
+      "step": 37830
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.145498649823791e-05,
+      "loss": 2.2094,
+      "step": 37840
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.145269806398462e-05,
+      "loss": 2.2227,
+      "step": 37850
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.145040962973134e-05,
+      "loss": 2.2341,
+      "step": 37860
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.144812119547806e-05,
+      "loss": 2.1889,
+      "step": 37870
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.144583276122477e-05,
+      "loss": 2.2434,
+      "step": 37880
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1443544326971486e-05,
+      "loss": 2.114,
+      "step": 37890
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.144125589271821e-05,
+      "loss": 2.3075,
+      "step": 37900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.143896745846492e-05,
+      "loss": 2.2841,
+      "step": 37910
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1436679024211636e-05,
+      "loss": 2.1142,
+      "step": 37920
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1434390589958357e-05,
+      "loss": 2.167,
+      "step": 37930
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1432102155705064e-05,
+      "loss": 2.1222,
+      "step": 37940
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1429813721451785e-05,
+      "loss": 2.1025,
+      "step": 37950
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.14275252871985e-05,
+      "loss": 2.241,
+      "step": 37960
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.142523685294521e-05,
+      "loss": 2.1692,
+      "step": 37970
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1422948418691934e-05,
+      "loss": 2.1924,
+      "step": 37980
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.142065998443865e-05,
+      "loss": 2.2379,
+      "step": 37990
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.141837155018536e-05,
+      "loss": 2.1543,
+      "step": 38000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.141608311593208e-05,
+      "loss": 2.218,
+      "step": 38010
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.14137946816788e-05,
+      "loss": 2.2276,
+      "step": 38020
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.141150624742551e-05,
+      "loss": 2.1639,
+      "step": 38030
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.140921781317223e-05,
+      "loss": 2.0979,
+      "step": 38040
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1406929378918946e-05,
+      "loss": 2.0996,
+      "step": 38050
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.140464094466566e-05,
+      "loss": 2.0435,
+      "step": 38060
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.140235251041238e-05,
+      "loss": 2.2641,
+      "step": 38070
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1400064076159096e-05,
+      "loss": 2.1746,
+      "step": 38080
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.139777564190581e-05,
+      "loss": 2.2327,
+      "step": 38090
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.139548720765253e-05,
+      "loss": 2.2072,
+      "step": 38100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1393198773399245e-05,
+      "loss": 2.234,
+      "step": 38110
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.139091033914596e-05,
+      "loss": 2.1555,
+      "step": 38120
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.138862190489268e-05,
+      "loss": 1.9149,
+      "step": 38130
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.138633347063939e-05,
+      "loss": 2.2824,
+      "step": 38140
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.13840450363861e-05,
+      "loss": 2.1314,
+      "step": 38150
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.138175660213282e-05,
+      "loss": 2.1325,
+      "step": 38160
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1379468167879536e-05,
+      "loss": 2.0525,
+      "step": 38170
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.137717973362625e-05,
+      "loss": 2.228,
+      "step": 38180
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.137489129937297e-05,
+      "loss": 2.1375,
+      "step": 38190
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1372602865119686e-05,
+      "loss": 2.1777,
+      "step": 38200
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.13703144308664e-05,
+      "loss": 2.0687,
+      "step": 38210
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.136802599661312e-05,
+      "loss": 2.1486,
+      "step": 38220
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1365737562359835e-05,
+      "loss": 2.2134,
+      "step": 38230
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.136344912810655e-05,
+      "loss": 2.1767,
+      "step": 38240
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.136116069385327e-05,
+      "loss": 2.1079,
+      "step": 38250
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1358872259599984e-05,
+      "loss": 2.0588,
+      "step": 38260
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.13565838253467e-05,
+      "loss": 2.2988,
+      "step": 38270
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.135429539109342e-05,
+      "loss": 2.1837,
+      "step": 38280
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.135200695684013e-05,
+      "loss": 2.1671,
+      "step": 38290
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.134971852258685e-05,
+      "loss": 2.1453,
+      "step": 38300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.134743008833357e-05,
+      "loss": 2.141,
+      "step": 38310
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.134514165408028e-05,
+      "loss": 2.2167,
+      "step": 38320
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1342853219826996e-05,
+      "loss": 2.1915,
+      "step": 38330
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.134056478557371e-05,
+      "loss": 2.18,
+      "step": 38340
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1338276351320425e-05,
+      "loss": 2.2206,
+      "step": 38350
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1335987917067146e-05,
+      "loss": 2.2051,
+      "step": 38360
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.133369948281386e-05,
+      "loss": 2.0144,
+      "step": 38370
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1331411048560574e-05,
+      "loss": 2.1683,
+      "step": 38380
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1329122614307295e-05,
+      "loss": 2.1199,
+      "step": 38390
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.132683418005401e-05,
+      "loss": 2.2722,
+      "step": 38400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.132454574580072e-05,
+      "loss": 2.2842,
+      "step": 38410
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1322257311547444e-05,
+      "loss": 2.0474,
+      "step": 38420
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.131996887729416e-05,
+      "loss": 2.2541,
+      "step": 38430
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.131768044304087e-05,
+      "loss": 2.1254,
+      "step": 38440
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.131539200878759e-05,
+      "loss": 2.3087,
+      "step": 38450
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.131310357453431e-05,
+      "loss": 2.0168,
+      "step": 38460
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.131081514028102e-05,
+      "loss": 2.0471,
+      "step": 38470
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.130852670602774e-05,
+      "loss": 2.3601,
+      "step": 38480
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1306238271774456e-05,
+      "loss": 2.1867,
+      "step": 38490
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.130394983752117e-05,
+      "loss": 2.192,
+      "step": 38500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.130166140326789e-05,
+      "loss": 2.145,
+      "step": 38510
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1299372969014606e-05,
+      "loss": 2.2573,
+      "step": 38520
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.129708453476131e-05,
+      "loss": 2.1622,
+      "step": 38530
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1294796100508034e-05,
+      "loss": 2.126,
+      "step": 38540
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.129250766625475e-05,
+      "loss": 2.1473,
+      "step": 38550
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.129021923200146e-05,
+      "loss": 2.3138,
+      "step": 38560
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.128793079774818e-05,
+      "loss": 2.2366,
+      "step": 38570
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.12856423634949e-05,
+      "loss": 2.238,
+      "step": 38580
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.128335392924161e-05,
+      "loss": 2.1841,
+      "step": 38590
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.128106549498833e-05,
+      "loss": 2.1524,
+      "step": 38600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1278777060735046e-05,
+      "loss": 2.1959,
+      "step": 38610
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.127648862648176e-05,
+      "loss": 2.2588,
+      "step": 38620
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.127420019222848e-05,
+      "loss": 2.3358,
+      "step": 38630
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1271911757975196e-05,
+      "loss": 2.1211,
+      "step": 38640
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.126962332372191e-05,
+      "loss": 2.199,
+      "step": 38650
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.126733488946863e-05,
+      "loss": 2.1338,
+      "step": 38660
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1265046455215345e-05,
+      "loss": 2.1359,
+      "step": 38670
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.126275802096206e-05,
+      "loss": 2.2716,
+      "step": 38680
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.126046958670878e-05,
+      "loss": 2.264,
+      "step": 38690
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1258181152455494e-05,
+      "loss": 2.0558,
+      "step": 38700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.125589271820221e-05,
+      "loss": 2.3346,
+      "step": 38710
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.125360428394893e-05,
+      "loss": 2.1571,
+      "step": 38720
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1251315849695636e-05,
+      "loss": 2.1586,
+      "step": 38730
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.124902741544236e-05,
+      "loss": 2.2251,
+      "step": 38740
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.124673898118907e-05,
+      "loss": 2.0975,
+      "step": 38750
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1244450546935785e-05,
+      "loss": 2.2225,
+      "step": 38760
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1242162112682506e-05,
+      "loss": 2.1715,
+      "step": 38770
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.123987367842922e-05,
+      "loss": 2.1969,
+      "step": 38780
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1237585244175935e-05,
+      "loss": 2.134,
+      "step": 38790
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1235296809922656e-05,
+      "loss": 2.2035,
+      "step": 38800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.123300837566937e-05,
+      "loss": 2.0994,
+      "step": 38810
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1230719941416084e-05,
+      "loss": 2.1816,
+      "step": 38820
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1228431507162805e-05,
+      "loss": 2.1874,
+      "step": 38830
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.122614307290952e-05,
+      "loss": 2.3196,
+      "step": 38840
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.122385463865623e-05,
+      "loss": 2.1777,
+      "step": 38850
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.122156620440295e-05,
+      "loss": 2.2047,
+      "step": 38860
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.121927777014967e-05,
+      "loss": 2.1705,
+      "step": 38870
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.121698933589638e-05,
+      "loss": 2.0595,
+      "step": 38880
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1214700901643096e-05,
+      "loss": 2.299,
+      "step": 38890
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.121241246738982e-05,
+      "loss": 2.1413,
+      "step": 38900
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.121012403313653e-05,
+      "loss": 2.0972,
+      "step": 38910
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1207835598883245e-05,
+      "loss": 2.21,
+      "step": 38920
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.120554716462996e-05,
+      "loss": 2.3894,
+      "step": 38930
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1203258730376674e-05,
+      "loss": 2.1682,
+      "step": 38940
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1200970296123395e-05,
+      "loss": 2.2609,
+      "step": 38950
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.119868186187011e-05,
+      "loss": 2.1922,
+      "step": 38960
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.119639342761682e-05,
+      "loss": 2.2527,
+      "step": 38970
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1194104993363544e-05,
+      "loss": 2.1982,
+      "step": 38980
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.119181655911026e-05,
+      "loss": 2.15,
+      "step": 38990
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.118952812485697e-05,
+      "loss": 2.0954,
+      "step": 39000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.118723969060369e-05,
+      "loss": 2.2751,
+      "step": 39010
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.118495125635041e-05,
+      "loss": 2.2016,
+      "step": 39020
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.118266282209712e-05,
+      "loss": 2.1387,
+      "step": 39030
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.118037438784384e-05,
+      "loss": 2.2879,
+      "step": 39040
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1178085953590556e-05,
+      "loss": 2.1391,
+      "step": 39050
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.117579751933727e-05,
+      "loss": 2.0418,
+      "step": 39060
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.117350908508399e-05,
+      "loss": 2.2904,
+      "step": 39070
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1171220650830705e-05,
+      "loss": 2.188,
+      "step": 39080
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.116893221657742e-05,
+      "loss": 2.06,
+      "step": 39090
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.116664378232414e-05,
+      "loss": 2.0672,
+      "step": 39100
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1164355348070855e-05,
+      "loss": 2.2613,
+      "step": 39110
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.116206691381757e-05,
+      "loss": 2.1939,
+      "step": 39120
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.115977847956428e-05,
+      "loss": 2.2416,
+      "step": 39130
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1157490045311e-05,
+      "loss": 2.2868,
+      "step": 39140
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.115520161105771e-05,
+      "loss": 2.2093,
+      "step": 39150
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.115291317680443e-05,
+      "loss": 2.1679,
+      "step": 39160
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1150624742551146e-05,
+      "loss": 2.0698,
+      "step": 39170
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.114833630829786e-05,
+      "loss": 2.1162,
+      "step": 39180
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.114604787404458e-05,
+      "loss": 2.2055,
+      "step": 39190
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1143759439791295e-05,
+      "loss": 2.1557,
+      "step": 39200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.114147100553801e-05,
+      "loss": 2.2299,
+      "step": 39210
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.113918257128473e-05,
+      "loss": 2.2486,
+      "step": 39220
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1136894137031445e-05,
+      "loss": 2.2951,
+      "step": 39230
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.113460570277816e-05,
+      "loss": 2.1963,
+      "step": 39240
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.113231726852488e-05,
+      "loss": 2.3398,
+      "step": 39250
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1130028834271594e-05,
+      "loss": 2.1408,
+      "step": 39260
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.112774040001831e-05,
+      "loss": 2.2601,
+      "step": 39270
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.112545196576503e-05,
+      "loss": 2.31,
+      "step": 39280
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.112316353151174e-05,
+      "loss": 2.1315,
+      "step": 39290
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.112087509725846e-05,
+      "loss": 2.0875,
+      "step": 39300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.111858666300518e-05,
+      "loss": 2.1373,
+      "step": 39310
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1116298228751885e-05,
+      "loss": 2.1675,
+      "step": 39320
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1114009794498606e-05,
+      "loss": 2.0679,
+      "step": 39330
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.111172136024532e-05,
+      "loss": 2.2478,
+      "step": 39340
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1109432925992035e-05,
+      "loss": 2.1606,
+      "step": 39350
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1107144491738755e-05,
+      "loss": 2.1569,
+      "step": 39360
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.110485605748547e-05,
+      "loss": 2.1405,
+      "step": 39370
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1102567623232184e-05,
+      "loss": 2.0152,
+      "step": 39380
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1100279188978905e-05,
+      "loss": 2.1416,
+      "step": 39390
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.109799075472562e-05,
+      "loss": 2.1287,
+      "step": 39400
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.109570232047233e-05,
+      "loss": 2.257,
+      "step": 39410
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1093413886219054e-05,
+      "loss": 2.3246,
+      "step": 39420
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.109112545196577e-05,
+      "loss": 2.1754,
+      "step": 39430
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.108883701771248e-05,
+      "loss": 2.2396,
+      "step": 39440
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.10865485834592e-05,
+      "loss": 2.1357,
+      "step": 39450
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.108426014920592e-05,
+      "loss": 2.2138,
+      "step": 39460
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.108197171495263e-05,
+      "loss": 2.154,
+      "step": 39470
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.107968328069935e-05,
+      "loss": 2.2399,
+      "step": 39480
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1077394846446066e-05,
+      "loss": 2.2361,
+      "step": 39490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.107510641219278e-05,
+      "loss": 2.256,
+      "step": 39500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.10728179779395e-05,
+      "loss": 2.3559,
+      "step": 39510
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.107052954368621e-05,
+      "loss": 2.1513,
+      "step": 39520
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.106824110943292e-05,
+      "loss": 2.1737,
+      "step": 39530
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1065952675179644e-05,
+      "loss": 2.255,
+      "step": 39540
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.106366424092636e-05,
+      "loss": 2.2463,
+      "step": 39550
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.106137580667307e-05,
+      "loss": 2.1985,
+      "step": 39560
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.105908737241979e-05,
+      "loss": 1.9957,
+      "step": 39570
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.105679893816651e-05,
+      "loss": 2.2229,
+      "step": 39580
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.105451050391322e-05,
+      "loss": 2.2627,
+      "step": 39590
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.105222206965994e-05,
+      "loss": 2.1253,
+      "step": 39600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1049933635406656e-05,
+      "loss": 2.1862,
+      "step": 39610
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.104764520115337e-05,
+      "loss": 2.2142,
+      "step": 39620
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.104535676690009e-05,
+      "loss": 2.241,
+      "step": 39630
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1043068332646805e-05,
+      "loss": 2.3106,
+      "step": 39640
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.104077989839352e-05,
+      "loss": 2.0868,
+      "step": 39650
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.103849146414024e-05,
+      "loss": 2.093,
+      "step": 39660
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1036203029886955e-05,
+      "loss": 2.337,
+      "step": 39670
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.103391459563367e-05,
+      "loss": 2.1856,
+      "step": 39680
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.103162616138039e-05,
+      "loss": 2.2584,
+      "step": 39690
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1029337727127104e-05,
+      "loss": 2.2237,
+      "step": 39700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.102704929287382e-05,
+      "loss": 2.1445,
+      "step": 39710
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.102476085862053e-05,
+      "loss": 2.1212,
+      "step": 39720
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1022472424367246e-05,
+      "loss": 2.3353,
+      "step": 39730
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.102018399011397e-05,
+      "loss": 2.299,
+      "step": 39740
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.101789555586068e-05,
+      "loss": 2.1968,
+      "step": 39750
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1015607121607395e-05,
+      "loss": 2.2367,
+      "step": 39760
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1013318687354116e-05,
+      "loss": 2.1524,
+      "step": 39770
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.101103025310083e-05,
+      "loss": 2.0985,
+      "step": 39780
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1008741818847544e-05,
+      "loss": 2.2321,
+      "step": 39790
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1006453384594265e-05,
+      "loss": 2.2173,
+      "step": 39800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.100416495034098e-05,
+      "loss": 2.1457,
+      "step": 39810
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1001876516087694e-05,
+      "loss": 2.1098,
+      "step": 39820
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0999588081834415e-05,
+      "loss": 2.1585,
+      "step": 39830
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.099729964758113e-05,
+      "loss": 2.1385,
+      "step": 39840
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.099501121332784e-05,
+      "loss": 2.2087,
+      "step": 39850
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0992722779074564e-05,
+      "loss": 2.1661,
+      "step": 39860
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.099043434482128e-05,
+      "loss": 2.3008,
+      "step": 39870
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.098814591056799e-05,
+      "loss": 2.1823,
+      "step": 39880
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.098585747631471e-05,
+      "loss": 2.1454,
+      "step": 39890
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.098356904206143e-05,
+      "loss": 2.4635,
+      "step": 39900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.098128060780814e-05,
+      "loss": 2.1268,
+      "step": 39910
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0978992173554855e-05,
+      "loss": 2.2736,
+      "step": 39920
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.097670373930157e-05,
+      "loss": 2.1329,
+      "step": 39930
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0974415305048284e-05,
+      "loss": 2.1597,
+      "step": 39940
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0972126870795005e-05,
+      "loss": 2.2167,
+      "step": 39950
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.096983843654172e-05,
+      "loss": 2.2999,
+      "step": 39960
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.096755000228843e-05,
+      "loss": 2.1745,
+      "step": 39970
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0965261568035154e-05,
+      "loss": 2.1963,
+      "step": 39980
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.096297313378187e-05,
+      "loss": 2.2926,
+      "step": 39990
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.096068469952858e-05,
+      "loss": 2.2,
+      "step": 40000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.09583962652753e-05,
+      "loss": 2.3226,
+      "step": 40010
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.095610783102202e-05,
+      "loss": 2.0848,
+      "step": 40020
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.095381939676873e-05,
+      "loss": 2.0445,
+      "step": 40030
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.095153096251545e-05,
+      "loss": 2.0948,
+      "step": 40040
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0949242528262166e-05,
+      "loss": 2.3205,
+      "step": 40050
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.094695409400888e-05,
+      "loss": 2.2164,
+      "step": 40060
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.09446656597556e-05,
+      "loss": 2.2814,
+      "step": 40070
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0942377225502315e-05,
+      "loss": 2.0768,
+      "step": 40080
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.094008879124903e-05,
+      "loss": 2.2799,
+      "step": 40090
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.093780035699575e-05,
+      "loss": 2.1764,
+      "step": 40100
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.093551192274246e-05,
+      "loss": 2.2403,
+      "step": 40110
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.093322348848917e-05,
+      "loss": 2.1449,
+      "step": 40120
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.093093505423589e-05,
+      "loss": 2.0456,
+      "step": 40130
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.092864661998261e-05,
+      "loss": 2.2687,
+      "step": 40140
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.092635818572932e-05,
+      "loss": 2.1992,
+      "step": 40150
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.092406975147604e-05,
+      "loss": 2.083,
+      "step": 40160
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0921781317222756e-05,
+      "loss": 2.2419,
+      "step": 40170
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.091949288296947e-05,
+      "loss": 2.376,
+      "step": 40180
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.091720444871619e-05,
+      "loss": 2.2283,
+      "step": 40190
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0914916014462905e-05,
+      "loss": 2.1643,
+      "step": 40200
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.091262758020962e-05,
+      "loss": 2.1368,
+      "step": 40210
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.091033914595634e-05,
+      "loss": 2.1561,
+      "step": 40220
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0908050711703054e-05,
+      "loss": 2.1866,
+      "step": 40230
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.090576227744977e-05,
+      "loss": 2.1337,
+      "step": 40240
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.090347384319649e-05,
+      "loss": 2.3041,
+      "step": 40250
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0901185408943204e-05,
+      "loss": 2.2959,
+      "step": 40260
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.089889697468992e-05,
+      "loss": 2.2023,
+      "step": 40270
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.089660854043664e-05,
+      "loss": 2.2773,
+      "step": 40280
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.089432010618335e-05,
+      "loss": 2.1231,
+      "step": 40290
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.089203167193007e-05,
+      "loss": 2.2303,
+      "step": 40300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.088974323767678e-05,
+      "loss": 2.239,
+      "step": 40310
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0887454803423495e-05,
+      "loss": 2.1627,
+      "step": 40320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0885166369170216e-05,
+      "loss": 2.1551,
+      "step": 40330
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.088287793491693e-05,
+      "loss": 2.2244,
+      "step": 40340
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0880589500663644e-05,
+      "loss": 2.0253,
+      "step": 40350
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0878301066410365e-05,
+      "loss": 2.1681,
+      "step": 40360
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.087601263215708e-05,
+      "loss": 2.0612,
+      "step": 40370
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0873724197903794e-05,
+      "loss": 2.1681,
+      "step": 40380
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0871435763650514e-05,
+      "loss": 2.1546,
+      "step": 40390
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.086914732939723e-05,
+      "loss": 2.2376,
+      "step": 40400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.086685889514394e-05,
+      "loss": 2.0168,
+      "step": 40410
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0864570460890664e-05,
+      "loss": 2.2137,
+      "step": 40420
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.086228202663738e-05,
+      "loss": 2.0738,
+      "step": 40430
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.085999359238409e-05,
+      "loss": 2.2133,
+      "step": 40440
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.085770515813081e-05,
+      "loss": 2.1627,
+      "step": 40450
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.085541672387753e-05,
+      "loss": 2.1162,
+      "step": 40460
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.085312828962424e-05,
+      "loss": 2.2313,
+      "step": 40470
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.085083985537096e-05,
+      "loss": 2.1265,
+      "step": 40480
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0848551421117676e-05,
+      "loss": 2.2138,
+      "step": 40490
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.084626298686439e-05,
+      "loss": 2.171,
+      "step": 40500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0843974552611104e-05,
+      "loss": 2.3195,
+      "step": 40510
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.084168611835782e-05,
+      "loss": 2.1439,
+      "step": 40520
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.083939768410453e-05,
+      "loss": 2.0584,
+      "step": 40530
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0837109249851254e-05,
+      "loss": 1.9797,
+      "step": 40540
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.083482081559797e-05,
+      "loss": 2.0572,
+      "step": 40550
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.083253238134468e-05,
+      "loss": 2.146,
+      "step": 40560
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.08302439470914e-05,
+      "loss": 2.1449,
+      "step": 40570
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.082795551283812e-05,
+      "loss": 2.1155,
+      "step": 40580
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.082566707858483e-05,
+      "loss": 2.0917,
+      "step": 40590
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.082337864433155e-05,
+      "loss": 2.2005,
+      "step": 40600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0821090210078266e-05,
+      "loss": 2.2283,
+      "step": 40610
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.081880177582498e-05,
+      "loss": 2.0801,
+      "step": 40620
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.08165133415717e-05,
+      "loss": 2.1057,
+      "step": 40630
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0814224907318415e-05,
+      "loss": 2.1016,
+      "step": 40640
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.081193647306513e-05,
+      "loss": 2.315,
+      "step": 40650
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.080964803881185e-05,
+      "loss": 2.1776,
+      "step": 40660
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0807359604558564e-05,
+      "loss": 2.2283,
+      "step": 40670
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.080507117030528e-05,
+      "loss": 2.208,
+      "step": 40680
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0802782736052e-05,
+      "loss": 2.1618,
+      "step": 40690
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0800494301798714e-05,
+      "loss": 2.2496,
+      "step": 40700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.079820586754543e-05,
+      "loss": 2.217,
+      "step": 40710
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.079591743329214e-05,
+      "loss": 2.2432,
+      "step": 40720
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0793628999038856e-05,
+      "loss": 2.369,
+      "step": 40730
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.079134056478558e-05,
+      "loss": 2.1811,
+      "step": 40740
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.078905213053229e-05,
+      "loss": 2.2291,
+      "step": 40750
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0786763696279005e-05,
+      "loss": 2.1766,
+      "step": 40760
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0784475262025726e-05,
+      "loss": 2.2116,
+      "step": 40770
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.078218682777244e-05,
+      "loss": 2.1091,
+      "step": 40780
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0779898393519154e-05,
+      "loss": 2.1593,
+      "step": 40790
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0777609959265875e-05,
+      "loss": 2.0119,
+      "step": 40800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.077532152501259e-05,
+      "loss": 2.0798,
+      "step": 40810
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0773033090759304e-05,
+      "loss": 2.159,
+      "step": 40820
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0770744656506024e-05,
+      "loss": 2.0073,
+      "step": 40830
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.076845622225274e-05,
+      "loss": 2.2946,
+      "step": 40840
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.076616778799945e-05,
+      "loss": 2.168,
+      "step": 40850
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0763879353746174e-05,
+      "loss": 2.3139,
+      "step": 40860
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.076159091949289e-05,
+      "loss": 2.194,
+      "step": 40870
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.07593024852396e-05,
+      "loss": 2.1492,
+      "step": 40880
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.075701405098632e-05,
+      "loss": 2.1483,
+      "step": 40890
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.075472561673303e-05,
+      "loss": 2.2923,
+      "step": 40900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0752437182479744e-05,
+      "loss": 1.9377,
+      "step": 40910
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0750148748226465e-05,
+      "loss": 2.2351,
+      "step": 40920
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.074786031397318e-05,
+      "loss": 2.1274,
+      "step": 40930
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0745571879719893e-05,
+      "loss": 2.1766,
+      "step": 40940
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0743283445466614e-05,
+      "loss": 2.1575,
+      "step": 40950
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.074099501121333e-05,
+      "loss": 2.1387,
+      "step": 40960
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.073870657696004e-05,
+      "loss": 2.2917,
+      "step": 40970
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0736418142706764e-05,
+      "loss": 2.0953,
+      "step": 40980
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.073412970845348e-05,
+      "loss": 2.1902,
+      "step": 40990
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.073184127420019e-05,
+      "loss": 2.1813,
+      "step": 41000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.072955283994691e-05,
+      "loss": 2.2685,
+      "step": 41010
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.072726440569363e-05,
+      "loss": 2.0449,
+      "step": 41020
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.072497597144034e-05,
+      "loss": 2.3287,
+      "step": 41030
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.072268753718706e-05,
+      "loss": 2.0608,
+      "step": 41040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0720399102933776e-05,
+      "loss": 2.103,
+      "step": 41050
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.071811066868049e-05,
+      "loss": 2.1679,
+      "step": 41060
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.071582223442721e-05,
+      "loss": 2.0936,
+      "step": 41070
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0713533800173925e-05,
+      "loss": 2.074,
+      "step": 41080
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.071124536592064e-05,
+      "loss": 2.1266,
+      "step": 41090
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0708956931667353e-05,
+      "loss": 2.078,
+      "step": 41100
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.070666849741407e-05,
+      "loss": 2.1907,
+      "step": 41110
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.070438006316079e-05,
+      "loss": 2.209,
+      "step": 41120
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.07020916289075e-05,
+      "loss": 2.1901,
+      "step": 41130
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.069980319465422e-05,
+      "loss": 2.132,
+      "step": 41140
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.069751476040094e-05,
+      "loss": 2.1383,
+      "step": 41150
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.069522632614765e-05,
+      "loss": 2.1652,
+      "step": 41160
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0692937891894366e-05,
+      "loss": 2.0922,
+      "step": 41170
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.069064945764109e-05,
+      "loss": 2.2562,
+      "step": 41180
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.06883610233878e-05,
+      "loss": 2.2509,
+      "step": 41190
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0686072589134515e-05,
+      "loss": 2.1104,
+      "step": 41200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.068378415488123e-05,
+      "loss": 2.1704,
+      "step": 41210
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.068149572062795e-05,
+      "loss": 2.1106,
+      "step": 41220
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0679207286374664e-05,
+      "loss": 2.1684,
+      "step": 41230
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.067691885212138e-05,
+      "loss": 2.0325,
+      "step": 41240
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.06746304178681e-05,
+      "loss": 2.1113,
+      "step": 41250
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0672341983614813e-05,
+      "loss": 2.2521,
+      "step": 41260
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.067005354936153e-05,
+      "loss": 2.3279,
+      "step": 41270
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.066776511510825e-05,
+      "loss": 2.2887,
+      "step": 41280
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.066547668085496e-05,
+      "loss": 2.1898,
+      "step": 41290
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.066318824660168e-05,
+      "loss": 2.0979,
+      "step": 41300
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.066089981234839e-05,
+      "loss": 2.1183,
+      "step": 41310
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0658611378095105e-05,
+      "loss": 2.0995,
+      "step": 41320
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0656322943841826e-05,
+      "loss": 2.0172,
+      "step": 41330
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.065403450958854e-05,
+      "loss": 2.1919,
+      "step": 41340
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0651746075335254e-05,
+      "loss": 2.1032,
+      "step": 41350
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0649457641081975e-05,
+      "loss": 2.1869,
+      "step": 41360
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.064716920682869e-05,
+      "loss": 2.1594,
+      "step": 41370
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0644880772575403e-05,
+      "loss": 2.1721,
+      "step": 41380
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0642592338322124e-05,
+      "loss": 2.2419,
+      "step": 41390
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.064030390406884e-05,
+      "loss": 2.1785,
+      "step": 41400
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.063801546981555e-05,
+      "loss": 2.1928,
+      "step": 41410
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0635727035562274e-05,
+      "loss": 2.2218,
+      "step": 41420
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.063343860130899e-05,
+      "loss": 2.2249,
+      "step": 41430
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.06311501670557e-05,
+      "loss": 2.3241,
+      "step": 41440
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.062886173280242e-05,
+      "loss": 1.9521,
+      "step": 41450
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.062657329854914e-05,
+      "loss": 2.1092,
+      "step": 41460
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.062428486429585e-05,
+      "loss": 1.9185,
+      "step": 41470
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.062199643004257e-05,
+      "loss": 2.2682,
+      "step": 41480
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0619707995789286e-05,
+      "loss": 2.0564,
+      "step": 41490
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.061741956153599e-05,
+      "loss": 2.0998,
+      "step": 41500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0615131127282714e-05,
+      "loss": 2.0906,
+      "step": 41510
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.061284269302943e-05,
+      "loss": 2.1912,
+      "step": 41520
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.061055425877614e-05,
+      "loss": 2.157,
+      "step": 41530
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0608265824522863e-05,
+      "loss": 2.2169,
+      "step": 41540
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.060597739026958e-05,
+      "loss": 2.1296,
+      "step": 41550
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.060368895601629e-05,
+      "loss": 2.1135,
+      "step": 41560
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.060140052176301e-05,
+      "loss": 2.0258,
+      "step": 41570
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.059911208750973e-05,
+      "loss": 2.213,
+      "step": 41580
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.059682365325644e-05,
+      "loss": 2.1745,
+      "step": 41590
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.059453521900316e-05,
+      "loss": 2.2026,
+      "step": 41600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0592246784749876e-05,
+      "loss": 2.2578,
+      "step": 41610
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.058995835049659e-05,
+      "loss": 2.2816,
+      "step": 41620
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.058766991624331e-05,
+      "loss": 2.0974,
+      "step": 41630
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0585381481990025e-05,
+      "loss": 2.2395,
+      "step": 41640
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.058309304773674e-05,
+      "loss": 2.0941,
+      "step": 41650
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.058080461348346e-05,
+      "loss": 1.9998,
+      "step": 41660
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0578516179230174e-05,
+      "loss": 2.2752,
+      "step": 41670
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.057622774497689e-05,
+      "loss": 2.2109,
+      "step": 41680
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.05739393107236e-05,
+      "loss": 2.163,
+      "step": 41690
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.057165087647032e-05,
+      "loss": 2.3022,
+      "step": 41700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.056936244221704e-05,
+      "loss": 2.2563,
+      "step": 41710
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.056707400796375e-05,
+      "loss": 2.3956,
+      "step": 41720
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0564785573710466e-05,
+      "loss": 2.2816,
+      "step": 41730
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.056249713945719e-05,
+      "loss": 2.104,
+      "step": 41740
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.05602087052039e-05,
+      "loss": 2.1239,
+      "step": 41750
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0557920270950615e-05,
+      "loss": 2.2892,
+      "step": 41760
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0555631836697336e-05,
+      "loss": 2.1902,
+      "step": 41770
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.055334340244405e-05,
+      "loss": 2.1027,
+      "step": 41780
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0551054968190764e-05,
+      "loss": 2.2769,
+      "step": 41790
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0548766533937485e-05,
+      "loss": 2.2215,
+      "step": 41800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.05464780996842e-05,
+      "loss": 2.1912,
+      "step": 41810
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0544189665430913e-05,
+      "loss": 2.1736,
+      "step": 41820
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0541901231177634e-05,
+      "loss": 2.1454,
+      "step": 41830
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.053961279692435e-05,
+      "loss": 2.1703,
+      "step": 41840
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.053732436267106e-05,
+      "loss": 1.9324,
+      "step": 41850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0535035928417783e-05,
+      "loss": 2.1962,
+      "step": 41860
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.05327474941645e-05,
+      "loss": 2.296,
+      "step": 41870
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.053045905991121e-05,
+      "loss": 2.2037,
+      "step": 41880
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0528170625657926e-05,
+      "loss": 2.0787,
+      "step": 41890
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.052588219140464e-05,
+      "loss": 2.1846,
+      "step": 41900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0523593757151354e-05,
+      "loss": 2.1403,
+      "step": 41910
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0521305322898075e-05,
+      "loss": 2.2452,
+      "step": 41920
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.051901688864479e-05,
+      "loss": 2.1944,
+      "step": 41930
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.05167284543915e-05,
+      "loss": 2.1472,
+      "step": 41940
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0514440020138224e-05,
+      "loss": 2.2134,
+      "step": 41950
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.051215158588494e-05,
+      "loss": 2.1777,
+      "step": 41960
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.050986315163165e-05,
+      "loss": 2.1252,
+      "step": 41970
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0507574717378373e-05,
+      "loss": 2.2653,
+      "step": 41980
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.050528628312509e-05,
+      "loss": 2.3367,
+      "step": 41990
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.05029978488718e-05,
+      "loss": 2.1017,
+      "step": 42000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.050070941461852e-05,
+      "loss": 2.2148,
+      "step": 42010
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.049842098036524e-05,
+      "loss": 2.1486,
+      "step": 42020
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.049613254611195e-05,
+      "loss": 2.2734,
+      "step": 42030
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.049384411185867e-05,
+      "loss": 2.1611,
+      "step": 42040
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0491555677605386e-05,
+      "loss": 2.0876,
+      "step": 42050
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.04892672433521e-05,
+      "loss": 2.1585,
+      "step": 42060
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.048697880909882e-05,
+      "loss": 2.1434,
+      "step": 42070
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0484690374845535e-05,
+      "loss": 2.224,
+      "step": 42080
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.048240194059225e-05,
+      "loss": 2.1643,
+      "step": 42090
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.048011350633896e-05,
+      "loss": 2.0263,
+      "step": 42100
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.047782507208568e-05,
+      "loss": 2.1302,
+      "step": 42110
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.04755366378324e-05,
+      "loss": 2.1036,
+      "step": 42120
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.047324820357911e-05,
+      "loss": 2.2291,
+      "step": 42130
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.047095976932583e-05,
+      "loss": 2.0645,
+      "step": 42140
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.046867133507255e-05,
+      "loss": 2.0748,
+      "step": 42150
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.046638290081926e-05,
+      "loss": 2.1632,
+      "step": 42160
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0464094466565976e-05,
+      "loss": 2.1606,
+      "step": 42170
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.04618060323127e-05,
+      "loss": 2.1809,
+      "step": 42180
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.045951759805941e-05,
+      "loss": 2.2701,
+      "step": 42190
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0457229163806125e-05,
+      "loss": 2.2346,
+      "step": 42200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0454940729552846e-05,
+      "loss": 2.0805,
+      "step": 42210
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.045265229529956e-05,
+      "loss": 2.244,
+      "step": 42220
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0450363861046274e-05,
+      "loss": 2.2349,
+      "step": 42230
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0448075426792995e-05,
+      "loss": 2.15,
+      "step": 42240
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.044578699253971e-05,
+      "loss": 2.1146,
+      "step": 42250
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.044349855828642e-05,
+      "loss": 2.1266,
+      "step": 42260
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0441210124033144e-05,
+      "loss": 2.1063,
+      "step": 42270
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.043892168977986e-05,
+      "loss": 2.1298,
+      "step": 42280
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0436633255526566e-05,
+      "loss": 2.0812,
+      "step": 42290
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.043434482127329e-05,
+      "loss": 2.3123,
+      "step": 42300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.043205638702e-05,
+      "loss": 2.2748,
+      "step": 42310
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0429767952766715e-05,
+      "loss": 2.201,
+      "step": 42320
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0427479518513436e-05,
+      "loss": 2.0585,
+      "step": 42330
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.042519108426015e-05,
+      "loss": 2.1917,
+      "step": 42340
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0422902650006864e-05,
+      "loss": 2.1519,
+      "step": 42350
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0420614215753585e-05,
+      "loss": 2.1532,
+      "step": 42360
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.04183257815003e-05,
+      "loss": 2.2369,
+      "step": 42370
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.041603734724701e-05,
+      "loss": 2.0533,
+      "step": 42380
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0413748912993734e-05,
+      "loss": 2.0653,
+      "step": 42390
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.041146047874045e-05,
+      "loss": 2.1549,
+      "step": 42400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.040917204448716e-05,
+      "loss": 2.1142,
+      "step": 42410
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.040688361023388e-05,
+      "loss": 2.0491,
+      "step": 42420
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.04045951759806e-05,
+      "loss": 2.1749,
+      "step": 42430
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.040230674172731e-05,
+      "loss": 2.2594,
+      "step": 42440
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.040001830747403e-05,
+      "loss": 2.1713,
+      "step": 42450
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.039772987322075e-05,
+      "loss": 2.1126,
+      "step": 42460
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.039544143896746e-05,
+      "loss": 2.1659,
+      "step": 42470
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0393153004714175e-05,
+      "loss": 2.0997,
+      "step": 42480
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.039086457046089e-05,
+      "loss": 2.1626,
+      "step": 42490
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.03885761362076e-05,
+      "loss": 2.1459,
+      "step": 42500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0386287701954324e-05,
+      "loss": 2.2279,
+      "step": 42510
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.038399926770104e-05,
+      "loss": 2.2609,
+      "step": 42520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.038171083344775e-05,
+      "loss": 1.9888,
+      "step": 42530
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.037942239919447e-05,
+      "loss": 2.101,
+      "step": 42540
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.037713396494119e-05,
+      "loss": 2.0828,
+      "step": 42550
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.03748455306879e-05,
+      "loss": 2.3087,
+      "step": 42560
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.037255709643462e-05,
+      "loss": 2.0694,
+      "step": 42570
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.037026866218134e-05,
+      "loss": 2.1631,
+      "step": 42580
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.036798022792805e-05,
+      "loss": 2.1853,
+      "step": 42590
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.036569179367477e-05,
+      "loss": 2.329,
+      "step": 42600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0363403359421486e-05,
+      "loss": 2.0256,
+      "step": 42610
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.03611149251682e-05,
+      "loss": 2.1325,
+      "step": 42620
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.035882649091492e-05,
+      "loss": 2.2292,
+      "step": 42630
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0356538056661635e-05,
+      "loss": 2.276,
+      "step": 42640
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.035424962240835e-05,
+      "loss": 2.0988,
+      "step": 42650
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.035196118815507e-05,
+      "loss": 2.1693,
+      "step": 42660
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0349672753901784e-05,
+      "loss": 2.0843,
+      "step": 42670
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.03473843196485e-05,
+      "loss": 2.0869,
+      "step": 42680
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.034509588539521e-05,
+      "loss": 2.3715,
+      "step": 42690
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0342807451141927e-05,
+      "loss": 2.2155,
+      "step": 42700
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.034051901688865e-05,
+      "loss": 2.0673,
+      "step": 42710
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.033823058263536e-05,
+      "loss": 2.247,
+      "step": 42720
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0335942148382076e-05,
+      "loss": 2.2714,
+      "step": 42730
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.03336537141288e-05,
+      "loss": 2.2215,
+      "step": 42740
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.033136527987551e-05,
+      "loss": 2.2548,
+      "step": 42750
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0329076845622225e-05,
+      "loss": 2.1048,
+      "step": 42760
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0326788411368946e-05,
+      "loss": 2.1272,
+      "step": 42770
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.032449997711566e-05,
+      "loss": 2.2124,
+      "step": 42780
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0322211542862374e-05,
+      "loss": 2.3664,
+      "step": 42790
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0319923108609095e-05,
+      "loss": 2.035,
+      "step": 42800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.031763467435581e-05,
+      "loss": 2.3679,
+      "step": 42810
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.031534624010252e-05,
+      "loss": 2.1594,
+      "step": 42820
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0313057805849244e-05,
+      "loss": 2.1219,
+      "step": 42830
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.031076937159596e-05,
+      "loss": 2.0039,
+      "step": 42840
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.030848093734267e-05,
+      "loss": 2.0945,
+      "step": 42850
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.030619250308939e-05,
+      "loss": 2.1949,
+      "step": 42860
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.030390406883611e-05,
+      "loss": 2.1756,
+      "step": 42870
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0301615634582815e-05,
+      "loss": 2.077,
+      "step": 42880
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0299327200329536e-05,
+      "loss": 2.1333,
+      "step": 42890
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.029703876607625e-05,
+      "loss": 2.0794,
+      "step": 42900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0294750331822964e-05,
+      "loss": 2.1696,
+      "step": 42910
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0292461897569685e-05,
+      "loss": 2.1866,
+      "step": 42920
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.02901734633164e-05,
+      "loss": 1.9593,
+      "step": 42930
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.028788502906311e-05,
+      "loss": 2.1689,
+      "step": 42940
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0285596594809834e-05,
+      "loss": 2.1732,
+      "step": 42950
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.028330816055655e-05,
+      "loss": 2.0622,
+      "step": 42960
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.028101972630326e-05,
+      "loss": 2.1236,
+      "step": 42970
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.027873129204998e-05,
+      "loss": 2.0977,
+      "step": 42980
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.02764428577967e-05,
+      "loss": 2.1726,
+      "step": 42990
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.027415442354341e-05,
+      "loss": 2.285,
+      "step": 43000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.027186598929013e-05,
+      "loss": 2.1645,
+      "step": 43010
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0269577555036847e-05,
+      "loss": 1.9987,
+      "step": 43020
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.026728912078356e-05,
+      "loss": 2.1143,
+      "step": 43030
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.026500068653028e-05,
+      "loss": 2.2973,
+      "step": 43040
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0262712252276996e-05,
+      "loss": 2.1446,
+      "step": 43050
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.026042381802371e-05,
+      "loss": 2.1876,
+      "step": 43060
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.025813538377043e-05,
+      "loss": 2.1576,
+      "step": 43070
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.025584694951714e-05,
+      "loss": 2.1068,
+      "step": 43080
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.025355851526386e-05,
+      "loss": 2.096,
+      "step": 43090
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.025127008101057e-05,
+      "loss": 2.1522,
+      "step": 43100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.024898164675729e-05,
+      "loss": 2.2757,
+      "step": 43110
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.024669321250401e-05,
+      "loss": 2.114,
+      "step": 43120
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.024440477825072e-05,
+      "loss": 2.2968,
+      "step": 43130
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0242116343997437e-05,
+      "loss": 2.1175,
+      "step": 43140
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.023982790974416e-05,
+      "loss": 2.2303,
+      "step": 43150
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.023753947549087e-05,
+      "loss": 2.05,
+      "step": 43160
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0235251041237586e-05,
+      "loss": 2.2767,
+      "step": 43170
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0232962606984307e-05,
+      "loss": 2.1456,
+      "step": 43180
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.023067417273102e-05,
+      "loss": 2.2191,
+      "step": 43190
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0228385738477735e-05,
+      "loss": 2.1719,
+      "step": 43200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0226097304224456e-05,
+      "loss": 2.2707,
+      "step": 43210
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.022380886997117e-05,
+      "loss": 2.1729,
+      "step": 43220
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0221520435717884e-05,
+      "loss": 2.2156,
+      "step": 43230
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0219232001464605e-05,
+      "loss": 2.1239,
+      "step": 43240
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.021694356721132e-05,
+      "loss": 2.3339,
+      "step": 43250
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.021465513295803e-05,
+      "loss": 2.1526,
+      "step": 43260
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.021236669870475e-05,
+      "loss": 2.1376,
+      "step": 43270
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.021007826445146e-05,
+      "loss": 2.2015,
+      "step": 43280
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0207789830198176e-05,
+      "loss": 2.0855,
+      "step": 43290
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0205501395944897e-05,
+      "loss": 2.0382,
+      "step": 43300
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.020321296169161e-05,
+      "loss": 2.1346,
+      "step": 43310
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0200924527438325e-05,
+      "loss": 2.1081,
+      "step": 43320
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0198636093185046e-05,
+      "loss": 2.1694,
+      "step": 43330
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.019634765893176e-05,
+      "loss": 2.2509,
+      "step": 43340
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0194059224678474e-05,
+      "loss": 2.1679,
+      "step": 43350
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0191770790425195e-05,
+      "loss": 2.2324,
+      "step": 43360
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.018948235617191e-05,
+      "loss": 2.2324,
+      "step": 43370
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.018719392191862e-05,
+      "loss": 2.0115,
+      "step": 43380
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0184905487665344e-05,
+      "loss": 2.2971,
+      "step": 43390
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.018261705341206e-05,
+      "loss": 2.0845,
+      "step": 43400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.018032861915877e-05,
+      "loss": 2.3713,
+      "step": 43410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.017804018490549e-05,
+      "loss": 2.3598,
+      "step": 43420
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.017575175065221e-05,
+      "loss": 1.9928,
+      "step": 43430
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.017346331639892e-05,
+      "loss": 2.1497,
+      "step": 43440
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.017117488214564e-05,
+      "loss": 2.2359,
+      "step": 43450
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0168886447892357e-05,
+      "loss": 2.1742,
+      "step": 43460
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.016659801363907e-05,
+      "loss": 2.1448,
+      "step": 43470
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0164309579385785e-05,
+      "loss": 2.1013,
+      "step": 43480
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.01620211451325e-05,
+      "loss": 2.1386,
+      "step": 43490
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.015973271087922e-05,
+      "loss": 2.1619,
+      "step": 43500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0157444276625934e-05,
+      "loss": 2.0753,
+      "step": 43510
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.015515584237265e-05,
+      "loss": 2.1612,
+      "step": 43520
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.015286740811936e-05,
+      "loss": 2.2096,
+      "step": 43530
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.015057897386608e-05,
+      "loss": 2.1334,
+      "step": 43540
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.01482905396128e-05,
+      "loss": 2.2236,
+      "step": 43550
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.014600210535951e-05,
+      "loss": 2.2397,
+      "step": 43560
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.014371367110623e-05,
+      "loss": 2.0609,
+      "step": 43570
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0141425236852946e-05,
+      "loss": 2.1561,
+      "step": 43580
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.013913680259966e-05,
+      "loss": 2.2108,
+      "step": 43590
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.013684836834638e-05,
+      "loss": 2.0336,
+      "step": 43600
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0134559934093096e-05,
+      "loss": 2.1386,
+      "step": 43610
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.013227149983981e-05,
+      "loss": 2.0658,
+      "step": 43620
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.012998306558653e-05,
+      "loss": 2.0606,
+      "step": 43630
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0127694631333245e-05,
+      "loss": 2.1555,
+      "step": 43640
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.012540619707996e-05,
+      "loss": 2.2199,
+      "step": 43650
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.012311776282668e-05,
+      "loss": 2.0866,
+      "step": 43660
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.012082932857339e-05,
+      "loss": 2.2384,
+      "step": 43670
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.011854089432011e-05,
+      "loss": 2.2946,
+      "step": 43680
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.011625246006682e-05,
+      "loss": 2.1082,
+      "step": 43690
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0113964025813536e-05,
+      "loss": 2.2524,
+      "step": 43700
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.011167559156026e-05,
+      "loss": 2.2067,
+      "step": 43710
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.010938715730697e-05,
+      "loss": 2.1595,
+      "step": 43720
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0107098723053686e-05,
+      "loss": 2.084,
+      "step": 43730
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0104810288800407e-05,
+      "loss": 2.1358,
+      "step": 43740
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.010252185454712e-05,
+      "loss": 2.082,
+      "step": 43750
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0100233420293835e-05,
+      "loss": 2.3063,
+      "step": 43760
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0097944986040556e-05,
+      "loss": 2.1795,
+      "step": 43770
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.009565655178727e-05,
+      "loss": 2.0539,
+      "step": 43780
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0093368117533984e-05,
+      "loss": 2.1856,
+      "step": 43790
+    },
+    {
+      "epoch": 2.0,
+      "eval_cer": 0.6803315908802483,
+      "eval_em": 0.00728476821192053,
+      "eval_f1": 0.00728476821192053,
+      "eval_loss": 2.0877115726470947,
+      "eval_runtime": 2655.7777,
+      "eval_samples_per_second": 3.98,
+      "eval_steps_per_second": 1.99,
+      "eval_wer": 0.9927152317880795,
+      "step": 43799
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0091079683280705e-05,
+      "loss": 2.068,
+      "step": 43800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.008879124902742e-05,
+      "loss": 2.0781,
+      "step": 43810
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.008650281477413e-05,
+      "loss": 2.0973,
+      "step": 43820
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0084214380520854e-05,
+      "loss": 2.2002,
+      "step": 43830
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.008192594626757e-05,
+      "loss": 2.2435,
+      "step": 43840
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.007963751201428e-05,
+      "loss": 2.1009,
+      "step": 43850
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0077349077761e-05,
+      "loss": 2.3265,
+      "step": 43860
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.007506064350771e-05,
+      "loss": 2.1283,
+      "step": 43870
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0072772209254425e-05,
+      "loss": 2.1477,
+      "step": 43880
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0070483775001146e-05,
+      "loss": 2.2385,
+      "step": 43890
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.006819534074786e-05,
+      "loss": 2.1562,
+      "step": 43900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0065906906494574e-05,
+      "loss": 2.2254,
+      "step": 43910
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0063618472241295e-05,
+      "loss": 2.2111,
+      "step": 43920
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.006133003798801e-05,
+      "loss": 2.0934,
+      "step": 43930
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.005904160373472e-05,
+      "loss": 2.0573,
+      "step": 43940
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0056753169481444e-05,
+      "loss": 2.1335,
+      "step": 43950
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.005446473522816e-05,
+      "loss": 2.2072,
+      "step": 43960
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.005217630097487e-05,
+      "loss": 2.1794,
+      "step": 43970
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.004988786672159e-05,
+      "loss": 1.9912,
+      "step": 43980
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.004759943246831e-05,
+      "loss": 1.9587,
+      "step": 43990
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.004531099821502e-05,
+      "loss": 2.1115,
+      "step": 44000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.004302256396174e-05,
+      "loss": 2.312,
+      "step": 44010
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0040734129708456e-05,
+      "loss": 2.1002,
+      "step": 44020
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.003844569545517e-05,
+      "loss": 2.0632,
+      "step": 44030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.003615726120189e-05,
+      "loss": 2.3169,
+      "step": 44040
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0033868826948606e-05,
+      "loss": 2.1994,
+      "step": 44050
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.003158039269532e-05,
+      "loss": 2.1358,
+      "step": 44060
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0029291958442034e-05,
+      "loss": 1.958,
+      "step": 44070
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.002700352418875e-05,
+      "loss": 2.1652,
+      "step": 44080
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.002471508993547e-05,
+      "loss": 2.1721,
+      "step": 44090
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.002242665568218e-05,
+      "loss": 2.1231,
+      "step": 44100
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.00201382214289e-05,
+      "loss": 2.0832,
+      "step": 44110
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.001784978717562e-05,
+      "loss": 2.1321,
+      "step": 44120
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.001556135292233e-05,
+      "loss": 2.2346,
+      "step": 44130
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.0013272918669046e-05,
+      "loss": 2.1374,
+      "step": 44140
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.001098448441577e-05,
+      "loss": 2.049,
+      "step": 44150
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.000869605016248e-05,
+      "loss": 2.1991,
+      "step": 44160
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.0006407615909196e-05,
+      "loss": 2.2057,
+      "step": 44170
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.0004119181655916e-05,
+      "loss": 2.1019,
+      "step": 44180
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.000183074740263e-05,
+      "loss": 2.1055,
+      "step": 44190
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9999542313149345e-05,
+      "loss": 2.1823,
+      "step": 44200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9997253878896066e-05,
+      "loss": 2.346,
+      "step": 44210
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.999496544464278e-05,
+      "loss": 2.0932,
+      "step": 44220
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9992677010389494e-05,
+      "loss": 2.1734,
+      "step": 44230
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9990388576136215e-05,
+      "loss": 2.1016,
+      "step": 44240
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.998810014188293e-05,
+      "loss": 2.0696,
+      "step": 44250
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9985811707629636e-05,
+      "loss": 2.19,
+      "step": 44260
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.998352327337636e-05,
+      "loss": 1.9556,
+      "step": 44270
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.998123483912307e-05,
+      "loss": 2.1191,
+      "step": 44280
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9978946404869785e-05,
+      "loss": 2.1409,
+      "step": 44290
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9976657970616506e-05,
+      "loss": 2.1453,
+      "step": 44300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.997436953636322e-05,
+      "loss": 2.2592,
+      "step": 44310
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9972081102109935e-05,
+      "loss": 1.9952,
+      "step": 44320
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9969792667856656e-05,
+      "loss": 2.1462,
+      "step": 44330
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.996750423360337e-05,
+      "loss": 2.1939,
+      "step": 44340
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9965215799350084e-05,
+      "loss": 2.1064,
+      "step": 44350
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9962927365096805e-05,
+      "loss": 2.1621,
+      "step": 44360
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.996063893084352e-05,
+      "loss": 2.2163,
+      "step": 44370
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.995835049659023e-05,
+      "loss": 2.0581,
+      "step": 44380
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9956062062336954e-05,
+      "loss": 2.075,
+      "step": 44390
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.995377362808367e-05,
+      "loss": 2.0796,
+      "step": 44400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.995148519383038e-05,
+      "loss": 2.0189,
+      "step": 44410
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.99491967595771e-05,
+      "loss": 1.9997,
+      "step": 44420
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.994690832532382e-05,
+      "loss": 2.3219,
+      "step": 44430
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.994461989107053e-05,
+      "loss": 2.0995,
+      "step": 44440
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.994233145681725e-05,
+      "loss": 2.2545,
+      "step": 44450
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.994004302256396e-05,
+      "loss": 2.2726,
+      "step": 44460
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.993775458831068e-05,
+      "loss": 2.3233,
+      "step": 44470
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9935466154057395e-05,
+      "loss": 2.1245,
+      "step": 44480
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.993317771980411e-05,
+      "loss": 2.0066,
+      "step": 44490
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.993088928555083e-05,
+      "loss": 2.1064,
+      "step": 44500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9928600851297544e-05,
+      "loss": 2.1372,
+      "step": 44510
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.992631241704426e-05,
+      "loss": 2.2571,
+      "step": 44520
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.992402398279098e-05,
+      "loss": 2.0638,
+      "step": 44530
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.992173554853769e-05,
+      "loss": 2.1493,
+      "step": 44540
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.991944711428441e-05,
+      "loss": 2.0984,
+      "step": 44550
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.991715868003113e-05,
+      "loss": 2.1422,
+      "step": 44560
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.991487024577784e-05,
+      "loss": 2.1596,
+      "step": 44570
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9912581811524556e-05,
+      "loss": 2.3457,
+      "step": 44580
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.991029337727128e-05,
+      "loss": 2.1048,
+      "step": 44590
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.990800494301799e-05,
+      "loss": 2.0255,
+      "step": 44600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9905716508764706e-05,
+      "loss": 2.1514,
+      "step": 44610
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9903428074511426e-05,
+      "loss": 2.0737,
+      "step": 44620
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.990113964025814e-05,
+      "loss": 2.1085,
+      "step": 44630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9898851206004855e-05,
+      "loss": 1.9769,
+      "step": 44640
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.989656277175157e-05,
+      "loss": 2.0106,
+      "step": 44650
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.989427433749828e-05,
+      "loss": 2.1209,
+      "step": 44660
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9891985903245e-05,
+      "loss": 2.0847,
+      "step": 44670
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.988969746899172e-05,
+      "loss": 2.157,
+      "step": 44680
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.988740903473843e-05,
+      "loss": 2.1885,
+      "step": 44690
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9885120600485146e-05,
+      "loss": 2.0697,
+      "step": 44700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.988283216623187e-05,
+      "loss": 2.1716,
+      "step": 44710
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.988054373197858e-05,
+      "loss": 2.1129,
+      "step": 44720
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9878255297725295e-05,
+      "loss": 2.0408,
+      "step": 44730
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9875966863472016e-05,
+      "loss": 1.997,
+      "step": 44740
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.987367842921873e-05,
+      "loss": 2.0894,
+      "step": 44750
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9871389994965445e-05,
+      "loss": 2.144,
+      "step": 44760
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9869101560712166e-05,
+      "loss": 2.0767,
+      "step": 44770
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.986681312645888e-05,
+      "loss": 2.1087,
+      "step": 44780
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9864524692205594e-05,
+      "loss": 2.1549,
+      "step": 44790
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9862236257952315e-05,
+      "loss": 2.2952,
+      "step": 44800
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.985994782369903e-05,
+      "loss": 2.1928,
+      "step": 44810
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.985765938944574e-05,
+      "loss": 2.1544,
+      "step": 44820
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9855370955192464e-05,
+      "loss": 2.1965,
+      "step": 44830
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.985308252093918e-05,
+      "loss": 2.059,
+      "step": 44840
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9850794086685885e-05,
+      "loss": 2.176,
+      "step": 44850
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9848505652432606e-05,
+      "loss": 2.0571,
+      "step": 44860
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.984621721817932e-05,
+      "loss": 2.1609,
+      "step": 44870
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9843928783926035e-05,
+      "loss": 2.2207,
+      "step": 44880
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9841640349672755e-05,
+      "loss": 2.3255,
+      "step": 44890
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.983935191541947e-05,
+      "loss": 2.0508,
+      "step": 44900
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9837063481166184e-05,
+      "loss": 2.096,
+      "step": 44910
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9834775046912905e-05,
+      "loss": 2.2331,
+      "step": 44920
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.983248661265962e-05,
+      "loss": 2.2008,
+      "step": 44930
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.983019817840633e-05,
+      "loss": 2.2905,
+      "step": 44940
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9827909744153054e-05,
+      "loss": 2.0257,
+      "step": 44950
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.982562130989977e-05,
+      "loss": 2.3328,
+      "step": 44960
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.982333287564648e-05,
+      "loss": 2.1619,
+      "step": 44970
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.98210444413932e-05,
+      "loss": 2.2712,
+      "step": 44980
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.981875600713992e-05,
+      "loss": 2.142,
+      "step": 44990
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.981646757288663e-05,
+      "loss": 2.2381,
+      "step": 45000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.981417913863335e-05,
+      "loss": 2.1213,
+      "step": 45010
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9811890704380066e-05,
+      "loss": 2.0988,
+      "step": 45020
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.980960227012678e-05,
+      "loss": 2.1052,
+      "step": 45030
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.98073138358735e-05,
+      "loss": 2.1316,
+      "step": 45040
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.980502540162021e-05,
+      "loss": 2.1439,
+      "step": 45050
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.980273696736693e-05,
+      "loss": 2.1584,
+      "step": 45060
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9800448533113644e-05,
+      "loss": 2.2457,
+      "step": 45070
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.979816009886036e-05,
+      "loss": 2.1363,
+      "step": 45080
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.979587166460708e-05,
+      "loss": 1.9597,
+      "step": 45090
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.979358323035379e-05,
+      "loss": 1.9604,
+      "step": 45100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.979129479610051e-05,
+      "loss": 2.2077,
+      "step": 45110
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.978900636184723e-05,
+      "loss": 2.1129,
+      "step": 45120
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.978671792759394e-05,
+      "loss": 2.2119,
+      "step": 45130
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9784429493340656e-05,
+      "loss": 2.148,
+      "step": 45140
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.978214105908738e-05,
+      "loss": 2.0753,
+      "step": 45150
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.977985262483409e-05,
+      "loss": 2.1517,
+      "step": 45160
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9777564190580805e-05,
+      "loss": 2.2632,
+      "step": 45170
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9775275756327526e-05,
+      "loss": 2.0414,
+      "step": 45180
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.977298732207424e-05,
+      "loss": 2.1213,
+      "step": 45190
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9770698887820955e-05,
+      "loss": 2.0444,
+      "step": 45200
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9768410453567676e-05,
+      "loss": 2.2012,
+      "step": 45210
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.976612201931439e-05,
+      "loss": 2.0186,
+      "step": 45220
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9763833585061104e-05,
+      "loss": 2.1593,
+      "step": 45230
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9761545150807825e-05,
+      "loss": 1.9909,
+      "step": 45240
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.975925671655453e-05,
+      "loss": 2.1351,
+      "step": 45250
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9756968282301246e-05,
+      "loss": 2.078,
+      "step": 45260
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.975467984804797e-05,
+      "loss": 2.1609,
+      "step": 45270
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.975239141379468e-05,
+      "loss": 1.9537,
+      "step": 45280
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9750102979541395e-05,
+      "loss": 2.1635,
+      "step": 45290
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9747814545288116e-05,
+      "loss": 2.2814,
+      "step": 45300
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.974552611103483e-05,
+      "loss": 2.264,
+      "step": 45310
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9743237676781545e-05,
+      "loss": 2.1667,
+      "step": 45320
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9740949242528265e-05,
+      "loss": 2.1343,
+      "step": 45330
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.973866080827498e-05,
+      "loss": 2.2072,
+      "step": 45340
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9736372374021694e-05,
+      "loss": 2.1273,
+      "step": 45350
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9734083939768415e-05,
+      "loss": 2.3258,
+      "step": 45360
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.973179550551513e-05,
+      "loss": 2.0087,
+      "step": 45370
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.972950707126184e-05,
+      "loss": 2.1923,
+      "step": 45380
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9727218637008564e-05,
+      "loss": 2.049,
+      "step": 45390
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.972493020275528e-05,
+      "loss": 2.2572,
+      "step": 45400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.972264176850199e-05,
+      "loss": 2.1778,
+      "step": 45410
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.972035333424871e-05,
+      "loss": 2.1429,
+      "step": 45420
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.971806489999543e-05,
+      "loss": 2.0242,
+      "step": 45430
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.971577646574214e-05,
+      "loss": 2.0323,
+      "step": 45440
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9713488031488855e-05,
+      "loss": 2.0182,
+      "step": 45450
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.971119959723557e-05,
+      "loss": 2.0533,
+      "step": 45460
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.970891116298229e-05,
+      "loss": 2.1162,
+      "step": 45470
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9706622728729005e-05,
+      "loss": 2.2697,
+      "step": 45480
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.970433429447572e-05,
+      "loss": 2.1407,
+      "step": 45490
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.970204586022244e-05,
+      "loss": 2.129,
+      "step": 45500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9699757425969154e-05,
+      "loss": 2.1573,
+      "step": 45510
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.969746899171587e-05,
+      "loss": 2.1791,
+      "step": 45520
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.969518055746259e-05,
+      "loss": 2.0688,
+      "step": 45530
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.96928921232093e-05,
+      "loss": 2.1886,
+      "step": 45540
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.969060368895602e-05,
+      "loss": 2.1915,
+      "step": 45550
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.968831525470274e-05,
+      "loss": 2.1243,
+      "step": 45560
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.968602682044945e-05,
+      "loss": 2.1967,
+      "step": 45570
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9683738386196166e-05,
+      "loss": 2.2301,
+      "step": 45580
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.968144995194289e-05,
+      "loss": 1.9758,
+      "step": 45590
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.96791615176896e-05,
+      "loss": 2.279,
+      "step": 45600
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9676873083436315e-05,
+      "loss": 2.1666,
+      "step": 45610
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9674584649183036e-05,
+      "loss": 2.3857,
+      "step": 45620
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.967229621492975e-05,
+      "loss": 2.1314,
+      "step": 45630
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9670007780676465e-05,
+      "loss": 2.2925,
+      "step": 45640
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.966771934642318e-05,
+      "loss": 2.1168,
+      "step": 45650
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.966543091216989e-05,
+      "loss": 2.1833,
+      "step": 45660
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.966314247791661e-05,
+      "loss": 2.057,
+      "step": 45670
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.966085404366333e-05,
+      "loss": 2.1769,
+      "step": 45680
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.965856560941004e-05,
+      "loss": 2.0007,
+      "step": 45690
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9656277175156756e-05,
+      "loss": 2.0927,
+      "step": 45700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.965398874090348e-05,
+      "loss": 2.0708,
+      "step": 45710
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.965170030665019e-05,
+      "loss": 2.2017,
+      "step": 45720
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9649411872396905e-05,
+      "loss": 2.1507,
+      "step": 45730
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9647123438143626e-05,
+      "loss": 2.078,
+      "step": 45740
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.964483500389034e-05,
+      "loss": 2.1122,
+      "step": 45750
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9642546569637055e-05,
+      "loss": 2.2172,
+      "step": 45760
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9640258135383775e-05,
+      "loss": 2.039,
+      "step": 45770
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.963796970113049e-05,
+      "loss": 2.1516,
+      "step": 45780
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9635681266877204e-05,
+      "loss": 2.2883,
+      "step": 45790
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9633392832623925e-05,
+      "loss": 2.0931,
+      "step": 45800
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.963110439837064e-05,
+      "loss": 2.2002,
+      "step": 45810
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.962881596411735e-05,
+      "loss": 2.223,
+      "step": 45820
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9626527529864074e-05,
+      "loss": 2.1603,
+      "step": 45830
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.962423909561078e-05,
+      "loss": 2.0405,
+      "step": 45840
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.96219506613575e-05,
+      "loss": 2.1259,
+      "step": 45850
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9619662227104216e-05,
+      "loss": 2.1603,
+      "step": 45860
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.961737379285093e-05,
+      "loss": 2.0888,
+      "step": 45870
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9615085358597644e-05,
+      "loss": 2.036,
+      "step": 45880
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9612796924344365e-05,
+      "loss": 2.0684,
+      "step": 45890
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.961050849009108e-05,
+      "loss": 2.1832,
+      "step": 45900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9608220055837794e-05,
+      "loss": 2.1936,
+      "step": 45910
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9605931621584515e-05,
+      "loss": 2.2078,
+      "step": 45920
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.960364318733123e-05,
+      "loss": 2.0127,
+      "step": 45930
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.960135475307794e-05,
+      "loss": 2.1972,
+      "step": 45940
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9599066318824664e-05,
+      "loss": 2.1599,
+      "step": 45950
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.959677788457138e-05,
+      "loss": 2.0825,
+      "step": 45960
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.959448945031809e-05,
+      "loss": 2.1752,
+      "step": 45970
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.959220101606481e-05,
+      "loss": 2.0778,
+      "step": 45980
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.958991258181153e-05,
+      "loss": 2.1243,
+      "step": 45990
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.958762414755824e-05,
+      "loss": 2.0431,
+      "step": 46000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.958533571330496e-05,
+      "loss": 2.088,
+      "step": 46010
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9583047279051676e-05,
+      "loss": 2.0745,
+      "step": 46020
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.958075884479839e-05,
+      "loss": 2.0858,
+      "step": 46030
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9578470410545104e-05,
+      "loss": 2.1764,
+      "step": 46040
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.957618197629182e-05,
+      "loss": 2.2337,
+      "step": 46050
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.957389354203854e-05,
+      "loss": 2.0471,
+      "step": 46060
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9571605107785254e-05,
+      "loss": 1.9367,
+      "step": 46070
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.956931667353197e-05,
+      "loss": 2.0568,
+      "step": 46080
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.956702823927869e-05,
+      "loss": 2.0782,
+      "step": 46090
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.95647398050254e-05,
+      "loss": 1.9797,
+      "step": 46100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.956245137077212e-05,
+      "loss": 2.0825,
+      "step": 46110
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.956016293651884e-05,
+      "loss": 2.1537,
+      "step": 46120
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.955787450226555e-05,
+      "loss": 2.2567,
+      "step": 46130
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9555586068012266e-05,
+      "loss": 2.299,
+      "step": 46140
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.955329763375899e-05,
+      "loss": 2.0935,
+      "step": 46150
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.95510091995057e-05,
+      "loss": 2.148,
+      "step": 46160
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9548720765252415e-05,
+      "loss": 2.2409,
+      "step": 46170
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9546432330999136e-05,
+      "loss": 2.1539,
+      "step": 46180
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.954414389674585e-05,
+      "loss": 2.0304,
+      "step": 46190
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9541855462492564e-05,
+      "loss": 2.0006,
+      "step": 46200
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9539567028239285e-05,
+      "loss": 2.1048,
+      "step": 46210
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9537278593986e-05,
+      "loss": 2.1856,
+      "step": 46220
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9534990159732714e-05,
+      "loss": 2.1029,
+      "step": 46230
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.953270172547943e-05,
+      "loss": 2.2018,
+      "step": 46240
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.953041329122614e-05,
+      "loss": 2.2724,
+      "step": 46250
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9528124856972856e-05,
+      "loss": 1.8865,
+      "step": 46260
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.952583642271958e-05,
+      "loss": 2.1468,
+      "step": 46270
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.952354798846629e-05,
+      "loss": 2.2495,
+      "step": 46280
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9521259554213005e-05,
+      "loss": 2.1469,
+      "step": 46290
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9518971119959726e-05,
+      "loss": 2.0946,
+      "step": 46300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.951668268570644e-05,
+      "loss": 2.1642,
+      "step": 46310
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9514394251453154e-05,
+      "loss": 2.122,
+      "step": 46320
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9512105817199875e-05,
+      "loss": 2.1853,
+      "step": 46330
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.950981738294659e-05,
+      "loss": 2.1397,
+      "step": 46340
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9507528948693304e-05,
+      "loss": 1.9184,
+      "step": 46350
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9505240514440024e-05,
+      "loss": 2.2584,
+      "step": 46360
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.950295208018674e-05,
+      "loss": 1.9638,
+      "step": 46370
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.950066364593345e-05,
+      "loss": 2.1507,
+      "step": 46380
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9498375211680174e-05,
+      "loss": 2.111,
+      "step": 46390
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.949608677742689e-05,
+      "loss": 2.0705,
+      "step": 46400
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.94937983431736e-05,
+      "loss": 2.2116,
+      "step": 46410
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.949150990892032e-05,
+      "loss": 2.1783,
+      "step": 46420
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.948922147466704e-05,
+      "loss": 2.2152,
+      "step": 46430
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.948693304041375e-05,
+      "loss": 2.1482,
+      "step": 46440
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9484644606160465e-05,
+      "loss": 2.1685,
+      "step": 46450
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.948235617190718e-05,
+      "loss": 2.1667,
+      "step": 46460
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.94800677376539e-05,
+      "loss": 2.1654,
+      "step": 46470
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9477779303400614e-05,
+      "loss": 2.075,
+      "step": 46480
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.947549086914733e-05,
+      "loss": 2.2146,
+      "step": 46490
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.947320243489405e-05,
+      "loss": 2.0415,
+      "step": 46500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9470914000640764e-05,
+      "loss": 2.0157,
+      "step": 46510
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.946862556638748e-05,
+      "loss": 2.0698,
+      "step": 46520
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.94663371321342e-05,
+      "loss": 2.1039,
+      "step": 46530
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.946404869788091e-05,
+      "loss": 2.5307,
+      "step": 46540
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.946176026362763e-05,
+      "loss": 2.0225,
+      "step": 46550
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.945947182937435e-05,
+      "loss": 2.1736,
+      "step": 46560
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.945718339512106e-05,
+      "loss": 1.9992,
+      "step": 46570
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9454894960867776e-05,
+      "loss": 2.1352,
+      "step": 46580
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.94526065266145e-05,
+      "loss": 2.2198,
+      "step": 46590
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.945031809236121e-05,
+      "loss": 2.0998,
+      "step": 46600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9448029658107925e-05,
+      "loss": 2.0463,
+      "step": 46610
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9445741223854646e-05,
+      "loss": 2.1323,
+      "step": 46620
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9443452789601354e-05,
+      "loss": 2.0414,
+      "step": 46630
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.944116435534807e-05,
+      "loss": 2.1022,
+      "step": 46640
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.943887592109479e-05,
+      "loss": 1.9823,
+      "step": 46650
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.94365874868415e-05,
+      "loss": 2.1385,
+      "step": 46660
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.943429905258822e-05,
+      "loss": 1.9906,
+      "step": 46670
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.943201061833494e-05,
+      "loss": 2.0898,
+      "step": 46680
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.942972218408165e-05,
+      "loss": 2.0516,
+      "step": 46690
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9427433749828366e-05,
+      "loss": 1.9933,
+      "step": 46700
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.942514531557509e-05,
+      "loss": 2.1243,
+      "step": 46710
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.94228568813218e-05,
+      "loss": 2.1474,
+      "step": 46720
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9420568447068515e-05,
+      "loss": 2.2017,
+      "step": 46730
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9418280012815236e-05,
+      "loss": 2.1995,
+      "step": 46740
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.941599157856195e-05,
+      "loss": 2.223,
+      "step": 46750
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9413703144308664e-05,
+      "loss": 2.0641,
+      "step": 46760
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9411414710055385e-05,
+      "loss": 2.1539,
+      "step": 46770
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.94091262758021e-05,
+      "loss": 1.962,
+      "step": 46780
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9406837841548814e-05,
+      "loss": 2.1593,
+      "step": 46790
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9404549407295534e-05,
+      "loss": 2.0852,
+      "step": 46800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.940226097304225e-05,
+      "loss": 2.1276,
+      "step": 46810
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.939997253878896e-05,
+      "loss": 2.1777,
+      "step": 46820
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.939768410453568e-05,
+      "loss": 2.1439,
+      "step": 46830
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.939539567028239e-05,
+      "loss": 2.0445,
+      "step": 46840
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.939310723602911e-05,
+      "loss": 2.3366,
+      "step": 46850
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9390818801775826e-05,
+      "loss": 1.9631,
+      "step": 46860
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.938853036752254e-05,
+      "loss": 2.0999,
+      "step": 46870
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.938624193326926e-05,
+      "loss": 2.1832,
+      "step": 46880
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9383953499015975e-05,
+      "loss": 2.1353,
+      "step": 46890
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.938166506476269e-05,
+      "loss": 2.2176,
+      "step": 46900
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.937937663050941e-05,
+      "loss": 2.1035,
+      "step": 46910
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9377088196256124e-05,
+      "loss": 2.0285,
+      "step": 46920
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.937479976200284e-05,
+      "loss": 2.268,
+      "step": 46930
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.937251132774956e-05,
+      "loss": 2.1951,
+      "step": 46940
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9370222893496274e-05,
+      "loss": 2.0705,
+      "step": 46950
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.936793445924299e-05,
+      "loss": 2.1762,
+      "step": 46960
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.93656460249897e-05,
+      "loss": 2.1088,
+      "step": 46970
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.936335759073642e-05,
+      "loss": 2.2773,
+      "step": 46980
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.936106915648314e-05,
+      "loss": 2.1462,
+      "step": 46990
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.935878072222985e-05,
+      "loss": 2.0329,
+      "step": 47000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.935649228797657e-05,
+      "loss": 2.1224,
+      "step": 47010
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9354203853723286e-05,
+      "loss": 2.2246,
+      "step": 47020
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.935191541947e-05,
+      "loss": 2.007,
+      "step": 47030
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9349626985216714e-05,
+      "loss": 2.2435,
+      "step": 47040
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.934733855096343e-05,
+      "loss": 2.2642,
+      "step": 47050
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.934505011671015e-05,
+      "loss": 2.0647,
+      "step": 47060
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9342761682456864e-05,
+      "loss": 2.1486,
+      "step": 47070
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.934047324820358e-05,
+      "loss": 2.1652,
+      "step": 47080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.93381848139503e-05,
+      "loss": 2.1048,
+      "step": 47090
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.933589637969701e-05,
+      "loss": 2.0926,
+      "step": 47100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.933360794544373e-05,
+      "loss": 2.0286,
+      "step": 47110
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.933131951119045e-05,
+      "loss": 2.2732,
+      "step": 47120
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.932903107693716e-05,
+      "loss": 2.2196,
+      "step": 47130
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9326742642683876e-05,
+      "loss": 2.0621,
+      "step": 47140
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.93244542084306e-05,
+      "loss": 2.2639,
+      "step": 47150
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.932216577417731e-05,
+      "loss": 2.0945,
+      "step": 47160
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9319877339924025e-05,
+      "loss": 2.0831,
+      "step": 47170
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9317588905670746e-05,
+      "loss": 2.1228,
+      "step": 47180
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.931530047141746e-05,
+      "loss": 2.1384,
+      "step": 47190
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9313012037164174e-05,
+      "loss": 2.1789,
+      "step": 47200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9310723602910895e-05,
+      "loss": 2.2875,
+      "step": 47210
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.930843516865761e-05,
+      "loss": 2.0704,
+      "step": 47220
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.930614673440432e-05,
+      "loss": 2.0716,
+      "step": 47230
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.930385830015104e-05,
+      "loss": 2.0722,
+      "step": 47240
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.930156986589775e-05,
+      "loss": 2.1781,
+      "step": 47250
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9299281431644466e-05,
+      "loss": 2.2662,
+      "step": 47260
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.929699299739119e-05,
+      "loss": 2.1319,
+      "step": 47270
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.92947045631379e-05,
+      "loss": 1.9706,
+      "step": 47280
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9292416128884615e-05,
+      "loss": 2.0597,
+      "step": 47290
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9290127694631336e-05,
+      "loss": 2.1308,
+      "step": 47300
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.928783926037805e-05,
+      "loss": 2.1059,
+      "step": 47310
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9285550826124764e-05,
+      "loss": 2.1393,
+      "step": 47320
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9283262391871485e-05,
+      "loss": 2.0494,
+      "step": 47330
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.92809739576182e-05,
+      "loss": 2.1099,
+      "step": 47340
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9278685523364913e-05,
+      "loss": 2.0949,
+      "step": 47350
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9276397089111634e-05,
+      "loss": 2.0336,
+      "step": 47360
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.927410865485835e-05,
+      "loss": 2.0894,
+      "step": 47370
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.927182022060506e-05,
+      "loss": 2.1,
+      "step": 47380
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9269531786351784e-05,
+      "loss": 2.0308,
+      "step": 47390
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.92672433520985e-05,
+      "loss": 2.0726,
+      "step": 47400
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.926495491784521e-05,
+      "loss": 2.0852,
+      "step": 47410
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9262666483591926e-05,
+      "loss": 2.0567,
+      "step": 47420
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.926037804933864e-05,
+      "loss": 2.119,
+      "step": 47430
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.925808961508536e-05,
+      "loss": 2.1944,
+      "step": 47440
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9255801180832075e-05,
+      "loss": 2.0604,
+      "step": 47450
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.925351274657879e-05,
+      "loss": 2.2514,
+      "step": 47460
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.925122431232551e-05,
+      "loss": 2.1075,
+      "step": 47470
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9248935878072224e-05,
+      "loss": 2.1664,
+      "step": 47480
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.924664744381894e-05,
+      "loss": 2.081,
+      "step": 47490
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.924435900956566e-05,
+      "loss": 2.1088,
+      "step": 47500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9242070575312373e-05,
+      "loss": 2.0347,
+      "step": 47510
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.923978214105909e-05,
+      "loss": 2.0807,
+      "step": 47520
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.923749370680581e-05,
+      "loss": 2.0928,
+      "step": 47530
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.923520527255252e-05,
+      "loss": 2.1616,
+      "step": 47540
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.923291683829924e-05,
+      "loss": 2.0876,
+      "step": 47550
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.923062840404596e-05,
+      "loss": 1.9977,
+      "step": 47560
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.922833996979267e-05,
+      "loss": 2.1167,
+      "step": 47570
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9226051535539386e-05,
+      "loss": 2.1003,
+      "step": 47580
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.922376310128611e-05,
+      "loss": 2.1648,
+      "step": 47590
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.922147466703282e-05,
+      "loss": 2.1422,
+      "step": 47600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9219186232779535e-05,
+      "loss": 2.1331,
+      "step": 47610
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.921689779852625e-05,
+      "loss": 2.0277,
+      "step": 47620
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9214609364272963e-05,
+      "loss": 2.2131,
+      "step": 47630
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.921232093001968e-05,
+      "loss": 2.1237,
+      "step": 47640
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.92100324957664e-05,
+      "loss": 2.1095,
+      "step": 47650
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.920774406151311e-05,
+      "loss": 2.0905,
+      "step": 47660
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.920545562725983e-05,
+      "loss": 2.0226,
+      "step": 47670
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.920316719300655e-05,
+      "loss": 2.1161,
+      "step": 47680
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.920087875875326e-05,
+      "loss": 2.0459,
+      "step": 47690
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9198590324499976e-05,
+      "loss": 1.9904,
+      "step": 47700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.91963018902467e-05,
+      "loss": 2.0351,
+      "step": 47710
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.919401345599341e-05,
+      "loss": 1.9596,
+      "step": 47720
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9191725021740125e-05,
+      "loss": 2.149,
+      "step": 47730
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9189436587486846e-05,
+      "loss": 2.3412,
+      "step": 47740
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.918714815323356e-05,
+      "loss": 2.1434,
+      "step": 47750
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9184859718980274e-05,
+      "loss": 2.1307,
+      "step": 47760
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9182571284726995e-05,
+      "loss": 2.1789,
+      "step": 47770
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.918028285047371e-05,
+      "loss": 2.067,
+      "step": 47780
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9177994416220423e-05,
+      "loss": 1.9824,
+      "step": 47790
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9175705981967144e-05,
+      "loss": 2.1465,
+      "step": 47800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.917341754771386e-05,
+      "loss": 2.0306,
+      "step": 47810
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.917112911346057e-05,
+      "loss": 2.2313,
+      "step": 47820
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.916884067920729e-05,
+      "loss": 2.2388,
+      "step": 47830
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9166552244954e-05,
+      "loss": 2.2025,
+      "step": 47840
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.916426381070072e-05,
+      "loss": 2.1539,
+      "step": 47850
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9161975376447436e-05,
+      "loss": 2.2095,
+      "step": 47860
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.915968694219415e-05,
+      "loss": 2.0798,
+      "step": 47870
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.915739850794087e-05,
+      "loss": 2.0238,
+      "step": 47880
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9155110073687585e-05,
+      "loss": 2.0203,
+      "step": 47890
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.91528216394343e-05,
+      "loss": 2.2616,
+      "step": 47900
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.915053320518102e-05,
+      "loss": 2.0437,
+      "step": 47910
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9148244770927734e-05,
+      "loss": 2.2033,
+      "step": 47920
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.914595633667445e-05,
+      "loss": 2.1951,
+      "step": 47930
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.914366790242117e-05,
+      "loss": 2.1867,
+      "step": 47940
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9141379468167883e-05,
+      "loss": 2.1719,
+      "step": 47950
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.91390910339146e-05,
+      "loss": 2.0207,
+      "step": 47960
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.913680259966132e-05,
+      "loss": 1.9708,
+      "step": 47970
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.913451416540803e-05,
+      "loss": 2.0069,
+      "step": 47980
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.913222573115475e-05,
+      "loss": 1.9804,
+      "step": 47990
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.912993729690147e-05,
+      "loss": 2.1072,
+      "step": 48000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.912764886264818e-05,
+      "loss": 2.141,
+      "step": 48010
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.912536042839489e-05,
+      "loss": 2.0372,
+      "step": 48020
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.912307199414161e-05,
+      "loss": 1.9954,
+      "step": 48030
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9120783559888324e-05,
+      "loss": 2.1736,
+      "step": 48040
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.911849512563504e-05,
+      "loss": 2.0503,
+      "step": 48050
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.911620669138176e-05,
+      "loss": 2.0308,
+      "step": 48060
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.911391825712847e-05,
+      "loss": 2.0254,
+      "step": 48070
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.911162982287519e-05,
+      "loss": 2.225,
+      "step": 48080
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.910934138862191e-05,
+      "loss": 2.1156,
+      "step": 48090
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.910705295436862e-05,
+      "loss": 2.0516,
+      "step": 48100
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.910476452011534e-05,
+      "loss": 2.1484,
+      "step": 48110
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.910247608586206e-05,
+      "loss": 2.1058,
+      "step": 48120
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.910018765160877e-05,
+      "loss": 2.2021,
+      "step": 48130
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9097899217355486e-05,
+      "loss": 2.0625,
+      "step": 48140
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.909561078310221e-05,
+      "loss": 2.1299,
+      "step": 48150
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.909332234884892e-05,
+      "loss": 2.1301,
+      "step": 48160
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9091033914595635e-05,
+      "loss": 2.0908,
+      "step": 48170
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9088745480342356e-05,
+      "loss": 2.088,
+      "step": 48180
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.908645704608907e-05,
+      "loss": 2.2526,
+      "step": 48190
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9084168611835784e-05,
+      "loss": 1.9042,
+      "step": 48200
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.90818801775825e-05,
+      "loss": 2.091,
+      "step": 48210
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.907959174332921e-05,
+      "loss": 2.0557,
+      "step": 48220
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9077303309075927e-05,
+      "loss": 2.139,
+      "step": 48230
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.907501487482265e-05,
+      "loss": 2.1324,
+      "step": 48240
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.907272644056936e-05,
+      "loss": 2.0967,
+      "step": 48250
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9070438006316076e-05,
+      "loss": 2.2388,
+      "step": 48260
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.90681495720628e-05,
+      "loss": 2.1678,
+      "step": 48270
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.906586113780951e-05,
+      "loss": 2.158,
+      "step": 48280
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9063572703556225e-05,
+      "loss": 2.2548,
+      "step": 48290
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9061284269302946e-05,
+      "loss": 2.0729,
+      "step": 48300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.905899583504966e-05,
+      "loss": 2.0849,
+      "step": 48310
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9056707400796374e-05,
+      "loss": 2.1063,
+      "step": 48320
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9054418966543095e-05,
+      "loss": 2.1273,
+      "step": 48330
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.905213053228981e-05,
+      "loss": 2.2323,
+      "step": 48340
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.904984209803652e-05,
+      "loss": 2.1298,
+      "step": 48350
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9047553663783244e-05,
+      "loss": 2.203,
+      "step": 48360
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.904526522952996e-05,
+      "loss": 2.1235,
+      "step": 48370
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.904297679527667e-05,
+      "loss": 1.9759,
+      "step": 48380
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9040688361023393e-05,
+      "loss": 2.153,
+      "step": 48390
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.903839992677011e-05,
+      "loss": 2.1184,
+      "step": 48400
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.903611149251682e-05,
+      "loss": 2.1196,
+      "step": 48410
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9033823058263536e-05,
+      "loss": 2.1118,
+      "step": 48420
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.903153462401025e-05,
+      "loss": 2.2099,
+      "step": 48430
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.902924618975697e-05,
+      "loss": 2.2601,
+      "step": 48440
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9026957755503685e-05,
+      "loss": 2.2459,
+      "step": 48450
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.90246693212504e-05,
+      "loss": 2.0059,
+      "step": 48460
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.902238088699712e-05,
+      "loss": 2.2662,
+      "step": 48470
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9020092452743834e-05,
+      "loss": 2.1621,
+      "step": 48480
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.901780401849055e-05,
+      "loss": 2.0599,
+      "step": 48490
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.901551558423727e-05,
+      "loss": 1.9925,
+      "step": 48500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.901322714998398e-05,
+      "loss": 1.9964,
+      "step": 48510
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.90109387157307e-05,
+      "loss": 2.136,
+      "step": 48520
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.900865028147742e-05,
+      "loss": 2.1398,
+      "step": 48530
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.900636184722413e-05,
+      "loss": 2.0984,
+      "step": 48540
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.900407341297085e-05,
+      "loss": 2.1067,
+      "step": 48550
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.900178497871757e-05,
+      "loss": 2.149,
+      "step": 48560
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.899949654446428e-05,
+      "loss": 2.0383,
+      "step": 48570
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8997208110210996e-05,
+      "loss": 2.2194,
+      "step": 48580
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.899491967595772e-05,
+      "loss": 2.0549,
+      "step": 48590
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.899263124170443e-05,
+      "loss": 2.1741,
+      "step": 48600
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.899034280745114e-05,
+      "loss": 2.1794,
+      "step": 48610
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.898805437319786e-05,
+      "loss": 2.1417,
+      "step": 48620
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.898576593894457e-05,
+      "loss": 2.1563,
+      "step": 48630
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.898347750469129e-05,
+      "loss": 2.1586,
+      "step": 48640
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.898118907043801e-05,
+      "loss": 2.0013,
+      "step": 48650
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.897890063618472e-05,
+      "loss": 2.2536,
+      "step": 48660
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8976612201931437e-05,
+      "loss": 2.1487,
+      "step": 48670
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.897432376767816e-05,
+      "loss": 2.2718,
+      "step": 48680
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.897203533342487e-05,
+      "loss": 2.116,
+      "step": 48690
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8969746899171586e-05,
+      "loss": 2.0268,
+      "step": 48700
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.896745846491831e-05,
+      "loss": 2.0296,
+      "step": 48710
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.896517003066502e-05,
+      "loss": 2.2047,
+      "step": 48720
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8962881596411735e-05,
+      "loss": 2.1616,
+      "step": 48730
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8960593162158456e-05,
+      "loss": 2.1995,
+      "step": 48740
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.895830472790517e-05,
+      "loss": 2.1266,
+      "step": 48750
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8956016293651884e-05,
+      "loss": 2.1413,
+      "step": 48760
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8953727859398605e-05,
+      "loss": 2.0414,
+      "step": 48770
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.895143942514532e-05,
+      "loss": 2.1119,
+      "step": 48780
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.894915099089203e-05,
+      "loss": 2.2525,
+      "step": 48790
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8946862556638754e-05,
+      "loss": 2.1574,
+      "step": 48800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.894457412238546e-05,
+      "loss": 2.0673,
+      "step": 48810
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.894228568813218e-05,
+      "loss": 2.0946,
+      "step": 48820
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8939997253878897e-05,
+      "loss": 2.1382,
+      "step": 48830
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.893770881962561e-05,
+      "loss": 2.0205,
+      "step": 48840
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.893542038537233e-05,
+      "loss": 2.0873,
+      "step": 48850
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8933131951119046e-05,
+      "loss": 2.096,
+      "step": 48860
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.893084351686576e-05,
+      "loss": 2.2307,
+      "step": 48870
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.892855508261248e-05,
+      "loss": 2.1105,
+      "step": 48880
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8926266648359195e-05,
+      "loss": 2.0669,
+      "step": 48890
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.892397821410591e-05,
+      "loss": 2.0435,
+      "step": 48900
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.892168977985263e-05,
+      "loss": 2.1155,
+      "step": 48910
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8919401345599344e-05,
+      "loss": 2.2491,
+      "step": 48920
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.891711291134606e-05,
+      "loss": 2.1616,
+      "step": 48930
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.891482447709278e-05,
+      "loss": 2.2592,
+      "step": 48940
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.891253604283949e-05,
+      "loss": 2.1397,
+      "step": 48950
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.891024760858621e-05,
+      "loss": 2.1768,
+      "step": 48960
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.890795917433293e-05,
+      "loss": 2.1034,
+      "step": 48970
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.890567074007964e-05,
+      "loss": 2.2465,
+      "step": 48980
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.890338230582636e-05,
+      "loss": 2.1126,
+      "step": 48990
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.890109387157307e-05,
+      "loss": 2.1296,
+      "step": 49000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8898805437319785e-05,
+      "loss": 2.1098,
+      "step": 49010
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.88965170030665e-05,
+      "loss": 2.0403,
+      "step": 49020
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.889422856881322e-05,
+      "loss": 2.1242,
+      "step": 49030
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8891940134559934e-05,
+      "loss": 1.9809,
+      "step": 49040
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.888965170030665e-05,
+      "loss": 1.9988,
+      "step": 49050
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.888736326605337e-05,
+      "loss": 2.0248,
+      "step": 49060
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.888507483180008e-05,
+      "loss": 2.1326,
+      "step": 49070
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.88827863975468e-05,
+      "loss": 2.1674,
+      "step": 49080
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.888049796329352e-05,
+      "loss": 2.0436,
+      "step": 49090
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.887820952904023e-05,
+      "loss": 2.0515,
+      "step": 49100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8875921094786947e-05,
+      "loss": 2.1244,
+      "step": 49110
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.887363266053367e-05,
+      "loss": 2.1201,
+      "step": 49120
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.887134422628038e-05,
+      "loss": 2.1938,
+      "step": 49130
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8869055792027096e-05,
+      "loss": 2.1548,
+      "step": 49140
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.886676735777382e-05,
+      "loss": 2.0863,
+      "step": 49150
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.886447892352053e-05,
+      "loss": 2.0893,
+      "step": 49160
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8862190489267245e-05,
+      "loss": 2.1011,
+      "step": 49170
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8859902055013966e-05,
+      "loss": 2.1514,
+      "step": 49180
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.885761362076068e-05,
+      "loss": 2.0747,
+      "step": 49190
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8855325186507394e-05,
+      "loss": 2.2197,
+      "step": 49200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.885303675225411e-05,
+      "loss": 2.0238,
+      "step": 49210
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.885074831800082e-05,
+      "loss": 2.157,
+      "step": 49220
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.884845988374754e-05,
+      "loss": 1.936,
+      "step": 49230
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.884617144949426e-05,
+      "loss": 2.242,
+      "step": 49240
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.884388301524097e-05,
+      "loss": 2.1498,
+      "step": 49250
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.884159458098769e-05,
+      "loss": 2.0668,
+      "step": 49260
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8839306146734407e-05,
+      "loss": 2.1587,
+      "step": 49270
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.883701771248112e-05,
+      "loss": 2.0574,
+      "step": 49280
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.883472927822784e-05,
+      "loss": 1.98,
+      "step": 49290
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8832440843974556e-05,
+      "loss": 2.1095,
+      "step": 49300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.883015240972127e-05,
+      "loss": 2.0554,
+      "step": 49310
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8827863975467984e-05,
+      "loss": 2.1567,
+      "step": 49320
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8825575541214705e-05,
+      "loss": 2.0909,
+      "step": 49330
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.882328710696142e-05,
+      "loss": 1.9946,
+      "step": 49340
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.882099867270813e-05,
+      "loss": 2.1261,
+      "step": 49350
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8818710238454854e-05,
+      "loss": 2.2894,
+      "step": 49360
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.881642180420157e-05,
+      "loss": 2.2577,
+      "step": 49370
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.881413336994828e-05,
+      "loss": 2.0698,
+      "step": 49380
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8811844935695e-05,
+      "loss": 2.1284,
+      "step": 49390
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.880955650144171e-05,
+      "loss": 2.0674,
+      "step": 49400
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.880726806718843e-05,
+      "loss": 2.1366,
+      "step": 49410
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8804979632935146e-05,
+      "loss": 2.0845,
+      "step": 49420
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.880269119868186e-05,
+      "loss": 2.1531,
+      "step": 49430
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.880040276442858e-05,
+      "loss": 2.1782,
+      "step": 49440
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8798114330175295e-05,
+      "loss": 1.9761,
+      "step": 49450
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.879582589592201e-05,
+      "loss": 2.2302,
+      "step": 49460
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.879353746166873e-05,
+      "loss": 2.3322,
+      "step": 49470
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8791249027415444e-05,
+      "loss": 2.1389,
+      "step": 49480
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.878896059316216e-05,
+      "loss": 2.1608,
+      "step": 49490
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.878667215890888e-05,
+      "loss": 2.1195,
+      "step": 49500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.878438372465559e-05,
+      "loss": 2.2983,
+      "step": 49510
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.878209529040231e-05,
+      "loss": 2.1761,
+      "step": 49520
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.877980685614903e-05,
+      "loss": 2.161,
+      "step": 49530
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.877751842189574e-05,
+      "loss": 2.0599,
+      "step": 49540
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8775229987642457e-05,
+      "loss": 2.0318,
+      "step": 49550
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.877294155338918e-05,
+      "loss": 2.2334,
+      "step": 49560
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.877065311913589e-05,
+      "loss": 2.1773,
+      "step": 49570
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8768364684882606e-05,
+      "loss": 2.258,
+      "step": 49580
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8766076250629327e-05,
+      "loss": 2.2573,
+      "step": 49590
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8763787816376034e-05,
+      "loss": 2.3803,
+      "step": 49600
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.876149938212275e-05,
+      "loss": 2.1301,
+      "step": 49610
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.875921094786947e-05,
+      "loss": 2.1291,
+      "step": 49620
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.875692251361618e-05,
+      "loss": 2.2625,
+      "step": 49630
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.87546340793629e-05,
+      "loss": 2.1612,
+      "step": 49640
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.875234564510962e-05,
+      "loss": 2.2474,
+      "step": 49650
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.875005721085633e-05,
+      "loss": 2.1393,
+      "step": 49660
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8747768776603046e-05,
+      "loss": 2.253,
+      "step": 49670
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.874548034234977e-05,
+      "loss": 2.2509,
+      "step": 49680
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.874319190809648e-05,
+      "loss": 2.1933,
+      "step": 49690
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8740903473843196e-05,
+      "loss": 2.0035,
+      "step": 49700
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8738615039589917e-05,
+      "loss": 2.0659,
+      "step": 49710
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.873632660533663e-05,
+      "loss": 2.0804,
+      "step": 49720
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8734038171083345e-05,
+      "loss": 2.0729,
+      "step": 49730
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8731749736830066e-05,
+      "loss": 2.0256,
+      "step": 49740
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.872946130257678e-05,
+      "loss": 2.2274,
+      "step": 49750
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8727172868323494e-05,
+      "loss": 2.1318,
+      "step": 49760
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8724884434070215e-05,
+      "loss": 2.1691,
+      "step": 49770
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.872259599981693e-05,
+      "loss": 2.1749,
+      "step": 49780
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.872030756556364e-05,
+      "loss": 2.2132,
+      "step": 49790
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.871801913131036e-05,
+      "loss": 2.1117,
+      "step": 49800
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.871573069705707e-05,
+      "loss": 2.2657,
+      "step": 49810
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.871344226280379e-05,
+      "loss": 1.9524,
+      "step": 49820
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8711153828550506e-05,
+      "loss": 2.102,
+      "step": 49830
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.870886539429722e-05,
+      "loss": 2.0142,
+      "step": 49840
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.870657696004394e-05,
+      "loss": 1.9817,
+      "step": 49850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8704288525790656e-05,
+      "loss": 2.1695,
+      "step": 49860
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.870200009153737e-05,
+      "loss": 2.2847,
+      "step": 49870
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.869971165728409e-05,
+      "loss": 2.0244,
+      "step": 49880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8697423223030805e-05,
+      "loss": 2.3591,
+      "step": 49890
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.869513478877752e-05,
+      "loss": 2.1937,
+      "step": 49900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.869284635452424e-05,
+      "loss": 2.1039,
+      "step": 49910
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8690557920270954e-05,
+      "loss": 2.0935,
+      "step": 49920
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.868826948601767e-05,
+      "loss": 2.2592,
+      "step": 49930
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.868598105176439e-05,
+      "loss": 2.1644,
+      "step": 49940
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.86836926175111e-05,
+      "loss": 2.184,
+      "step": 49950
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.868140418325782e-05,
+      "loss": 2.0912,
+      "step": 49960
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.867911574900454e-05,
+      "loss": 2.1489,
+      "step": 49970
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.867682731475125e-05,
+      "loss": 2.1004,
+      "step": 49980
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.867453888049796e-05,
+      "loss": 2.1049,
+      "step": 49990
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.867225044624468e-05,
+      "loss": 2.1222,
+      "step": 50000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8669962011991395e-05,
+      "loss": 2.0034,
+      "step": 50010
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.866767357773811e-05,
+      "loss": 2.2468,
+      "step": 50020
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.866538514348483e-05,
+      "loss": 2.1985,
+      "step": 50030
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8663096709231544e-05,
+      "loss": 2.2666,
+      "step": 50040
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.866080827497826e-05,
+      "loss": 1.9806,
+      "step": 50050
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.865851984072498e-05,
+      "loss": 2.2344,
+      "step": 50060
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.865623140647169e-05,
+      "loss": 2.1659,
+      "step": 50070
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.865394297221841e-05,
+      "loss": 2.1695,
+      "step": 50080
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.865165453796513e-05,
+      "loss": 1.9414,
+      "step": 50090
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.864936610371184e-05,
+      "loss": 2.0546,
+      "step": 50100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8647077669458556e-05,
+      "loss": 2.0776,
+      "step": 50110
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.864478923520528e-05,
+      "loss": 2.1789,
+      "step": 50120
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.864250080095199e-05,
+      "loss": 2.1126,
+      "step": 50130
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8640212366698706e-05,
+      "loss": 2.1298,
+      "step": 50140
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8637923932445426e-05,
+      "loss": 2.0951,
+      "step": 50150
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.863563549819214e-05,
+      "loss": 2.1212,
+      "step": 50160
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8633347063938855e-05,
+      "loss": 2.1231,
+      "step": 50170
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8631058629685576e-05,
+      "loss": 2.0869,
+      "step": 50180
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.862877019543228e-05,
+      "loss": 2.1711,
+      "step": 50190
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8626481761179004e-05,
+      "loss": 2.0133,
+      "step": 50200
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.862419332692572e-05,
+      "loss": 2.0554,
+      "step": 50210
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.862190489267243e-05,
+      "loss": 2.189,
+      "step": 50220
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.861961645841915e-05,
+      "loss": 2.0093,
+      "step": 50230
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.861732802416587e-05,
+      "loss": 1.981,
+      "step": 50240
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.861503958991258e-05,
+      "loss": 2.1032,
+      "step": 50250
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.86127511556593e-05,
+      "loss": 1.971,
+      "step": 50260
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8610462721406016e-05,
+      "loss": 2.11,
+      "step": 50270
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.860817428715273e-05,
+      "loss": 1.9814,
+      "step": 50280
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.860588585289945e-05,
+      "loss": 1.8928,
+      "step": 50290
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8603597418646166e-05,
+      "loss": 2.1477,
+      "step": 50300
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.860130898439288e-05,
+      "loss": 2.0402,
+      "step": 50310
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.85990205501396e-05,
+      "loss": 2.1096,
+      "step": 50320
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8596732115886315e-05,
+      "loss": 2.0049,
+      "step": 50330
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.859444368163303e-05,
+      "loss": 2.1294,
+      "step": 50340
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.859215524737975e-05,
+      "loss": 2.163,
+      "step": 50350
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8589866813126464e-05,
+      "loss": 2.0552,
+      "step": 50360
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.858757837887318e-05,
+      "loss": 2.1217,
+      "step": 50370
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.85852899446199e-05,
+      "loss": 2.2756,
+      "step": 50380
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8583001510366606e-05,
+      "loss": 2.1641,
+      "step": 50390
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.858071307611332e-05,
+      "loss": 2.0336,
+      "step": 50400
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.857842464186004e-05,
+      "loss": 2.1027,
+      "step": 50410
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8576136207606756e-05,
+      "loss": 2.0484,
+      "step": 50420
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.857384777335347e-05,
+      "loss": 2.2565,
+      "step": 50430
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.857155933910019e-05,
+      "loss": 2.1195,
+      "step": 50440
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8569270904846905e-05,
+      "loss": 1.9386,
+      "step": 50450
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.856698247059362e-05,
+      "loss": 2.137,
+      "step": 50460
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.856469403634034e-05,
+      "loss": 2.1935,
+      "step": 50470
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8562405602087054e-05,
+      "loss": 2.1224,
+      "step": 50480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.856011716783377e-05,
+      "loss": 2.1669,
+      "step": 50490
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.855782873358049e-05,
+      "loss": 2.1013,
+      "step": 50500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.85555402993272e-05,
+      "loss": 2.0976,
+      "step": 50510
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.855325186507392e-05,
+      "loss": 2.2225,
+      "step": 50520
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.855096343082064e-05,
+      "loss": 2.0513,
+      "step": 50530
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.854867499656735e-05,
+      "loss": 2.032,
+      "step": 50540
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8546386562314066e-05,
+      "loss": 2.2771,
+      "step": 50550
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.854409812806079e-05,
+      "loss": 2.1413,
+      "step": 50560
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.85418096938075e-05,
+      "loss": 2.3045,
+      "step": 50570
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.853952125955421e-05,
+      "loss": 2.182,
+      "step": 50580
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.853723282530093e-05,
+      "loss": 2.0662,
+      "step": 50590
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8534944391047644e-05,
+      "loss": 2.1612,
+      "step": 50600
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.853265595679436e-05,
+      "loss": 2.096,
+      "step": 50610
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.853036752254108e-05,
+      "loss": 2.128,
+      "step": 50620
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.852807908828779e-05,
+      "loss": 2.2188,
+      "step": 50630
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.852579065403451e-05,
+      "loss": 2.1602,
+      "step": 50640
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.852350221978123e-05,
+      "loss": 2.0541,
+      "step": 50650
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.852121378552794e-05,
+      "loss": 2.0798,
+      "step": 50660
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8518925351274656e-05,
+      "loss": 2.2255,
+      "step": 50670
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.851663691702138e-05,
+      "loss": 1.976,
+      "step": 50680
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.851434848276809e-05,
+      "loss": 2.2851,
+      "step": 50690
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8512060048514805e-05,
+      "loss": 2.1,
+      "step": 50700
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8509771614261526e-05,
+      "loss": 2.2103,
+      "step": 50710
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.850748318000824e-05,
+      "loss": 2.049,
+      "step": 50720
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8505194745754955e-05,
+      "loss": 2.2592,
+      "step": 50730
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8502906311501676e-05,
+      "loss": 2.0832,
+      "step": 50740
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.850061787724839e-05,
+      "loss": 2.2371,
+      "step": 50750
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8498329442995104e-05,
+      "loss": 2.1011,
+      "step": 50760
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8496041008741825e-05,
+      "loss": 2.142,
+      "step": 50770
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.849375257448853e-05,
+      "loss": 2.1086,
+      "step": 50780
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.849146414023525e-05,
+      "loss": 2.1281,
+      "step": 50790
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.848917570598197e-05,
+      "loss": 2.0689,
+      "step": 50800
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.848688727172868e-05,
+      "loss": 2.1096,
+      "step": 50810
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.84845988374754e-05,
+      "loss": 2.108,
+      "step": 50820
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8482310403222116e-05,
+      "loss": 2.0857,
+      "step": 50830
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.848002196896883e-05,
+      "loss": 2.0518,
+      "step": 50840
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.847773353471555e-05,
+      "loss": 2.1641,
+      "step": 50850
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8475445100462266e-05,
+      "loss": 2.1298,
+      "step": 50860
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.847315666620898e-05,
+      "loss": 2.0757,
+      "step": 50870
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.84708682319557e-05,
+      "loss": 1.9765,
+      "step": 50880
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8468579797702415e-05,
+      "loss": 2.0261,
+      "step": 50890
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.846629136344913e-05,
+      "loss": 2.1579,
+      "step": 50900
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.846400292919585e-05,
+      "loss": 2.1198,
+      "step": 50910
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8461714494942564e-05,
+      "loss": 2.1848,
+      "step": 50920
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.845942606068928e-05,
+      "loss": 2.0203,
+      "step": 50930
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8457137626436e-05,
+      "loss": 2.2157,
+      "step": 50940
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.845484919218271e-05,
+      "loss": 2.0742,
+      "step": 50950
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.845256075792943e-05,
+      "loss": 2.1496,
+      "step": 50960
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.845027232367615e-05,
+      "loss": 2.2359,
+      "step": 50970
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8447983889422855e-05,
+      "loss": 2.0105,
+      "step": 50980
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.844569545516957e-05,
+      "loss": 2.0373,
+      "step": 50990
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.844340702091629e-05,
+      "loss": 2.1069,
+      "step": 51000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8441118586663005e-05,
+      "loss": 2.0813,
+      "step": 51010
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.843883015240972e-05,
+      "loss": 2.0863,
+      "step": 51020
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.843654171815644e-05,
+      "loss": 2.1473,
+      "step": 51030
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8434253283903154e-05,
+      "loss": 2.1806,
+      "step": 51040
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.843196484964987e-05,
+      "loss": 2.0446,
+      "step": 51050
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.842967641539659e-05,
+      "loss": 2.3122,
+      "step": 51060
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.84273879811433e-05,
+      "loss": 2.1379,
+      "step": 51070
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.842509954689002e-05,
+      "loss": 1.9715,
+      "step": 51080
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.842281111263674e-05,
+      "loss": 2.0568,
+      "step": 51090
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.842052267838345e-05,
+      "loss": 2.0714,
+      "step": 51100
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8418234244130166e-05,
+      "loss": 1.9473,
+      "step": 51110
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.841594580987689e-05,
+      "loss": 2.0012,
+      "step": 51120
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.84136573756236e-05,
+      "loss": 2.0751,
+      "step": 51130
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8411368941370315e-05,
+      "loss": 2.2669,
+      "step": 51140
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8409080507117036e-05,
+      "loss": 2.3476,
+      "step": 51150
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.840679207286375e-05,
+      "loss": 2.0941,
+      "step": 51160
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8404503638610465e-05,
+      "loss": 2.1436,
+      "step": 51170
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.840221520435718e-05,
+      "loss": 2.1463,
+      "step": 51180
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.839992677010389e-05,
+      "loss": 2.0512,
+      "step": 51190
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8397638335850614e-05,
+      "loss": 2.1011,
+      "step": 51200
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.839534990159733e-05,
+      "loss": 2.0,
+      "step": 51210
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.839306146734404e-05,
+      "loss": 2.1432,
+      "step": 51220
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.839077303309076e-05,
+      "loss": 2.1033,
+      "step": 51230
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.838848459883748e-05,
+      "loss": 2.2186,
+      "step": 51240
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.838619616458419e-05,
+      "loss": 2.132,
+      "step": 51250
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.838390773033091e-05,
+      "loss": 2.0954,
+      "step": 51260
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8381619296077626e-05,
+      "loss": 2.0263,
+      "step": 51270
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.837933086182434e-05,
+      "loss": 2.0679,
+      "step": 51280
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.837704242757106e-05,
+      "loss": 2.1228,
+      "step": 51290
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8374753993317775e-05,
+      "loss": 2.0461,
+      "step": 51300
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.837246555906449e-05,
+      "loss": 2.0341,
+      "step": 51310
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.837017712481121e-05,
+      "loss": 2.0894,
+      "step": 51320
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8367888690557925e-05,
+      "loss": 2.1274,
+      "step": 51330
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.836560025630464e-05,
+      "loss": 2.1241,
+      "step": 51340
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.836331182205136e-05,
+      "loss": 1.9904,
+      "step": 51350
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8361023387798074e-05,
+      "loss": 2.1198,
+      "step": 51360
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.835873495354478e-05,
+      "loss": 2.0308,
+      "step": 51370
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.83564465192915e-05,
+      "loss": 2.2135,
+      "step": 51380
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8354158085038216e-05,
+      "loss": 2.292,
+      "step": 51390
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.835186965078493e-05,
+      "loss": 2.009,
+      "step": 51400
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.834958121653165e-05,
+      "loss": 2.2131,
+      "step": 51410
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8347292782278365e-05,
+      "loss": 2.0647,
+      "step": 51420
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.834500434802508e-05,
+      "loss": 1.9326,
+      "step": 51430
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.83427159137718e-05,
+      "loss": 2.2176,
+      "step": 51440
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8340427479518515e-05,
+      "loss": 2.0945,
+      "step": 51450
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.833813904526523e-05,
+      "loss": 2.0784,
+      "step": 51460
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.833585061101195e-05,
+      "loss": 2.0984,
+      "step": 51470
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8333562176758664e-05,
+      "loss": 2.1016,
+      "step": 51480
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.833127374250538e-05,
+      "loss": 1.9336,
+      "step": 51490
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.83289853082521e-05,
+      "loss": 1.9483,
+      "step": 51500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.832669687399881e-05,
+      "loss": 2.1782,
+      "step": 51510
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.832440843974553e-05,
+      "loss": 2.0435,
+      "step": 51520
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.832212000549225e-05,
+      "loss": 2.1124,
+      "step": 51530
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.831983157123896e-05,
+      "loss": 2.1835,
+      "step": 51540
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8317543136985676e-05,
+      "loss": 2.0966,
+      "step": 51550
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.83152547027324e-05,
+      "loss": 2.0575,
+      "step": 51560
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8312966268479105e-05,
+      "loss": 2.1056,
+      "step": 51570
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8310677834225825e-05,
+      "loss": 2.1733,
+      "step": 51580
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.830838939997254e-05,
+      "loss": 2.2471,
+      "step": 51590
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8306100965719254e-05,
+      "loss": 2.1342,
+      "step": 51600
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8303812531465975e-05,
+      "loss": 2.106,
+      "step": 51610
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.830152409721269e-05,
+      "loss": 2.006,
+      "step": 51620
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.82992356629594e-05,
+      "loss": 2.2349,
+      "step": 51630
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8296947228706124e-05,
+      "loss": 2.0237,
+      "step": 51640
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.829465879445284e-05,
+      "loss": 2.1679,
+      "step": 51650
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.829237036019955e-05,
+      "loss": 2.121,
+      "step": 51660
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8290081925946266e-05,
+      "loss": 2.139,
+      "step": 51670
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.828779349169299e-05,
+      "loss": 1.9834,
+      "step": 51680
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.82855050574397e-05,
+      "loss": 2.1793,
+      "step": 51690
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8283216623186415e-05,
+      "loss": 2.0125,
+      "step": 51700
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8280928188933136e-05,
+      "loss": 2.2065,
+      "step": 51710
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.827863975467985e-05,
+      "loss": 2.047,
+      "step": 51720
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8276351320426565e-05,
+      "loss": 2.3328,
+      "step": 51730
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8274062886173285e-05,
+      "loss": 2.0506,
+      "step": 51740
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.827177445192e-05,
+      "loss": 1.9986,
+      "step": 51750
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8269486017666714e-05,
+      "loss": 1.8926,
+      "step": 51760
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.826719758341343e-05,
+      "loss": 1.9845,
+      "step": 51770
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.826490914916014e-05,
+      "loss": 2.1111,
+      "step": 51780
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.826262071490686e-05,
+      "loss": 2.1354,
+      "step": 51790
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.826033228065358e-05,
+      "loss": 2.1618,
+      "step": 51800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.825804384640029e-05,
+      "loss": 2.1757,
+      "step": 51810
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.825575541214701e-05,
+      "loss": 2.0079,
+      "step": 51820
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8253466977893726e-05,
+      "loss": 2.0823,
+      "step": 51830
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.825117854364044e-05,
+      "loss": 2.0434,
+      "step": 51840
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.824889010938716e-05,
+      "loss": 2.0382,
+      "step": 51850
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8246601675133875e-05,
+      "loss": 2.0787,
+      "step": 51860
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.824431324088059e-05,
+      "loss": 2.0355,
+      "step": 51870
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.824202480662731e-05,
+      "loss": 2.076,
+      "step": 51880
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8239736372374025e-05,
+      "loss": 2.124,
+      "step": 51890
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.823744793812074e-05,
+      "loss": 2.0418,
+      "step": 51900
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.823515950386746e-05,
+      "loss": 2.0612,
+      "step": 51910
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8232871069614174e-05,
+      "loss": 2.0437,
+      "step": 51920
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.823058263536089e-05,
+      "loss": 2.0563,
+      "step": 51930
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.822829420110761e-05,
+      "loss": 2.0505,
+      "step": 51940
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.822600576685432e-05,
+      "loss": 2.1083,
+      "step": 51950
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.822371733260104e-05,
+      "loss": 2.0186,
+      "step": 51960
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.822142889834775e-05,
+      "loss": 2.0287,
+      "step": 51970
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8219140464094465e-05,
+      "loss": 2.0659,
+      "step": 51980
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.821685202984118e-05,
+      "loss": 2.0006,
+      "step": 51990
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.82145635955879e-05,
+      "loss": 2.0655,
+      "step": 52000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8212275161334614e-05,
+      "loss": 2.0427,
+      "step": 52010
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.820998672708133e-05,
+      "loss": 2.0517,
+      "step": 52020
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.820769829282805e-05,
+      "loss": 2.2595,
+      "step": 52030
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8205409858574764e-05,
+      "loss": 2.2415,
+      "step": 52040
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.820312142432148e-05,
+      "loss": 2.0681,
+      "step": 52050
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.82008329900682e-05,
+      "loss": 2.1711,
+      "step": 52060
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.819854455581491e-05,
+      "loss": 2.0362,
+      "step": 52070
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.819625612156163e-05,
+      "loss": 2.0014,
+      "step": 52080
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.819396768730835e-05,
+      "loss": 2.1327,
+      "step": 52090
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.819167925305506e-05,
+      "loss": 1.9867,
+      "step": 52100
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8189390818801776e-05,
+      "loss": 2.1712,
+      "step": 52110
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.81871023845485e-05,
+      "loss": 2.0778,
+      "step": 52120
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.818481395029521e-05,
+      "loss": 2.1349,
+      "step": 52130
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8182525516041925e-05,
+      "loss": 2.0145,
+      "step": 52140
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8180237081788646e-05,
+      "loss": 2.0326,
+      "step": 52150
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8177948647535354e-05,
+      "loss": 2.2197,
+      "step": 52160
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8175660213282074e-05,
+      "loss": 2.1302,
+      "step": 52170
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.817337177902879e-05,
+      "loss": 2.1389,
+      "step": 52180
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.81710833447755e-05,
+      "loss": 2.1834,
+      "step": 52190
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8168794910522224e-05,
+      "loss": 2.2194,
+      "step": 52200
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.816650647626894e-05,
+      "loss": 2.0031,
+      "step": 52210
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.816421804201565e-05,
+      "loss": 2.0231,
+      "step": 52220
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.816192960776237e-05,
+      "loss": 2.0943,
+      "step": 52230
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.815964117350909e-05,
+      "loss": 2.0155,
+      "step": 52240
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.81573527392558e-05,
+      "loss": 2.1102,
+      "step": 52250
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.815506430500252e-05,
+      "loss": 1.9898,
+      "step": 52260
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8152775870749236e-05,
+      "loss": 2.129,
+      "step": 52270
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.815048743649595e-05,
+      "loss": 2.1677,
+      "step": 52280
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.814819900224267e-05,
+      "loss": 2.1734,
+      "step": 52290
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8145910567989385e-05,
+      "loss": 2.0654,
+      "step": 52300
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.81436221337361e-05,
+      "loss": 2.0678,
+      "step": 52310
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.814133369948282e-05,
+      "loss": 2.0278,
+      "step": 52320
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8139045265229535e-05,
+      "loss": 1.8912,
+      "step": 52330
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.813675683097625e-05,
+      "loss": 2.1537,
+      "step": 52340
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.813446839672297e-05,
+      "loss": 2.0677,
+      "step": 52350
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.813217996246968e-05,
+      "loss": 2.1631,
+      "step": 52360
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.812989152821639e-05,
+      "loss": 2.1152,
+      "step": 52370
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.812760309396311e-05,
+      "loss": 1.9674,
+      "step": 52380
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8125314659709826e-05,
+      "loss": 2.0749,
+      "step": 52390
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.812302622545654e-05,
+      "loss": 2.0694,
+      "step": 52400
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.812073779120326e-05,
+      "loss": 2.2792,
+      "step": 52410
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8118449356949975e-05,
+      "loss": 1.9582,
+      "step": 52420
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.811616092269669e-05,
+      "loss": 2.0275,
+      "step": 52430
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.811387248844341e-05,
+      "loss": 2.2523,
+      "step": 52440
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8111584054190124e-05,
+      "loss": 2.1268,
+      "step": 52450
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.810929561993684e-05,
+      "loss": 2.1986,
+      "step": 52460
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.810700718568356e-05,
+      "loss": 2.1972,
+      "step": 52470
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8104718751430274e-05,
+      "loss": 2.1031,
+      "step": 52480
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.810243031717699e-05,
+      "loss": 2.1115,
+      "step": 52490
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.810014188292371e-05,
+      "loss": 2.155,
+      "step": 52500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.809785344867042e-05,
+      "loss": 2.2326,
+      "step": 52510
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.809556501441714e-05,
+      "loss": 2.0367,
+      "step": 52520
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.809327658016386e-05,
+      "loss": 2.1638,
+      "step": 52530
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.809098814591057e-05,
+      "loss": 2.0417,
+      "step": 52540
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8088699711657286e-05,
+      "loss": 2.0686,
+      "step": 52550
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8086411277404e-05,
+      "loss": 2.0457,
+      "step": 52560
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8084122843150714e-05,
+      "loss": 2.175,
+      "step": 52570
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8081834408897435e-05,
+      "loss": 2.1104,
+      "step": 52580
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.807954597464415e-05,
+      "loss": 2.0676,
+      "step": 52590
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8077257540390864e-05,
+      "loss": 2.0664,
+      "step": 52600
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8074969106137584e-05,
+      "loss": 2.0509,
+      "step": 52610
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.80726806718843e-05,
+      "loss": 2.0892,
+      "step": 52620
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.807039223763101e-05,
+      "loss": 2.1214,
+      "step": 52630
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8068103803377734e-05,
+      "loss": 2.0521,
+      "step": 52640
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.806581536912445e-05,
+      "loss": 2.0869,
+      "step": 52650
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.806352693487116e-05,
+      "loss": 2.232,
+      "step": 52660
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.806123850061788e-05,
+      "loss": 2.0205,
+      "step": 52670
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.80589500663646e-05,
+      "loss": 2.221,
+      "step": 52680
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.805666163211131e-05,
+      "loss": 2.1396,
+      "step": 52690
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.805437319785803e-05,
+      "loss": 2.2324,
+      "step": 52700
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8052084763604746e-05,
+      "loss": 2.1654,
+      "step": 52710
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.804979632935146e-05,
+      "loss": 2.1377,
+      "step": 52720
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.804750789509818e-05,
+      "loss": 2.0661,
+      "step": 52730
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8045219460844895e-05,
+      "loss": 2.1598,
+      "step": 52740
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.804293102659161e-05,
+      "loss": 2.1578,
+      "step": 52750
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8040642592338324e-05,
+      "loss": 2.1061,
+      "step": 52760
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.803835415808504e-05,
+      "loss": 1.9702,
+      "step": 52770
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.803606572383175e-05,
+      "loss": 2.1122,
+      "step": 52780
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.803377728957847e-05,
+      "loss": 1.9899,
+      "step": 52790
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.803148885532519e-05,
+      "loss": 2.1356,
+      "step": 52800
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.80292004210719e-05,
+      "loss": 2.1551,
+      "step": 52810
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.802691198681862e-05,
+      "loss": 2.0418,
+      "step": 52820
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8024623552565336e-05,
+      "loss": 2.2507,
+      "step": 52830
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.802233511831205e-05,
+      "loss": 2.1581,
+      "step": 52840
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.802004668405877e-05,
+      "loss": 2.0773,
+      "step": 52850
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8017758249805485e-05,
+      "loss": 2.1546,
+      "step": 52860
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.80154698155522e-05,
+      "loss": 1.9898,
+      "step": 52870
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.801318138129892e-05,
+      "loss": 2.1816,
+      "step": 52880
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8010892947045634e-05,
+      "loss": 2.1261,
+      "step": 52890
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.800860451279235e-05,
+      "loss": 1.9197,
+      "step": 52900
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.800631607853907e-05,
+      "loss": 2.1006,
+      "step": 52910
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8004027644285784e-05,
+      "loss": 2.0812,
+      "step": 52920
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.80017392100325e-05,
+      "loss": 2.2013,
+      "step": 52930
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.799945077577922e-05,
+      "loss": 1.9934,
+      "step": 52940
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7997162341525926e-05,
+      "loss": 2.0336,
+      "step": 52950
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.799487390727264e-05,
+      "loss": 2.2171,
+      "step": 52960
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.799258547301936e-05,
+      "loss": 2.065,
+      "step": 52970
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7990297038766075e-05,
+      "loss": 2.0223,
+      "step": 52980
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.798800860451279e-05,
+      "loss": 1.9035,
+      "step": 52990
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.798572017025951e-05,
+      "loss": 2.0178,
+      "step": 53000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7983431736006224e-05,
+      "loss": 2.2346,
+      "step": 53010
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.798114330175294e-05,
+      "loss": 2.0685,
+      "step": 53020
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.797885486749966e-05,
+      "loss": 2.1521,
+      "step": 53030
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7976566433246374e-05,
+      "loss": 2.0657,
+      "step": 53040
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.797427799899309e-05,
+      "loss": 2.0312,
+      "step": 53050
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.797198956473981e-05,
+      "loss": 2.2251,
+      "step": 53060
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.796970113048652e-05,
+      "loss": 2.0927,
+      "step": 53070
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.796741269623324e-05,
+      "loss": 1.9918,
+      "step": 53080
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.796512426197996e-05,
+      "loss": 2.23,
+      "step": 53090
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.796283582772667e-05,
+      "loss": 2.0668,
+      "step": 53100
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7960547393473386e-05,
+      "loss": 2.1497,
+      "step": 53110
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.795825895922011e-05,
+      "loss": 2.1952,
+      "step": 53120
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.795597052496682e-05,
+      "loss": 2.0981,
+      "step": 53130
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7953682090713535e-05,
+      "loss": 2.1518,
+      "step": 53140
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.795139365646025e-05,
+      "loss": 2.1024,
+      "step": 53150
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7949105222206963e-05,
+      "loss": 2.3129,
+      "step": 53160
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7946816787953684e-05,
+      "loss": 2.1279,
+      "step": 53170
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.79445283537004e-05,
+      "loss": 2.0288,
+      "step": 53180
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.794223991944711e-05,
+      "loss": 1.966,
+      "step": 53190
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7939951485193834e-05,
+      "loss": 2.1596,
+      "step": 53200
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.793766305094055e-05,
+      "loss": 2.1315,
+      "step": 53210
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.793537461668726e-05,
+      "loss": 2.1474,
+      "step": 53220
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.793308618243398e-05,
+      "loss": 2.1883,
+      "step": 53230
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.79307977481807e-05,
+      "loss": 2.1089,
+      "step": 53240
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.792850931392741e-05,
+      "loss": 2.1806,
+      "step": 53250
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.792622087967413e-05,
+      "loss": 2.1233,
+      "step": 53260
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7923932445420846e-05,
+      "loss": 2.156,
+      "step": 53270
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.792164401116756e-05,
+      "loss": 2.1202,
+      "step": 53280
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.791935557691428e-05,
+      "loss": 2.1214,
+      "step": 53290
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7917067142660995e-05,
+      "loss": 2.0905,
+      "step": 53300
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.791477870840771e-05,
+      "loss": 2.1636,
+      "step": 53310
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.791249027415443e-05,
+      "loss": 2.047,
+      "step": 53320
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7910201839901144e-05,
+      "loss": 2.1588,
+      "step": 53330
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.790791340564786e-05,
+      "loss": 2.0122,
+      "step": 53340
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.790562497139457e-05,
+      "loss": 1.8912,
+      "step": 53350
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.790333653714129e-05,
+      "loss": 2.0633,
+      "step": 53360
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7901048102888e-05,
+      "loss": 2.0528,
+      "step": 53370
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.789875966863472e-05,
+      "loss": 1.9656,
+      "step": 53380
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7896471234381436e-05,
+      "loss": 2.2616,
+      "step": 53390
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.789418280012815e-05,
+      "loss": 2.1235,
+      "step": 53400
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.789189436587487e-05,
+      "loss": 2.1273,
+      "step": 53410
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7889605931621585e-05,
+      "loss": 2.0519,
+      "step": 53420
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.78873174973683e-05,
+      "loss": 2.2067,
+      "step": 53430
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.788502906311502e-05,
+      "loss": 2.0375,
+      "step": 53440
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7882740628861734e-05,
+      "loss": 2.1687,
+      "step": 53450
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.788045219460845e-05,
+      "loss": 2.0649,
+      "step": 53460
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.787816376035517e-05,
+      "loss": 2.115,
+      "step": 53470
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7875875326101883e-05,
+      "loss": 2.0573,
+      "step": 53480
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.78735868918486e-05,
+      "loss": 1.995,
+      "step": 53490
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.787129845759532e-05,
+      "loss": 2.0309,
+      "step": 53500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.786901002334203e-05,
+      "loss": 2.3506,
+      "step": 53510
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.786672158908875e-05,
+      "loss": 2.0166,
+      "step": 53520
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.786443315483547e-05,
+      "loss": 2.2474,
+      "step": 53530
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.786214472058218e-05,
+      "loss": 2.0506,
+      "step": 53540
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7859856286328896e-05,
+      "loss": 1.9023,
+      "step": 53550
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.785756785207561e-05,
+      "loss": 2.1758,
+      "step": 53560
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7855279417822324e-05,
+      "loss": 2.2241,
+      "step": 53570
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7852990983569045e-05,
+      "loss": 2.0969,
+      "step": 53580
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.785070254931576e-05,
+      "loss": 2.0847,
+      "step": 53590
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7848414115062473e-05,
+      "loss": 1.9857,
+      "step": 53600
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7846125680809194e-05,
+      "loss": 2.0399,
+      "step": 53610
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.784383724655591e-05,
+      "loss": 2.0017,
+      "step": 53620
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.784154881230262e-05,
+      "loss": 2.1251,
+      "step": 53630
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7839260378049344e-05,
+      "loss": 2.2302,
+      "step": 53640
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.783697194379606e-05,
+      "loss": 2.142,
+      "step": 53650
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.783468350954277e-05,
+      "loss": 2.1778,
+      "step": 53660
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.783239507528949e-05,
+      "loss": 2.0409,
+      "step": 53670
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.783010664103621e-05,
+      "loss": 2.2633,
+      "step": 53680
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.782781820678292e-05,
+      "loss": 2.1282,
+      "step": 53690
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.782552977252964e-05,
+      "loss": 2.0592,
+      "step": 53700
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7823241338276356e-05,
+      "loss": 2.0712,
+      "step": 53710
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.782095290402307e-05,
+      "loss": 2.2526,
+      "step": 53720
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.781866446976979e-05,
+      "loss": 2.0818,
+      "step": 53730
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.78163760355165e-05,
+      "loss": 1.9357,
+      "step": 53740
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.781408760126321e-05,
+      "loss": 1.9911,
+      "step": 53750
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7811799167009933e-05,
+      "loss": 2.0886,
+      "step": 53760
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.780951073275665e-05,
+      "loss": 2.1458,
+      "step": 53770
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.780722229850336e-05,
+      "loss": 1.9885,
+      "step": 53780
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.780493386425008e-05,
+      "loss": 2.131,
+      "step": 53790
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.78026454299968e-05,
+      "loss": 2.0626,
+      "step": 53800
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.780035699574351e-05,
+      "loss": 2.1246,
+      "step": 53810
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.779806856149023e-05,
+      "loss": 2.0163,
+      "step": 53820
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7795780127236946e-05,
+      "loss": 2.2968,
+      "step": 53830
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.779349169298366e-05,
+      "loss": 2.0274,
+      "step": 53840
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.779120325873038e-05,
+      "loss": 2.1396,
+      "step": 53850
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7788914824477095e-05,
+      "loss": 2.1114,
+      "step": 53860
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.778662639022381e-05,
+      "loss": 2.1231,
+      "step": 53870
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.778433795597053e-05,
+      "loss": 2.2463,
+      "step": 53880
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7782049521717244e-05,
+      "loss": 1.9275,
+      "step": 53890
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.777976108746396e-05,
+      "loss": 2.1139,
+      "step": 53900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.777747265321068e-05,
+      "loss": 2.1483,
+      "step": 53910
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7775184218957393e-05,
+      "loss": 2.1499,
+      "step": 53920
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.777289578470411e-05,
+      "loss": 2.1123,
+      "step": 53930
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.777060735045082e-05,
+      "loss": 2.0921,
+      "step": 53940
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7768318916197536e-05,
+      "loss": 2.0257,
+      "step": 53950
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.776603048194426e-05,
+      "loss": 2.1954,
+      "step": 53960
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.776374204769097e-05,
+      "loss": 1.8288,
+      "step": 53970
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7761453613437685e-05,
+      "loss": 2.132,
+      "step": 53980
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.77591651791844e-05,
+      "loss": 2.1294,
+      "step": 53990
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.775687674493112e-05,
+      "loss": 2.1889,
+      "step": 54000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7754588310677834e-05,
+      "loss": 1.9717,
+      "step": 54010
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.775229987642455e-05,
+      "loss": 2.2544,
+      "step": 54020
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.775001144217127e-05,
+      "loss": 1.971,
+      "step": 54030
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7747723007917983e-05,
+      "loss": 2.0851,
+      "step": 54040
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.77454345736647e-05,
+      "loss": 2.1712,
+      "step": 54050
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.774314613941142e-05,
+      "loss": 2.1703,
+      "step": 54060
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.774085770515813e-05,
+      "loss": 2.0685,
+      "step": 54070
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.773856927090485e-05,
+      "loss": 2.0789,
+      "step": 54080
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.773628083665157e-05,
+      "loss": 2.1583,
+      "step": 54090
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.773399240239828e-05,
+      "loss": 2.0998,
+      "step": 54100
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7731703968144996e-05,
+      "loss": 2.0634,
+      "step": 54110
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.772941553389172e-05,
+      "loss": 2.0597,
+      "step": 54120
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.772712709963843e-05,
+      "loss": 2.2742,
+      "step": 54130
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7724838665385145e-05,
+      "loss": 2.1188,
+      "step": 54140
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.772255023113186e-05,
+      "loss": 2.1912,
+      "step": 54150
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.772026179687857e-05,
+      "loss": 2.0167,
+      "step": 54160
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7717973362625294e-05,
+      "loss": 2.1042,
+      "step": 54170
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.771568492837201e-05,
+      "loss": 2.1339,
+      "step": 54180
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.771339649411872e-05,
+      "loss": 2.061,
+      "step": 54190
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7711108059865443e-05,
+      "loss": 1.9906,
+      "step": 54200
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.770881962561216e-05,
+      "loss": 2.1004,
+      "step": 54210
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.770653119135887e-05,
+      "loss": 2.0601,
+      "step": 54220
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.770424275710559e-05,
+      "loss": 2.1492,
+      "step": 54230
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.770195432285231e-05,
+      "loss": 2.059,
+      "step": 54240
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.769966588859902e-05,
+      "loss": 2.1444,
+      "step": 54250
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.769737745434574e-05,
+      "loss": 2.2549,
+      "step": 54260
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7695089020092456e-05,
+      "loss": 2.1385,
+      "step": 54270
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.769280058583917e-05,
+      "loss": 2.217,
+      "step": 54280
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.769051215158589e-05,
+      "loss": 2.107,
+      "step": 54290
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7688223717332605e-05,
+      "loss": 2.0783,
+      "step": 54300
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.768593528307932e-05,
+      "loss": 2.0685,
+      "step": 54310
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.768364684882604e-05,
+      "loss": 2.0873,
+      "step": 54320
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7681358414572754e-05,
+      "loss": 2.2005,
+      "step": 54330
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.767906998031946e-05,
+      "loss": 2.1855,
+      "step": 54340
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.767678154606618e-05,
+      "loss": 2.0212,
+      "step": 54350
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.76744931118129e-05,
+      "loss": 2.0476,
+      "step": 54360
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.767220467755961e-05,
+      "loss": 2.1564,
+      "step": 54370
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.766991624330633e-05,
+      "loss": 2.0888,
+      "step": 54380
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7667627809053046e-05,
+      "loss": 2.0253,
+      "step": 54390
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.766533937479976e-05,
+      "loss": 2.0812,
+      "step": 54400
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.766305094054648e-05,
+      "loss": 2.1892,
+      "step": 54410
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7660762506293195e-05,
+      "loss": 2.0332,
+      "step": 54420
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.765847407203991e-05,
+      "loss": 2.0473,
+      "step": 54430
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.765618563778663e-05,
+      "loss": 2.024,
+      "step": 54440
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7653897203533344e-05,
+      "loss": 1.9344,
+      "step": 54450
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.765160876928006e-05,
+      "loss": 1.9635,
+      "step": 54460
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.764932033502678e-05,
+      "loss": 2.0122,
+      "step": 54470
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.764703190077349e-05,
+      "loss": 2.0122,
+      "step": 54480
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.764474346652021e-05,
+      "loss": 2.0958,
+      "step": 54490
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.764245503226693e-05,
+      "loss": 2.2532,
+      "step": 54500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.764016659801364e-05,
+      "loss": 2.0298,
+      "step": 54510
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.763787816376036e-05,
+      "loss": 2.1321,
+      "step": 54520
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.763558972950707e-05,
+      "loss": 2.0334,
+      "step": 54530
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7633301295253785e-05,
+      "loss": 2.0792,
+      "step": 54540
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7631012861000506e-05,
+      "loss": 2.1192,
+      "step": 54550
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.762872442674722e-05,
+      "loss": 2.044,
+      "step": 54560
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7626435992493934e-05,
+      "loss": 2.1887,
+      "step": 54570
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7624147558240655e-05,
+      "loss": 2.0137,
+      "step": 54580
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.762185912398737e-05,
+      "loss": 2.2364,
+      "step": 54590
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.761957068973408e-05,
+      "loss": 2.0129,
+      "step": 54600
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7617282255480804e-05,
+      "loss": 1.9886,
+      "step": 54610
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.761499382122752e-05,
+      "loss": 2.1535,
+      "step": 54620
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.761270538697423e-05,
+      "loss": 2.1311,
+      "step": 54630
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.761041695272095e-05,
+      "loss": 2.0224,
+      "step": 54640
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.760812851846767e-05,
+      "loss": 2.023,
+      "step": 54650
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.760584008421438e-05,
+      "loss": 2.0654,
+      "step": 54660
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.76035516499611e-05,
+      "loss": 2.1174,
+      "step": 54670
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.760126321570782e-05,
+      "loss": 1.9262,
+      "step": 54680
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.759897478145453e-05,
+      "loss": 2.0423,
+      "step": 54690
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.759668634720125e-05,
+      "loss": 2.1999,
+      "step": 54700
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7594397912947966e-05,
+      "loss": 2.0796,
+      "step": 54710
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.759210947869468e-05,
+      "loss": 2.0349,
+      "step": 54720
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7589821044441394e-05,
+      "loss": 2.0472,
+      "step": 54730
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.758753261018811e-05,
+      "loss": 2.0753,
+      "step": 54740
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.758524417593482e-05,
+      "loss": 2.0556,
+      "step": 54750
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.758295574168154e-05,
+      "loss": 2.061,
+      "step": 54760
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.758066730742826e-05,
+      "loss": 2.0391,
+      "step": 54770
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.757837887317497e-05,
+      "loss": 1.9065,
+      "step": 54780
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.757609043892169e-05,
+      "loss": 1.9818,
+      "step": 54790
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.757380200466841e-05,
+      "loss": 2.2574,
+      "step": 54800
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.757151357041512e-05,
+      "loss": 2.1606,
+      "step": 54810
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.756922513616184e-05,
+      "loss": 2.198,
+      "step": 54820
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7566936701908556e-05,
+      "loss": 2.1161,
+      "step": 54830
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.756464826765527e-05,
+      "loss": 2.192,
+      "step": 54840
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.756235983340199e-05,
+      "loss": 2.0151,
+      "step": 54850
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7560071399148705e-05,
+      "loss": 2.0372,
+      "step": 54860
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.755778296489542e-05,
+      "loss": 2.0069,
+      "step": 54870
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.755549453064214e-05,
+      "loss": 2.0536,
+      "step": 54880
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7553206096388854e-05,
+      "loss": 2.0625,
+      "step": 54890
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.755091766213557e-05,
+      "loss": 2.1191,
+      "step": 54900
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.754862922788229e-05,
+      "loss": 2.0426,
+      "step": 54910
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7546340793629e-05,
+      "loss": 2.0312,
+      "step": 54920
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.754405235937572e-05,
+      "loss": 2.1156,
+      "step": 54930
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.754176392512243e-05,
+      "loss": 2.1274,
+      "step": 54940
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7539475490869146e-05,
+      "loss": 2.0928,
+      "step": 54950
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.753718705661587e-05,
+      "loss": 2.0822,
+      "step": 54960
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.753489862236258e-05,
+      "loss": 2.0791,
+      "step": 54970
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7532610188109295e-05,
+      "loss": 2.0343,
+      "step": 54980
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7530321753856016e-05,
+      "loss": 1.9317,
+      "step": 54990
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.752803331960273e-05,
+      "loss": 2.0684,
+      "step": 55000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7525744885349444e-05,
+      "loss": 2.1137,
+      "step": 55010
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7523456451096165e-05,
+      "loss": 2.1406,
+      "step": 55020
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.752116801684288e-05,
+      "loss": 2.1325,
+      "step": 55030
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.751887958258959e-05,
+      "loss": 1.9589,
+      "step": 55040
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7516591148336314e-05,
+      "loss": 2.2374,
+      "step": 55050
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.751430271408303e-05,
+      "loss": 2.1404,
+      "step": 55060
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.751201427982974e-05,
+      "loss": 2.12,
+      "step": 55070
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.750972584557646e-05,
+      "loss": 2.1285,
+      "step": 55080
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.750743741132318e-05,
+      "loss": 2.0181,
+      "step": 55090
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.750514897706989e-05,
+      "loss": 2.1276,
+      "step": 55100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7502860542816606e-05,
+      "loss": 2.258,
+      "step": 55110
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.750057210856333e-05,
+      "loss": 2.0513,
+      "step": 55120
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7498283674310034e-05,
+      "loss": 2.144,
+      "step": 55130
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7495995240056755e-05,
+      "loss": 2.0709,
+      "step": 55140
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.749370680580347e-05,
+      "loss": 1.9705,
+      "step": 55150
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.749141837155018e-05,
+      "loss": 2.1022,
+      "step": 55160
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7489129937296904e-05,
+      "loss": 2.2126,
+      "step": 55170
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.748684150304362e-05,
+      "loss": 2.0057,
+      "step": 55180
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.748455306879033e-05,
+      "loss": 2.0227,
+      "step": 55190
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.748226463453705e-05,
+      "loss": 1.9519,
+      "step": 55200
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.747997620028377e-05,
+      "loss": 2.1013,
+      "step": 55210
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.747768776603048e-05,
+      "loss": 2.0136,
+      "step": 55220
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.74753993317772e-05,
+      "loss": 1.9668,
+      "step": 55230
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7473110897523917e-05,
+      "loss": 2.1216,
+      "step": 55240
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.747082246327063e-05,
+      "loss": 2.0082,
+      "step": 55250
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.746853402901735e-05,
+      "loss": 2.0234,
+      "step": 55260
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7466245594764066e-05,
+      "loss": 2.1334,
+      "step": 55270
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.746395716051078e-05,
+      "loss": 2.117,
+      "step": 55280
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.74616687262575e-05,
+      "loss": 2.0839,
+      "step": 55290
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7459380292004215e-05,
+      "loss": 2.2928,
+      "step": 55300
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.745709185775093e-05,
+      "loss": 2.0782,
+      "step": 55310
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.745480342349765e-05,
+      "loss": 2.0692,
+      "step": 55320
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.745251498924436e-05,
+      "loss": 2.1831,
+      "step": 55330
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.745022655499107e-05,
+      "loss": 2.1831,
+      "step": 55340
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.744793812073779e-05,
+      "loss": 2.0251,
+      "step": 55350
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7445649686484507e-05,
+      "loss": 2.229,
+      "step": 55360
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.744336125223122e-05,
+      "loss": 2.0796,
+      "step": 55370
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.744107281797794e-05,
+      "loss": 1.9666,
+      "step": 55380
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7438784383724656e-05,
+      "loss": 1.9538,
+      "step": 55390
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.743649594947137e-05,
+      "loss": 2.2513,
+      "step": 55400
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.743420751521809e-05,
+      "loss": 2.0895,
+      "step": 55410
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7431919080964805e-05,
+      "loss": 2.1402,
+      "step": 55420
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.742963064671152e-05,
+      "loss": 1.9313,
+      "step": 55430
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.742734221245824e-05,
+      "loss": 2.0794,
+      "step": 55440
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7425053778204954e-05,
+      "loss": 2.0605,
+      "step": 55450
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.742276534395167e-05,
+      "loss": 2.0145,
+      "step": 55460
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.742047690969839e-05,
+      "loss": 2.082,
+      "step": 55470
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.74181884754451e-05,
+      "loss": 1.9713,
+      "step": 55480
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.741590004119182e-05,
+      "loss": 1.9474,
+      "step": 55490
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.741361160693854e-05,
+      "loss": 2.1112,
+      "step": 55500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.741132317268525e-05,
+      "loss": 2.1003,
+      "step": 55510
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7409034738431967e-05,
+      "loss": 1.9048,
+      "step": 55520
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.740674630417868e-05,
+      "loss": 2.0815,
+      "step": 55530
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7404457869925395e-05,
+      "loss": 2.1491,
+      "step": 55540
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7402169435672116e-05,
+      "loss": 2.0447,
+      "step": 55550
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.739988100141883e-05,
+      "loss": 2.133,
+      "step": 55560
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7397592567165544e-05,
+      "loss": 2.1885,
+      "step": 55570
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7395304132912265e-05,
+      "loss": 2.0825,
+      "step": 55580
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.739301569865898e-05,
+      "loss": 2.062,
+      "step": 55590
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.739072726440569e-05,
+      "loss": 2.1198,
+      "step": 55600
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7388438830152414e-05,
+      "loss": 2.0793,
+      "step": 55610
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.738615039589913e-05,
+      "loss": 1.923,
+      "step": 55620
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.738386196164584e-05,
+      "loss": 2.0132,
+      "step": 55630
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.738157352739256e-05,
+      "loss": 2.1455,
+      "step": 55640
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.737928509313928e-05,
+      "loss": 2.1385,
+      "step": 55650
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.737699665888599e-05,
+      "loss": 2.0693,
+      "step": 55660
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.737470822463271e-05,
+      "loss": 2.2256,
+      "step": 55670
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7372419790379427e-05,
+      "loss": 2.0247,
+      "step": 55680
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.737013135612614e-05,
+      "loss": 2.1928,
+      "step": 55690
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.736784292187286e-05,
+      "loss": 2.0533,
+      "step": 55700
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7365554487619576e-05,
+      "loss": 2.0897,
+      "step": 55710
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.736326605336628e-05,
+      "loss": 2.2051,
+      "step": 55720
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7360977619113004e-05,
+      "loss": 2.0997,
+      "step": 55730
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.735868918485972e-05,
+      "loss": 1.9786,
+      "step": 55740
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.735640075060643e-05,
+      "loss": 2.0799,
+      "step": 55750
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.735411231635315e-05,
+      "loss": 2.1259,
+      "step": 55760
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.735182388209987e-05,
+      "loss": 2.07,
+      "step": 55770
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.734953544784658e-05,
+      "loss": 2.0415,
+      "step": 55780
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.73472470135933e-05,
+      "loss": 2.0219,
+      "step": 55790
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7344958579340016e-05,
+      "loss": 2.0663,
+      "step": 55800
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.734267014508673e-05,
+      "loss": 1.9442,
+      "step": 55810
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.734038171083345e-05,
+      "loss": 2.0578,
+      "step": 55820
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7338093276580166e-05,
+      "loss": 1.8445,
+      "step": 55830
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.733580484232688e-05,
+      "loss": 2.0965,
+      "step": 55840
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.73335164080736e-05,
+      "loss": 2.059,
+      "step": 55850
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7331227973820315e-05,
+      "loss": 2.116,
+      "step": 55860
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.732893953956703e-05,
+      "loss": 2.1164,
+      "step": 55870
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.732665110531375e-05,
+      "loss": 2.1189,
+      "step": 55880
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7324362671060464e-05,
+      "loss": 2.0734,
+      "step": 55890
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.732207423680718e-05,
+      "loss": 1.9508,
+      "step": 55900
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.73197858025539e-05,
+      "loss": 2.3485,
+      "step": 55910
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7317497368300606e-05,
+      "loss": 2.0454,
+      "step": 55920
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.731520893404733e-05,
+      "loss": 2.1684,
+      "step": 55930
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.731292049979404e-05,
+      "loss": 2.1762,
+      "step": 55940
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7310632065540756e-05,
+      "loss": 2.1283,
+      "step": 55950
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7308343631287477e-05,
+      "loss": 2.1153,
+      "step": 55960
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.730605519703419e-05,
+      "loss": 1.9631,
+      "step": 55970
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7303766762780905e-05,
+      "loss": 2.1562,
+      "step": 55980
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7301478328527626e-05,
+      "loss": 2.1698,
+      "step": 55990
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.729918989427434e-05,
+      "loss": 2.0108,
+      "step": 56000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7296901460021054e-05,
+      "loss": 2.0552,
+      "step": 56010
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7294613025767775e-05,
+      "loss": 2.06,
+      "step": 56020
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.729232459151449e-05,
+      "loss": 1.9603,
+      "step": 56030
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.72900361572612e-05,
+      "loss": 2.3348,
+      "step": 56040
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7287747723007924e-05,
+      "loss": 2.0553,
+      "step": 56050
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.728545928875464e-05,
+      "loss": 2.0116,
+      "step": 56060
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.728317085450135e-05,
+      "loss": 2.1814,
+      "step": 56070
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.728088242024807e-05,
+      "loss": 2.1096,
+      "step": 56080
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.727859398599479e-05,
+      "loss": 2.2713,
+      "step": 56090
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.72763055517415e-05,
+      "loss": 2.1631,
+      "step": 56100
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.727401711748822e-05,
+      "loss": 2.0889,
+      "step": 56110
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.727172868323493e-05,
+      "loss": 2.0393,
+      "step": 56120
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7269440248981644e-05,
+      "loss": 2.0818,
+      "step": 56130
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7267151814728365e-05,
+      "loss": 2.0089,
+      "step": 56140
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.726486338047508e-05,
+      "loss": 2.1336,
+      "step": 56150
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.726257494622179e-05,
+      "loss": 1.9101,
+      "step": 56160
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7260286511968514e-05,
+      "loss": 2.0744,
+      "step": 56170
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.725799807771523e-05,
+      "loss": 2.0823,
+      "step": 56180
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.725570964346194e-05,
+      "loss": 2.0133,
+      "step": 56190
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.725342120920866e-05,
+      "loss": 2.1695,
+      "step": 56200
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.725113277495538e-05,
+      "loss": 2.0733,
+      "step": 56210
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.724884434070209e-05,
+      "loss": 2.1198,
+      "step": 56220
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.724655590644881e-05,
+      "loss": 2.1105,
+      "step": 56230
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7244267472195526e-05,
+      "loss": 1.9211,
+      "step": 56240
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.724197903794224e-05,
+      "loss": 2.0019,
+      "step": 56250
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.723969060368896e-05,
+      "loss": 2.102,
+      "step": 56260
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7237402169435676e-05,
+      "loss": 2.1646,
+      "step": 56270
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.723511373518239e-05,
+      "loss": 2.0723,
+      "step": 56280
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.723282530092911e-05,
+      "loss": 2.1022,
+      "step": 56290
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7230536866675825e-05,
+      "loss": 1.9632,
+      "step": 56300
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.722824843242254e-05,
+      "loss": 2.004,
+      "step": 56310
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.722595999816925e-05,
+      "loss": 1.9706,
+      "step": 56320
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.722367156391597e-05,
+      "loss": 1.9945,
+      "step": 56330
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.722138312966268e-05,
+      "loss": 2.0379,
+      "step": 56340
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.72190946954094e-05,
+      "loss": 2.082,
+      "step": 56350
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7216806261156116e-05,
+      "loss": 2.0053,
+      "step": 56360
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.721451782690283e-05,
+      "loss": 2.2256,
+      "step": 56370
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.721222939264955e-05,
+      "loss": 2.1377,
+      "step": 56380
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7209940958396266e-05,
+      "loss": 2.1441,
+      "step": 56390
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.720765252414298e-05,
+      "loss": 2.097,
+      "step": 56400
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.72053640898897e-05,
+      "loss": 2.0194,
+      "step": 56410
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7203075655636415e-05,
+      "loss": 1.9461,
+      "step": 56420
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.720078722138313e-05,
+      "loss": 2.1546,
+      "step": 56430
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.719849878712985e-05,
+      "loss": 1.9252,
+      "step": 56440
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7196210352876564e-05,
+      "loss": 2.1071,
+      "step": 56450
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.719392191862328e-05,
+      "loss": 2.1992,
+      "step": 56460
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.719163348437e-05,
+      "loss": 2.0437,
+      "step": 56470
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.718934505011671e-05,
+      "loss": 2.0317,
+      "step": 56480
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.718705661586343e-05,
+      "loss": 2.1424,
+      "step": 56490
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.718476818161015e-05,
+      "loss": 1.916,
+      "step": 56500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7182479747356855e-05,
+      "loss": 2.004,
+      "step": 56510
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7180191313103576e-05,
+      "loss": 2.029,
+      "step": 56520
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.717790287885029e-05,
+      "loss": 1.9793,
+      "step": 56530
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7175614444597005e-05,
+      "loss": 1.9213,
+      "step": 56540
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7173326010343726e-05,
+      "loss": 2.0126,
+      "step": 56550
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.717103757609044e-05,
+      "loss": 1.9991,
+      "step": 56560
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7168749141837154e-05,
+      "loss": 2.2415,
+      "step": 56570
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7166460707583875e-05,
+      "loss": 1.9796,
+      "step": 56580
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.716417227333059e-05,
+      "loss": 2.1871,
+      "step": 56590
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.71618838390773e-05,
+      "loss": 2.0648,
+      "step": 56600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7159595404824024e-05,
+      "loss": 2.0426,
+      "step": 56610
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.715730697057074e-05,
+      "loss": 2.1107,
+      "step": 56620
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.715501853631745e-05,
+      "loss": 2.0771,
+      "step": 56630
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.715273010206417e-05,
+      "loss": 2.0962,
+      "step": 56640
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.715044166781089e-05,
+      "loss": 2.0463,
+      "step": 56650
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.71481532335576e-05,
+      "loss": 2.1098,
+      "step": 56660
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.714586479930432e-05,
+      "loss": 2.0142,
+      "step": 56670
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7143576365051036e-05,
+      "loss": 1.8776,
+      "step": 56680
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.714128793079775e-05,
+      "loss": 1.9612,
+      "step": 56690
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.713899949654447e-05,
+      "loss": 2.1125,
+      "step": 56700
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.713671106229118e-05,
+      "loss": 2.0839,
+      "step": 56710
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.713442262803789e-05,
+      "loss": 2.1675,
+      "step": 56720
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7132134193784614e-05,
+      "loss": 2.1217,
+      "step": 56730
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.712984575953133e-05,
+      "loss": 2.0722,
+      "step": 56740
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.712755732527804e-05,
+      "loss": 2.0234,
+      "step": 56750
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.712526889102476e-05,
+      "loss": 2.1482,
+      "step": 56760
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.712298045677148e-05,
+      "loss": 2.0112,
+      "step": 56770
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.712069202251819e-05,
+      "loss": 1.98,
+      "step": 56780
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.711840358826491e-05,
+      "loss": 2.0728,
+      "step": 56790
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7116115154011626e-05,
+      "loss": 2.1524,
+      "step": 56800
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.711382671975834e-05,
+      "loss": 2.1106,
+      "step": 56810
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.711153828550506e-05,
+      "loss": 1.9692,
+      "step": 56820
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7109249851251776e-05,
+      "loss": 2.092,
+      "step": 56830
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.710696141699849e-05,
+      "loss": 2.1573,
+      "step": 56840
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.710467298274521e-05,
+      "loss": 2.0802,
+      "step": 56850
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7102384548491925e-05,
+      "loss": 2.1592,
+      "step": 56860
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.710009611423864e-05,
+      "loss": 2.0554,
+      "step": 56870
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.709780767998536e-05,
+      "loss": 2.0852,
+      "step": 56880
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7095519245732074e-05,
+      "loss": 2.062,
+      "step": 56890
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.709323081147879e-05,
+      "loss": 2.0247,
+      "step": 56900
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.70909423772255e-05,
+      "loss": 2.155,
+      "step": 56910
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7088653942972216e-05,
+      "loss": 2.001,
+      "step": 56920
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.708636550871894e-05,
+      "loss": 1.944,
+      "step": 56930
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.708407707446565e-05,
+      "loss": 1.9979,
+      "step": 56940
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7081788640212365e-05,
+      "loss": 2.1044,
+      "step": 56950
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7079500205959086e-05,
+      "loss": 2.2294,
+      "step": 56960
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.70772117717058e-05,
+      "loss": 1.9362,
+      "step": 56970
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7074923337452515e-05,
+      "loss": 2.0825,
+      "step": 56980
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7072634903199236e-05,
+      "loss": 2.1283,
+      "step": 56990
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.707034646894595e-05,
+      "loss": 2.0789,
+      "step": 57000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7068058034692664e-05,
+      "loss": 2.0858,
+      "step": 57010
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7065769600439385e-05,
+      "loss": 1.9174,
+      "step": 57020
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.70634811661861e-05,
+      "loss": 1.9471,
+      "step": 57030
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.706119273193281e-05,
+      "loss": 2.159,
+      "step": 57040
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7058904297679534e-05,
+      "loss": 2.0339,
+      "step": 57050
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.705661586342625e-05,
+      "loss": 2.1899,
+      "step": 57060
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.705432742917296e-05,
+      "loss": 2.0512,
+      "step": 57070
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.705203899491968e-05,
+      "loss": 2.1914,
+      "step": 57080
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.70497505606664e-05,
+      "loss": 1.8672,
+      "step": 57090
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7047462126413105e-05,
+      "loss": 2.1735,
+      "step": 57100
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7045173692159825e-05,
+      "loss": 2.0316,
+      "step": 57110
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.704288525790654e-05,
+      "loss": 2.1796,
+      "step": 57120
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7040596823653254e-05,
+      "loss": 2.1074,
+      "step": 57130
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7038308389399975e-05,
+      "loss": 1.9527,
+      "step": 57140
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.703601995514669e-05,
+      "loss": 2.19,
+      "step": 57150
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.70337315208934e-05,
+      "loss": 2.1559,
+      "step": 57160
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7031443086640124e-05,
+      "loss": 1.9654,
+      "step": 57170
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.702915465238684e-05,
+      "loss": 2.1255,
+      "step": 57180
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.702686621813355e-05,
+      "loss": 1.9261,
+      "step": 57190
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.702457778388027e-05,
+      "loss": 2.0065,
+      "step": 57200
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.702228934962699e-05,
+      "loss": 2.0228,
+      "step": 57210
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.70200009153737e-05,
+      "loss": 2.2086,
+      "step": 57220
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.701771248112042e-05,
+      "loss": 2.1557,
+      "step": 57230
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7015424046867136e-05,
+      "loss": 2.1885,
+      "step": 57240
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.701313561261385e-05,
+      "loss": 1.9917,
+      "step": 57250
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.701084717836057e-05,
+      "loss": 2.1469,
+      "step": 57260
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7008558744107285e-05,
+      "loss": 2.1008,
+      "step": 57270
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7006270309854e-05,
+      "loss": 2.1187,
+      "step": 57280
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.700398187560072e-05,
+      "loss": 2.2132,
+      "step": 57290
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.700169344134743e-05,
+      "loss": 2.0774,
+      "step": 57300
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.699940500709415e-05,
+      "loss": 1.8832,
+      "step": 57310
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.699711657284086e-05,
+      "loss": 1.9734,
+      "step": 57320
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.699482813858758e-05,
+      "loss": 2.102,
+      "step": 57330
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.69925397043343e-05,
+      "loss": 2.0303,
+      "step": 57340
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.699025127008101e-05,
+      "loss": 2.0072,
+      "step": 57350
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6987962835827726e-05,
+      "loss": 2.013,
+      "step": 57360
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.698567440157445e-05,
+      "loss": 2.0624,
+      "step": 57370
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.698338596732116e-05,
+      "loss": 2.1109,
+      "step": 57380
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6981097533067875e-05,
+      "loss": 2.18,
+      "step": 57390
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6978809098814596e-05,
+      "loss": 2.096,
+      "step": 57400
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.697652066456131e-05,
+      "loss": 2.3225,
+      "step": 57410
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6974232230308025e-05,
+      "loss": 1.9349,
+      "step": 57420
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.697194379605474e-05,
+      "loss": 2.1478,
+      "step": 57430
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.696965536180146e-05,
+      "loss": 2.1915,
+      "step": 57440
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6967366927548174e-05,
+      "loss": 2.0166,
+      "step": 57450
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.696507849329489e-05,
+      "loss": 1.8968,
+      "step": 57460
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.696279005904161e-05,
+      "loss": 2.0278,
+      "step": 57470
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.696050162478832e-05,
+      "loss": 2.0988,
+      "step": 57480
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.695821319053504e-05,
+      "loss": 1.942,
+      "step": 57490
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.695592475628175e-05,
+      "loss": 2.0734,
+      "step": 57500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6953636322028465e-05,
+      "loss": 2.0561,
+      "step": 57510
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6951347887775186e-05,
+      "loss": 2.1839,
+      "step": 57520
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.69490594535219e-05,
+      "loss": 2.0426,
+      "step": 57530
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6946771019268615e-05,
+      "loss": 1.938,
+      "step": 57540
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6944482585015335e-05,
+      "loss": 2.0839,
+      "step": 57550
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.694219415076205e-05,
+      "loss": 2.1951,
+      "step": 57560
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6939905716508764e-05,
+      "loss": 1.9209,
+      "step": 57570
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6937617282255485e-05,
+      "loss": 1.9029,
+      "step": 57580
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.69353288480022e-05,
+      "loss": 2.073,
+      "step": 57590
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.693304041374891e-05,
+      "loss": 2.1439,
+      "step": 57600
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6930751979495634e-05,
+      "loss": 1.9993,
+      "step": 57610
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.692846354524235e-05,
+      "loss": 2.1121,
+      "step": 57620
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.692617511098906e-05,
+      "loss": 1.8303,
+      "step": 57630
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.692388667673578e-05,
+      "loss": 2.0225,
+      "step": 57640
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.69215982424825e-05,
+      "loss": 2.0697,
+      "step": 57650
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.691930980822921e-05,
+      "loss": 1.9887,
+      "step": 57660
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.691702137397593e-05,
+      "loss": 2.1407,
+      "step": 57670
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6914732939722646e-05,
+      "loss": 2.116,
+      "step": 57680
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.691244450546936e-05,
+      "loss": 2.1433,
+      "step": 57690
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6910156071216075e-05,
+      "loss": 2.1158,
+      "step": 57700
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.690786763696279e-05,
+      "loss": 2.0599,
+      "step": 57710
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.69055792027095e-05,
+      "loss": 2.0965,
+      "step": 57720
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6903290768456224e-05,
+      "loss": 2.0685,
+      "step": 57730
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.690100233420294e-05,
+      "loss": 2.1214,
+      "step": 57740
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.689871389994965e-05,
+      "loss": 2.1619,
+      "step": 57750
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.689642546569637e-05,
+      "loss": 2.1479,
+      "step": 57760
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.689413703144309e-05,
+      "loss": 2.0556,
+      "step": 57770
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.68918485971898e-05,
+      "loss": 2.0552,
+      "step": 57780
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.688956016293652e-05,
+      "loss": 2.0345,
+      "step": 57790
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6887271728683236e-05,
+      "loss": 2.1808,
+      "step": 57800
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.688498329442995e-05,
+      "loss": 2.0419,
+      "step": 57810
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.688269486017667e-05,
+      "loss": 2.0243,
+      "step": 57820
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6880406425923385e-05,
+      "loss": 2.0117,
+      "step": 57830
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.68781179916701e-05,
+      "loss": 2.1394,
+      "step": 57840
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.687582955741682e-05,
+      "loss": 2.0451,
+      "step": 57850
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6873541123163535e-05,
+      "loss": 1.9846,
+      "step": 57860
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.687125268891025e-05,
+      "loss": 2.0152,
+      "step": 57870
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.686896425465697e-05,
+      "loss": 2.0128,
+      "step": 57880
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.686667582040368e-05,
+      "loss": 1.9987,
+      "step": 57890
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.68643873861504e-05,
+      "loss": 2.0913,
+      "step": 57900
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.686209895189711e-05,
+      "loss": 2.0131,
+      "step": 57910
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6859810517643826e-05,
+      "loss": 2.1856,
+      "step": 57920
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.685752208339055e-05,
+      "loss": 1.9502,
+      "step": 57930
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.685523364913726e-05,
+      "loss": 2.0931,
+      "step": 57940
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6852945214883975e-05,
+      "loss": 2.1285,
+      "step": 57950
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6850656780630696e-05,
+      "loss": 2.0627,
+      "step": 57960
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.684836834637741e-05,
+      "loss": 2.1386,
+      "step": 57970
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6846079912124124e-05,
+      "loss": 2.0847,
+      "step": 57980
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6843791477870845e-05,
+      "loss": 2.1057,
+      "step": 57990
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.684150304361756e-05,
+      "loss": 1.9698,
+      "step": 58000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6839214609364274e-05,
+      "loss": 1.9382,
+      "step": 58010
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6836926175110995e-05,
+      "loss": 2.1756,
+      "step": 58020
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.683463774085771e-05,
+      "loss": 1.9959,
+      "step": 58030
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.683234930660442e-05,
+      "loss": 2.0233,
+      "step": 58040
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6830060872351144e-05,
+      "loss": 2.0524,
+      "step": 58050
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.682777243809786e-05,
+      "loss": 1.9634,
+      "step": 58060
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.682548400384457e-05,
+      "loss": 2.0909,
+      "step": 58070
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.682319556959129e-05,
+      "loss": 2.0212,
+      "step": 58080
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6820907135338e-05,
+      "loss": 2.0383,
+      "step": 58090
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6818618701084714e-05,
+      "loss": 2.1022,
+      "step": 58100
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6816330266831435e-05,
+      "loss": 2.07,
+      "step": 58110
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.681404183257815e-05,
+      "loss": 2.0416,
+      "step": 58120
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6811753398324864e-05,
+      "loss": 2.2765,
+      "step": 58130
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6809464964071585e-05,
+      "loss": 1.9543,
+      "step": 58140
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.68071765298183e-05,
+      "loss": 1.9225,
+      "step": 58150
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.680488809556501e-05,
+      "loss": 2.0151,
+      "step": 58160
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6802599661311734e-05,
+      "loss": 2.066,
+      "step": 58170
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.680031122705845e-05,
+      "loss": 2.0155,
+      "step": 58180
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.679802279280516e-05,
+      "loss": 2.054,
+      "step": 58190
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.679573435855188e-05,
+      "loss": 2.0897,
+      "step": 58200
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.67934459242986e-05,
+      "loss": 2.007,
+      "step": 58210
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.679115749004531e-05,
+      "loss": 1.9506,
+      "step": 58220
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.678886905579203e-05,
+      "loss": 2.1325,
+      "step": 58230
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6786580621538746e-05,
+      "loss": 2.0709,
+      "step": 58240
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.678429218728546e-05,
+      "loss": 2.0106,
+      "step": 58250
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.678200375303218e-05,
+      "loss": 2.0859,
+      "step": 58260
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6779715318778895e-05,
+      "loss": 1.9996,
+      "step": 58270
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.677742688452561e-05,
+      "loss": 1.9101,
+      "step": 58280
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6775138450272324e-05,
+      "loss": 1.9499,
+      "step": 58290
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.677285001601904e-05,
+      "loss": 2.1478,
+      "step": 58300
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.677056158176576e-05,
+      "loss": 1.9986,
+      "step": 58310
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.676827314751247e-05,
+      "loss": 2.0749,
+      "step": 58320
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.676598471325919e-05,
+      "loss": 1.9843,
+      "step": 58330
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.676369627900591e-05,
+      "loss": 1.9746,
+      "step": 58340
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.676140784475262e-05,
+      "loss": 2.0123,
+      "step": 58350
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6759119410499336e-05,
+      "loss": 2.0119,
+      "step": 58360
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.675683097624606e-05,
+      "loss": 2.1912,
+      "step": 58370
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.675454254199277e-05,
+      "loss": 1.9338,
+      "step": 58380
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6752254107739485e-05,
+      "loss": 2.0327,
+      "step": 58390
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6749965673486206e-05,
+      "loss": 2.0245,
+      "step": 58400
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.674767723923292e-05,
+      "loss": 2.0826,
+      "step": 58410
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6745388804979634e-05,
+      "loss": 1.9785,
+      "step": 58420
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6743100370726355e-05,
+      "loss": 1.9895,
+      "step": 58430
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.674081193647307e-05,
+      "loss": 1.9089,
+      "step": 58440
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6738523502219784e-05,
+      "loss": 1.9278,
+      "step": 58450
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6736235067966505e-05,
+      "loss": 1.9345,
+      "step": 58460
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.673394663371322e-05,
+      "loss": 1.9415,
+      "step": 58470
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.673165819945993e-05,
+      "loss": 2.0436,
+      "step": 58480
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.672936976520665e-05,
+      "loss": 2.0148,
+      "step": 58490
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.672708133095336e-05,
+      "loss": 1.9926,
+      "step": 58500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6724792896700075e-05,
+      "loss": 1.9266,
+      "step": 58510
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6722504462446796e-05,
+      "loss": 2.046,
+      "step": 58520
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.672021602819351e-05,
+      "loss": 2.0374,
+      "step": 58530
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6717927593940224e-05,
+      "loss": 2.0402,
+      "step": 58540
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6715639159686945e-05,
+      "loss": 2.0703,
+      "step": 58550
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.671335072543366e-05,
+      "loss": 1.9894,
+      "step": 58560
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6711062291180374e-05,
+      "loss": 2.0043,
+      "step": 58570
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6708773856927094e-05,
+      "loss": 2.0521,
+      "step": 58580
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.670648542267381e-05,
+      "loss": 2.0144,
+      "step": 58590
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.670419698842052e-05,
+      "loss": 1.9865,
+      "step": 58600
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6701908554167244e-05,
+      "loss": 1.9444,
+      "step": 58610
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.669962011991396e-05,
+      "loss": 2.091,
+      "step": 58620
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.669733168566067e-05,
+      "loss": 2.1615,
+      "step": 58630
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.669504325140739e-05,
+      "loss": 2.1239,
+      "step": 58640
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.669275481715411e-05,
+      "loss": 2.1091,
+      "step": 58650
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.669046638290082e-05,
+      "loss": 1.9792,
+      "step": 58660
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.668817794864754e-05,
+      "loss": 2.0113,
+      "step": 58670
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.668588951439425e-05,
+      "loss": 1.9905,
+      "step": 58680
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6683601080140964e-05,
+      "loss": 2.0136,
+      "step": 58690
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6681312645887684e-05,
+      "loss": 2.0869,
+      "step": 58700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.66790242116344e-05,
+      "loss": 2.1484,
+      "step": 58710
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.667673577738111e-05,
+      "loss": 2.0519,
+      "step": 58720
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6674447343127834e-05,
+      "loss": 2.0373,
+      "step": 58730
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.667215890887455e-05,
+      "loss": 2.0117,
+      "step": 58740
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.666987047462126e-05,
+      "loss": 2.0485,
+      "step": 58750
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.666758204036798e-05,
+      "loss": 2.1632,
+      "step": 58760
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.66652936061147e-05,
+      "loss": 2.0138,
+      "step": 58770
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.666300517186141e-05,
+      "loss": 1.9481,
+      "step": 58780
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.666071673760813e-05,
+      "loss": 2.1327,
+      "step": 58790
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6658428303354846e-05,
+      "loss": 1.9908,
+      "step": 58800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.665613986910156e-05,
+      "loss": 1.9587,
+      "step": 58810
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.665385143484828e-05,
+      "loss": 2.0894,
+      "step": 58820
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6651563000594995e-05,
+      "loss": 2.0452,
+      "step": 58830
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.664927456634171e-05,
+      "loss": 1.9466,
+      "step": 58840
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.664698613208843e-05,
+      "loss": 2.1746,
+      "step": 58850
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6644697697835144e-05,
+      "loss": 2.0489,
+      "step": 58860
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.664240926358186e-05,
+      "loss": 2.1759,
+      "step": 58870
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.664012082932857e-05,
+      "loss": 2.0698,
+      "step": 58880
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.663783239507529e-05,
+      "loss": 2.1073,
+      "step": 58890
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.663554396082201e-05,
+      "loss": 1.953,
+      "step": 58900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.663325552656872e-05,
+      "loss": 2.1363,
+      "step": 58910
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6630967092315436e-05,
+      "loss": 2.0793,
+      "step": 58920
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.662867865806216e-05,
+      "loss": 2.0648,
+      "step": 58930
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.662639022380887e-05,
+      "loss": 2.1383,
+      "step": 58940
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6624101789555585e-05,
+      "loss": 2.0385,
+      "step": 58950
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6621813355302306e-05,
+      "loss": 2.0749,
+      "step": 58960
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.661952492104902e-05,
+      "loss": 2.0044,
+      "step": 58970
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6617236486795734e-05,
+      "loss": 1.9786,
+      "step": 58980
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6614948052542455e-05,
+      "loss": 2.0958,
+      "step": 58990
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.661265961828917e-05,
+      "loss": 2.0193,
+      "step": 59000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6610371184035884e-05,
+      "loss": 1.9841,
+      "step": 59010
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6608082749782604e-05,
+      "loss": 2.1085,
+      "step": 59020
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.660579431552932e-05,
+      "loss": 2.1187,
+      "step": 59030
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.660350588127603e-05,
+      "loss": 2.1762,
+      "step": 59040
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6601217447022754e-05,
+      "loss": 1.9308,
+      "step": 59050
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.659892901276947e-05,
+      "loss": 1.9871,
+      "step": 59060
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.659664057851618e-05,
+      "loss": 1.8631,
+      "step": 59070
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6594352144262896e-05,
+      "loss": 1.9796,
+      "step": 59080
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.659206371000961e-05,
+      "loss": 2.0868,
+      "step": 59090
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6589775275756324e-05,
+      "loss": 2.0347,
+      "step": 59100
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6587486841503045e-05,
+      "loss": 1.8025,
+      "step": 59110
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.658519840724976e-05,
+      "loss": 1.9625,
+      "step": 59120
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6582909972996473e-05,
+      "loss": 2.1019,
+      "step": 59130
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6580621538743194e-05,
+      "loss": 1.9888,
+      "step": 59140
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.657833310448991e-05,
+      "loss": 2.0713,
+      "step": 59150
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.657604467023662e-05,
+      "loss": 2.051,
+      "step": 59160
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6573756235983344e-05,
+      "loss": 1.9585,
+      "step": 59170
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.657146780173006e-05,
+      "loss": 2.1061,
+      "step": 59180
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.656917936747677e-05,
+      "loss": 2.0879,
+      "step": 59190
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.656689093322349e-05,
+      "loss": 1.9454,
+      "step": 59200
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.656460249897021e-05,
+      "loss": 2.132,
+      "step": 59210
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.656231406471692e-05,
+      "loss": 1.9449,
+      "step": 59220
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.656002563046364e-05,
+      "loss": 1.9215,
+      "step": 59230
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6557737196210356e-05,
+      "loss": 2.073,
+      "step": 59240
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.655544876195707e-05,
+      "loss": 1.9028,
+      "step": 59250
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.655316032770379e-05,
+      "loss": 2.0405,
+      "step": 59260
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6550871893450505e-05,
+      "loss": 1.997,
+      "step": 59270
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.654858345919722e-05,
+      "loss": 1.9598,
+      "step": 59280
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6546295024943933e-05,
+      "loss": 1.866,
+      "step": 59290
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.654400659069065e-05,
+      "loss": 2.0209,
+      "step": 59300
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.654171815643737e-05,
+      "loss": 2.1247,
+      "step": 59310
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.653942972218408e-05,
+      "loss": 2.1026,
+      "step": 59320
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.65371412879308e-05,
+      "loss": 1.9478,
+      "step": 59330
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.653485285367752e-05,
+      "loss": 2.0552,
+      "step": 59340
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.653256441942423e-05,
+      "loss": 2.0881,
+      "step": 59350
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6530275985170946e-05,
+      "loss": 2.0333,
+      "step": 59360
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.652798755091767e-05,
+      "loss": 2.0921,
+      "step": 59370
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.652569911666438e-05,
+      "loss": 2.1748,
+      "step": 59380
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6523410682411095e-05,
+      "loss": 1.9012,
+      "step": 59390
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6521122248157816e-05,
+      "loss": 1.9632,
+      "step": 59400
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.651883381390453e-05,
+      "loss": 2.0397,
+      "step": 59410
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6516545379651244e-05,
+      "loss": 1.987,
+      "step": 59420
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6514256945397965e-05,
+      "loss": 2.0182,
+      "step": 59430
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.651196851114468e-05,
+      "loss": 2.098,
+      "step": 59440
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6509680076891394e-05,
+      "loss": 1.9469,
+      "step": 59450
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6507391642638114e-05,
+      "loss": 1.9248,
+      "step": 59460
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.650510320838482e-05,
+      "loss": 1.9711,
+      "step": 59470
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6502814774131536e-05,
+      "loss": 1.909,
+      "step": 59480
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.650052633987826e-05,
+      "loss": 1.998,
+      "step": 59490
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.649823790562497e-05,
+      "loss": 1.9103,
+      "step": 59500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6495949471371685e-05,
+      "loss": 2.1806,
+      "step": 59510
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6493661037118406e-05,
+      "loss": 2.0162,
+      "step": 59520
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.649137260286512e-05,
+      "loss": 1.9769,
+      "step": 59530
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6489084168611834e-05,
+      "loss": 1.8979,
+      "step": 59540
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6486795734358555e-05,
+      "loss": 1.8916,
+      "step": 59550
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.648450730010527e-05,
+      "loss": 2.0353,
+      "step": 59560
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6482218865851983e-05,
+      "loss": 2.0973,
+      "step": 59570
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6479930431598704e-05,
+      "loss": 2.0024,
+      "step": 59580
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.647764199734542e-05,
+      "loss": 2.1468,
+      "step": 59590
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.647535356309213e-05,
+      "loss": 2.1231,
+      "step": 59600
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6473065128838854e-05,
+      "loss": 1.9381,
+      "step": 59610
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.647077669458557e-05,
+      "loss": 1.9635,
+      "step": 59620
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.646848826033228e-05,
+      "loss": 2.0019,
+      "step": 59630
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6466199826079e-05,
+      "loss": 1.9443,
+      "step": 59640
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.646391139182572e-05,
+      "loss": 1.9899,
+      "step": 59650
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.646162295757243e-05,
+      "loss": 1.9079,
+      "step": 59660
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6459334523319145e-05,
+      "loss": 2.0221,
+      "step": 59670
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.645704608906586e-05,
+      "loss": 2.0524,
+      "step": 59680
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.645475765481258e-05,
+      "loss": 1.9912,
+      "step": 59690
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6452469220559294e-05,
+      "loss": 2.0082,
+      "step": 59700
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.645018078630601e-05,
+      "loss": 1.9967,
+      "step": 59710
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.644789235205273e-05,
+      "loss": 2.1516,
+      "step": 59720
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6445603917799443e-05,
+      "loss": 1.9955,
+      "step": 59730
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.644331548354616e-05,
+      "loss": 2.223,
+      "step": 59740
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.644102704929288e-05,
+      "loss": 1.8971,
+      "step": 59750
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.643873861503959e-05,
+      "loss": 2.0708,
+      "step": 59760
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.643645018078631e-05,
+      "loss": 2.0318,
+      "step": 59770
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.643416174653302e-05,
+      "loss": 1.939,
+      "step": 59780
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.643187331227974e-05,
+      "loss": 1.9177,
+      "step": 59790
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6429584878026456e-05,
+      "loss": 1.978,
+      "step": 59800
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.642729644377317e-05,
+      "loss": 2.1113,
+      "step": 59810
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.642500800951989e-05,
+      "loss": 1.9359,
+      "step": 59820
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6422719575266605e-05,
+      "loss": 2.0547,
+      "step": 59830
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.642043114101332e-05,
+      "loss": 1.9669,
+      "step": 59840
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.641814270676004e-05,
+      "loss": 2.021,
+      "step": 59850
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6415854272506754e-05,
+      "loss": 2.049,
+      "step": 59860
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.641356583825347e-05,
+      "loss": 2.0174,
+      "step": 59870
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.641127740400018e-05,
+      "loss": 1.9955,
+      "step": 59880
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.64089889697469e-05,
+      "loss": 1.9543,
+      "step": 59890
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.640670053549362e-05,
+      "loss": 2.0832,
+      "step": 59900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.640441210124033e-05,
+      "loss": 1.9544,
+      "step": 59910
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6402123666987046e-05,
+      "loss": 2.0689,
+      "step": 59920
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.639983523273377e-05,
+      "loss": 2.1265,
+      "step": 59930
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.639754679848048e-05,
+      "loss": 1.8493,
+      "step": 59940
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6395258364227195e-05,
+      "loss": 2.0218,
+      "step": 59950
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6392969929973916e-05,
+      "loss": 1.929,
+      "step": 59960
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.639068149572063e-05,
+      "loss": 2.0547,
+      "step": 59970
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6388393061467344e-05,
+      "loss": 1.9919,
+      "step": 59980
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6386104627214065e-05,
+      "loss": 2.0245,
+      "step": 59990
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.638381619296078e-05,
+      "loss": 1.9682,
+      "step": 60000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6381527758707493e-05,
+      "loss": 1.9584,
+      "step": 60010
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6379239324454214e-05,
+      "loss": 2.0457,
+      "step": 60020
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.637695089020093e-05,
+      "loss": 2.0441,
+      "step": 60030
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.637466245594764e-05,
+      "loss": 2.0462,
+      "step": 60040
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6372374021694363e-05,
+      "loss": 2.0593,
+      "step": 60050
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.637008558744108e-05,
+      "loss": 1.9978,
+      "step": 60060
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6367797153187785e-05,
+      "loss": 2.0878,
+      "step": 60070
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6365508718934506e-05,
+      "loss": 1.9447,
+      "step": 60080
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.636322028468122e-05,
+      "loss": 2.0362,
+      "step": 60090
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6360931850427934e-05,
+      "loss": 2.0925,
+      "step": 60100
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6358643416174655e-05,
+      "loss": 1.9102,
+      "step": 60110
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.635635498192137e-05,
+      "loss": 1.9583,
+      "step": 60120
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.635406654766808e-05,
+      "loss": 1.8651,
+      "step": 60130
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6351778113414804e-05,
+      "loss": 1.9928,
+      "step": 60140
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.634948967916152e-05,
+      "loss": 2.0876,
+      "step": 60150
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.634720124490823e-05,
+      "loss": 1.8945,
+      "step": 60160
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6344912810654953e-05,
+      "loss": 1.7745,
+      "step": 60170
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.634262437640167e-05,
+      "loss": 1.9067,
+      "step": 60180
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.634033594214838e-05,
+      "loss": 1.9254,
+      "step": 60190
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.63380475078951e-05,
+      "loss": 1.8875,
+      "step": 60200
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.633575907364182e-05,
+      "loss": 2.0375,
+      "step": 60210
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.633347063938853e-05,
+      "loss": 1.7968,
+      "step": 60220
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.633118220513525e-05,
+      "loss": 2.1611,
+      "step": 60230
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6328893770881966e-05,
+      "loss": 2.1146,
+      "step": 60240
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.632660533662868e-05,
+      "loss": 1.8987,
+      "step": 60250
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6324316902375394e-05,
+      "loss": 1.8894,
+      "step": 60260
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.632202846812211e-05,
+      "loss": 2.0244,
+      "step": 60270
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.631974003386883e-05,
+      "loss": 2.0659,
+      "step": 60280
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.631745159961554e-05,
+      "loss": 2.0322,
+      "step": 60290
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.631516316536226e-05,
+      "loss": 1.881,
+      "step": 60300
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.631287473110898e-05,
+      "loss": 2.0464,
+      "step": 60310
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.631058629685569e-05,
+      "loss": 1.9628,
+      "step": 60320
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.630829786260241e-05,
+      "loss": 1.8972,
+      "step": 60330
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.630600942834913e-05,
+      "loss": 2.088,
+      "step": 60340
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.630372099409584e-05,
+      "loss": 2.0569,
+      "step": 60350
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6301432559842556e-05,
+      "loss": 1.9792,
+      "step": 60360
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.629914412558928e-05,
+      "loss": 1.949,
+      "step": 60370
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.629685569133599e-05,
+      "loss": 1.9344,
+      "step": 60380
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6294567257082705e-05,
+      "loss": 2.1063,
+      "step": 60390
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6292278822829426e-05,
+      "loss": 1.9026,
+      "step": 60400
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.628999038857614e-05,
+      "loss": 2.0503,
+      "step": 60410
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6287701954322854e-05,
+      "loss": 2.1715,
+      "step": 60420
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6285413520069575e-05,
+      "loss": 2.0045,
+      "step": 60430
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.628312508581629e-05,
+      "loss": 2.0622,
+      "step": 60440
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6280836651563e-05,
+      "loss": 1.934,
+      "step": 60450
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.627854821730972e-05,
+      "loss": 1.8569,
+      "step": 60460
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.627625978305643e-05,
+      "loss": 1.9467,
+      "step": 60470
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6273971348803146e-05,
+      "loss": 2.0438,
+      "step": 60480
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.627168291454987e-05,
+      "loss": 2.0008,
+      "step": 60490
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.626939448029658e-05,
+      "loss": 1.985,
+      "step": 60500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6267106046043295e-05,
+      "loss": 2.0893,
+      "step": 60510
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6264817611790016e-05,
+      "loss": 1.9721,
+      "step": 60520
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.626252917753673e-05,
+      "loss": 1.9079,
+      "step": 60530
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6260240743283444e-05,
+      "loss": 1.9903,
+      "step": 60540
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6257952309030165e-05,
+      "loss": 2.163,
+      "step": 60550
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.625566387477688e-05,
+      "loss": 2.0605,
+      "step": 60560
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.625337544052359e-05,
+      "loss": 1.8531,
+      "step": 60570
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6251087006270314e-05,
+      "loss": 2.0791,
+      "step": 60580
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.624879857201703e-05,
+      "loss": 2.0383,
+      "step": 60590
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.624651013776374e-05,
+      "loss": 1.9614,
+      "step": 60600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6244221703510463e-05,
+      "loss": 2.0337,
+      "step": 60610
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.624193326925718e-05,
+      "loss": 1.9638,
+      "step": 60620
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.623964483500389e-05,
+      "loss": 1.9294,
+      "step": 60630
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.623735640075061e-05,
+      "loss": 2.0662,
+      "step": 60640
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.623506796649733e-05,
+      "loss": 1.9583,
+      "step": 60650
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.623277953224404e-05,
+      "loss": 1.91,
+      "step": 60660
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6230491097990755e-05,
+      "loss": 1.8304,
+      "step": 60670
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.622820266373747e-05,
+      "loss": 2.011,
+      "step": 60680
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.622591422948419e-05,
+      "loss": 1.9931,
+      "step": 60690
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6223625795230904e-05,
+      "loss": 2.157,
+      "step": 60700
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.622133736097762e-05,
+      "loss": 2.0211,
+      "step": 60710
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.621904892672434e-05,
+      "loss": 1.9471,
+      "step": 60720
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.621676049247105e-05,
+      "loss": 1.857,
+      "step": 60730
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.621447205821777e-05,
+      "loss": 1.9129,
+      "step": 60740
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.621218362396449e-05,
+      "loss": 1.8573,
+      "step": 60750
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.62098951897112e-05,
+      "loss": 1.9954,
+      "step": 60760
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.620760675545792e-05,
+      "loss": 1.9688,
+      "step": 60770
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.620531832120464e-05,
+      "loss": 2.0694,
+      "step": 60780
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.620302988695135e-05,
+      "loss": 1.9621,
+      "step": 60790
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6200741452698066e-05,
+      "loss": 2.0583,
+      "step": 60800
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.619845301844479e-05,
+      "loss": 1.9476,
+      "step": 60810
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.61961645841915e-05,
+      "loss": 1.9973,
+      "step": 60820
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6193876149938215e-05,
+      "loss": 1.9639,
+      "step": 60830
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6191587715684936e-05,
+      "loss": 2.0196,
+      "step": 60840
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.618929928143165e-05,
+      "loss": 2.1245,
+      "step": 60850
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.618701084717836e-05,
+      "loss": 2.0029,
+      "step": 60860
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.618472241292508e-05,
+      "loss": 1.9786,
+      "step": 60870
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.618243397867179e-05,
+      "loss": 2.0416,
+      "step": 60880
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6180145544418507e-05,
+      "loss": 1.945,
+      "step": 60890
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.617785711016523e-05,
+      "loss": 2.0444,
+      "step": 60900
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.617556867591194e-05,
+      "loss": 2.0061,
+      "step": 60910
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6173280241658656e-05,
+      "loss": 1.9984,
+      "step": 60920
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.617099180740538e-05,
+      "loss": 1.8808,
+      "step": 60930
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.616870337315209e-05,
+      "loss": 1.974,
+      "step": 60940
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6166414938898805e-05,
+      "loss": 2.0278,
+      "step": 60950
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6164126504645526e-05,
+      "loss": 1.8631,
+      "step": 60960
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.616183807039224e-05,
+      "loss": 1.924,
+      "step": 60970
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6159549636138954e-05,
+      "loss": 2.006,
+      "step": 60980
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6157261201885675e-05,
+      "loss": 1.955,
+      "step": 60990
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.615497276763239e-05,
+      "loss": 1.9427,
+      "step": 61000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.61526843333791e-05,
+      "loss": 1.9409,
+      "step": 61010
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6150395899125824e-05,
+      "loss": 1.9506,
+      "step": 61020
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.614810746487254e-05,
+      "loss": 1.8466,
+      "step": 61030
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.614581903061925e-05,
+      "loss": 1.943,
+      "step": 61040
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6143530596365967e-05,
+      "loss": 1.9642,
+      "step": 61050
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.614124216211268e-05,
+      "loss": 2.1687,
+      "step": 61060
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6138953727859395e-05,
+      "loss": 1.925,
+      "step": 61070
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6136665293606116e-05,
+      "loss": 2.0432,
+      "step": 61080
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.613437685935283e-05,
+      "loss": 2.1505,
+      "step": 61090
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6132088425099544e-05,
+      "loss": 1.9926,
+      "step": 61100
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6129799990846265e-05,
+      "loss": 2.1669,
+      "step": 61110
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.612751155659298e-05,
+      "loss": 1.9846,
+      "step": 61120
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.612522312233969e-05,
+      "loss": 1.9489,
+      "step": 61130
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6122934688086414e-05,
+      "loss": 2.0555,
+      "step": 61140
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.612064625383313e-05,
+      "loss": 2.0091,
+      "step": 61150
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.611835781957984e-05,
+      "loss": 1.9568,
+      "step": 61160
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.611606938532656e-05,
+      "loss": 1.9226,
+      "step": 61170
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.611378095107328e-05,
+      "loss": 2.1924,
+      "step": 61180
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.611149251681999e-05,
+      "loss": 1.8953,
+      "step": 61190
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.610920408256671e-05,
+      "loss": 2.0938,
+      "step": 61200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6106915648313427e-05,
+      "loss": 2.0289,
+      "step": 61210
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.610462721406014e-05,
+      "loss": 2.0128,
+      "step": 61220
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.610233877980686e-05,
+      "loss": 2.1055,
+      "step": 61230
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6100050345553576e-05,
+      "loss": 1.8905,
+      "step": 61240
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.609776191130029e-05,
+      "loss": 2.014,
+      "step": 61250
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6095473477047004e-05,
+      "loss": 2.1216,
+      "step": 61260
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.609318504279372e-05,
+      "loss": 1.9205,
+      "step": 61270
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.609089660854044e-05,
+      "loss": 1.954,
+      "step": 61280
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.608860817428715e-05,
+      "loss": 1.9841,
+      "step": 61290
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.608631974003387e-05,
+      "loss": 1.9075,
+      "step": 61300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.608403130578059e-05,
+      "loss": 1.9314,
+      "step": 61310
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.60817428715273e-05,
+      "loss": 2.0757,
+      "step": 61320
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6079454437274017e-05,
+      "loss": 2.0854,
+      "step": 61330
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.607716600302074e-05,
+      "loss": 2.0061,
+      "step": 61340
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.607487756876745e-05,
+      "loss": 1.8823,
+      "step": 61350
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6072589134514166e-05,
+      "loss": 1.9387,
+      "step": 61360
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.607030070026089e-05,
+      "loss": 2.0842,
+      "step": 61370
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.60680122660076e-05,
+      "loss": 1.9555,
+      "step": 61380
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6065723831754315e-05,
+      "loss": 2.0182,
+      "step": 61390
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6063435397501036e-05,
+      "loss": 2.1223,
+      "step": 61400
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.606114696324775e-05,
+      "loss": 2.0476,
+      "step": 61410
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6058858528994464e-05,
+      "loss": 2.0908,
+      "step": 61420
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6056570094741185e-05,
+      "loss": 1.9769,
+      "step": 61430
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.60542816604879e-05,
+      "loss": 2.0315,
+      "step": 61440
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6051993226234606e-05,
+      "loss": 2.0142,
+      "step": 61450
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.604970479198133e-05,
+      "loss": 1.8936,
+      "step": 61460
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.604741635772804e-05,
+      "loss": 2.0686,
+      "step": 61470
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6045127923474756e-05,
+      "loss": 2.04,
+      "step": 61480
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6042839489221477e-05,
+      "loss": 2.0724,
+      "step": 61490
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.604055105496819e-05,
+      "loss": 2.013,
+      "step": 61500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6038262620714905e-05,
+      "loss": 1.9822,
+      "step": 61510
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6035974186461626e-05,
+      "loss": 2.0616,
+      "step": 61520
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.603368575220834e-05,
+      "loss": 2.1496,
+      "step": 61530
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6031397317955054e-05,
+      "loss": 1.9661,
+      "step": 61540
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6029108883701775e-05,
+      "loss": 1.8144,
+      "step": 61550
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.602682044944849e-05,
+      "loss": 1.8169,
+      "step": 61560
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.60245320151952e-05,
+      "loss": 1.9258,
+      "step": 61570
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6022243580941924e-05,
+      "loss": 2.1131,
+      "step": 61580
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.601995514668864e-05,
+      "loss": 1.8251,
+      "step": 61590
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.601766671243535e-05,
+      "loss": 1.8831,
+      "step": 61600
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.601537827818207e-05,
+      "loss": 2.0656,
+      "step": 61610
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.601308984392879e-05,
+      "loss": 1.99,
+      "step": 61620
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.60108014096755e-05,
+      "loss": 1.9731,
+      "step": 61630
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.600851297542222e-05,
+      "loss": 1.8577,
+      "step": 61640
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.600622454116893e-05,
+      "loss": 1.8428,
+      "step": 61650
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.600393610691565e-05,
+      "loss": 2.0219,
+      "step": 61660
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6001647672662365e-05,
+      "loss": 1.9814,
+      "step": 61670
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.599935923840908e-05,
+      "loss": 2.0118,
+      "step": 61680
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.59970708041558e-05,
+      "loss": 2.0391,
+      "step": 61690
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5994782369902514e-05,
+      "loss": 1.9902,
+      "step": 61700
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.599249393564923e-05,
+      "loss": 1.8667,
+      "step": 61710
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.599020550139595e-05,
+      "loss": 2.0817,
+      "step": 61720
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.598791706714266e-05,
+      "loss": 1.9757,
+      "step": 61730
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.598562863288938e-05,
+      "loss": 1.9032,
+      "step": 61740
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.59833401986361e-05,
+      "loss": 1.9409,
+      "step": 61750
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.598105176438281e-05,
+      "loss": 1.9412,
+      "step": 61760
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5978763330129527e-05,
+      "loss": 2.0815,
+      "step": 61770
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.597647489587625e-05,
+      "loss": 1.9708,
+      "step": 61780
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.597418646162296e-05,
+      "loss": 1.9773,
+      "step": 61790
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5971898027369676e-05,
+      "loss": 1.9472,
+      "step": 61800
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5969609593116397e-05,
+      "loss": 1.953,
+      "step": 61810
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.596732115886311e-05,
+      "loss": 1.9493,
+      "step": 61820
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5965032724609825e-05,
+      "loss": 1.9448,
+      "step": 61830
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.596274429035654e-05,
+      "loss": 2.0161,
+      "step": 61840
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.596045585610325e-05,
+      "loss": 1.8877,
+      "step": 61850
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.595816742184997e-05,
+      "loss": 1.9639,
+      "step": 61860
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.595587898759669e-05,
+      "loss": 1.9716,
+      "step": 61870
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.59535905533434e-05,
+      "loss": 1.896,
+      "step": 61880
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5951302119090116e-05,
+      "loss": 2.0705,
+      "step": 61890
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.594901368483684e-05,
+      "loss": 1.9875,
+      "step": 61900
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.594672525058355e-05,
+      "loss": 1.9054,
+      "step": 61910
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5944436816330266e-05,
+      "loss": 1.934,
+      "step": 61920
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5942148382076987e-05,
+      "loss": 1.8307,
+      "step": 61930
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.59398599478237e-05,
+      "loss": 1.897,
+      "step": 61940
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5937571513570415e-05,
+      "loss": 1.949,
+      "step": 61950
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5935283079317136e-05,
+      "loss": 1.9394,
+      "step": 61960
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.593299464506385e-05,
+      "loss": 2.0158,
+      "step": 61970
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5930706210810564e-05,
+      "loss": 1.9855,
+      "step": 61980
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5928417776557285e-05,
+      "loss": 1.8759,
+      "step": 61990
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5926129342304e-05,
+      "loss": 1.991,
+      "step": 62000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.592384090805071e-05,
+      "loss": 2.0799,
+      "step": 62010
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5921552473797434e-05,
+      "loss": 2.1465,
+      "step": 62020
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.591926403954415e-05,
+      "loss": 2.0323,
+      "step": 62030
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.591697560529086e-05,
+      "loss": 1.8978,
+      "step": 62040
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5914687171037576e-05,
+      "loss": 1.8778,
+      "step": 62050
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.591239873678429e-05,
+      "loss": 1.9978,
+      "step": 62060
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.591011030253101e-05,
+      "loss": 1.9031,
+      "step": 62070
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5907821868277726e-05,
+      "loss": 1.8336,
+      "step": 62080
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.590553343402444e-05,
+      "loss": 1.9167,
+      "step": 62090
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.590324499977116e-05,
+      "loss": 1.9665,
+      "step": 62100
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5900956565517875e-05,
+      "loss": 1.8544,
+      "step": 62110
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.589866813126459e-05,
+      "loss": 1.9748,
+      "step": 62120
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.58963796970113e-05,
+      "loss": 1.9473,
+      "step": 62130
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5894091262758024e-05,
+      "loss": 2.0598,
+      "step": 62140
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.589180282850474e-05,
+      "loss": 1.8982,
+      "step": 62150
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.588951439425145e-05,
+      "loss": 1.9716,
+      "step": 62160
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.588722595999817e-05,
+      "loss": 1.9182,
+      "step": 62170
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.588493752574489e-05,
+      "loss": 1.9881,
+      "step": 62180
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.58826490914916e-05,
+      "loss": 1.947,
+      "step": 62190
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.588036065723832e-05,
+      "loss": 1.9551,
+      "step": 62200
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5878072222985036e-05,
+      "loss": 1.9064,
+      "step": 62210
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.587578378873175e-05,
+      "loss": 1.973,
+      "step": 62220
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.587349535447847e-05,
+      "loss": 1.9656,
+      "step": 62230
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.587120692022518e-05,
+      "loss": 1.8818,
+      "step": 62240
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.58689184859719e-05,
+      "loss": 2.0272,
+      "step": 62250
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5866630051718614e-05,
+      "loss": 1.9223,
+      "step": 62260
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.586434161746533e-05,
+      "loss": 2.0453,
+      "step": 62270
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.586205318321205e-05,
+      "loss": 1.8061,
+      "step": 62280
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.585976474895876e-05,
+      "loss": 2.005,
+      "step": 62290
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.585747631470548e-05,
+      "loss": 1.7388,
+      "step": 62300
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.58551878804522e-05,
+      "loss": 1.8706,
+      "step": 62310
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.585289944619891e-05,
+      "loss": 1.9684,
+      "step": 62320
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5850611011945626e-05,
+      "loss": 1.942,
+      "step": 62330
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.584832257769235e-05,
+      "loss": 1.8568,
+      "step": 62340
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.584603414343906e-05,
+      "loss": 1.9895,
+      "step": 62350
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5843745709185776e-05,
+      "loss": 1.9855,
+      "step": 62360
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5841457274932496e-05,
+      "loss": 1.883,
+      "step": 62370
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.583916884067921e-05,
+      "loss": 1.9529,
+      "step": 62380
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5836880406425925e-05,
+      "loss": 1.7964,
+      "step": 62390
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5834591972172646e-05,
+      "loss": 1.9558,
+      "step": 62400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.583230353791936e-05,
+      "loss": 2.1253,
+      "step": 62410
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5830015103666074e-05,
+      "loss": 1.9071,
+      "step": 62420
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5827726669412795e-05,
+      "loss": 1.8805,
+      "step": 62430
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.58254382351595e-05,
+      "loss": 1.8454,
+      "step": 62440
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5823149800906216e-05,
+      "loss": 2.0848,
+      "step": 62450
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.582086136665294e-05,
+      "loss": 1.8935,
+      "step": 62460
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.581857293239965e-05,
+      "loss": 2.0187,
+      "step": 62470
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5816284498146366e-05,
+      "loss": 1.7974,
+      "step": 62480
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5813996063893086e-05,
+      "loss": 2.0467,
+      "step": 62490
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.58117076296398e-05,
+      "loss": 1.9387,
+      "step": 62500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5809419195386515e-05,
+      "loss": 1.7828,
+      "step": 62510
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5807130761133236e-05,
+      "loss": 1.9787,
+      "step": 62520
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.580484232687995e-05,
+      "loss": 1.9266,
+      "step": 62530
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5802553892626664e-05,
+      "loss": 1.9047,
+      "step": 62540
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5800265458373385e-05,
+      "loss": 1.8257,
+      "step": 62550
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.57979770241201e-05,
+      "loss": 2.0659,
+      "step": 62560
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.579568858986681e-05,
+      "loss": 1.9568,
+      "step": 62570
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5793400155613534e-05,
+      "loss": 1.9103,
+      "step": 62580
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.579111172136025e-05,
+      "loss": 1.7248,
+      "step": 62590
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.578882328710696e-05,
+      "loss": 1.7813,
+      "step": 62600
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.578653485285368e-05,
+      "loss": 1.883,
+      "step": 62610
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.57842464186004e-05,
+      "loss": 2.0126,
+      "step": 62620
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.578195798434711e-05,
+      "loss": 2.0162,
+      "step": 62630
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5779669550093826e-05,
+      "loss": 1.9926,
+      "step": 62640
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.577738111584054e-05,
+      "loss": 2.0862,
+      "step": 62650
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.577509268158726e-05,
+      "loss": 1.961,
+      "step": 62660
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5772804247333975e-05,
+      "loss": 1.9156,
+      "step": 62670
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.577051581308069e-05,
+      "loss": 1.9592,
+      "step": 62680
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.576822737882741e-05,
+      "loss": 1.9383,
+      "step": 62690
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5765938944574124e-05,
+      "loss": 2.0883,
+      "step": 62700
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.576365051032084e-05,
+      "loss": 2.0203,
+      "step": 62710
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.576136207606756e-05,
+      "loss": 1.926,
+      "step": 62720
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.575907364181427e-05,
+      "loss": 2.1409,
+      "step": 62730
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.575678520756099e-05,
+      "loss": 1.9232,
+      "step": 62740
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.575449677330771e-05,
+      "loss": 1.8628,
+      "step": 62750
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.575220833905442e-05,
+      "loss": 2.0279,
+      "step": 62760
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5749919904801136e-05,
+      "loss": 1.9012,
+      "step": 62770
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.574763147054786e-05,
+      "loss": 2.0143,
+      "step": 62780
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.574534303629457e-05,
+      "loss": 1.8241,
+      "step": 62790
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5743054602041286e-05,
+      "loss": 1.9759,
+      "step": 62800
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5740766167788006e-05,
+      "loss": 1.7345,
+      "step": 62810
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.573847773353472e-05,
+      "loss": 1.9398,
+      "step": 62820
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.573618929928143e-05,
+      "loss": 2.0881,
+      "step": 62830
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.573390086502815e-05,
+      "loss": 1.9504,
+      "step": 62840
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.573161243077486e-05,
+      "loss": 1.9129,
+      "step": 62850
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.572932399652158e-05,
+      "loss": 1.797,
+      "step": 62860
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.57270355622683e-05,
+      "loss": 1.9459,
+      "step": 62870
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.572474712801501e-05,
+      "loss": 1.9564,
+      "step": 62880
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5722458693761726e-05,
+      "loss": 1.8502,
+      "step": 62890
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.572017025950845e-05,
+      "loss": 1.9123,
+      "step": 62900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.571788182525516e-05,
+      "loss": 2.0073,
+      "step": 62910
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5715593391001875e-05,
+      "loss": 1.9719,
+      "step": 62920
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5713304956748596e-05,
+      "loss": 1.9162,
+      "step": 62930
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.571101652249531e-05,
+      "loss": 1.9261,
+      "step": 62940
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5708728088242025e-05,
+      "loss": 1.9868,
+      "step": 62950
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5706439653988746e-05,
+      "loss": 1.762,
+      "step": 62960
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.570415121973546e-05,
+      "loss": 2.0028,
+      "step": 62970
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5701862785482174e-05,
+      "loss": 1.9391,
+      "step": 62980
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5699574351228895e-05,
+      "loss": 1.9854,
+      "step": 62990
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.569728591697561e-05,
+      "loss": 1.8794,
+      "step": 63000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.569499748272232e-05,
+      "loss": 2.0347,
+      "step": 63010
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5692709048469044e-05,
+      "loss": 1.9397,
+      "step": 63020
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.569042061421575e-05,
+      "loss": 1.9203,
+      "step": 63030
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.568813217996247e-05,
+      "loss": 1.8024,
+      "step": 63040
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5685843745709186e-05,
+      "loss": 2.0544,
+      "step": 63050
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.56835553114559e-05,
+      "loss": 1.8847,
+      "step": 63060
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.568126687720262e-05,
+      "loss": 1.9498,
+      "step": 63070
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5678978442949335e-05,
+      "loss": 1.7492,
+      "step": 63080
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.567669000869605e-05,
+      "loss": 1.8658,
+      "step": 63090
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.567440157444277e-05,
+      "loss": 1.8613,
+      "step": 63100
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5672113140189485e-05,
+      "loss": 1.7266,
+      "step": 63110
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.56698247059362e-05,
+      "loss": 1.8717,
+      "step": 63120
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.566753627168292e-05,
+      "loss": 2.0591,
+      "step": 63130
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5665247837429634e-05,
+      "loss": 1.8792,
+      "step": 63140
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.566295940317635e-05,
+      "loss": 2.0118,
+      "step": 63150
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.566067096892307e-05,
+      "loss": 1.9258,
+      "step": 63160
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.565838253466978e-05,
+      "loss": 1.8229,
+      "step": 63170
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.56560941004165e-05,
+      "loss": 1.8949,
+      "step": 63180
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.565380566616322e-05,
+      "loss": 1.8369,
+      "step": 63190
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.565151723190993e-05,
+      "loss": 1.9096,
+      "step": 63200
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5649228797656646e-05,
+      "loss": 2.0014,
+      "step": 63210
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.564694036340336e-05,
+      "loss": 1.9616,
+      "step": 63220
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5644651929150075e-05,
+      "loss": 2.0339,
+      "step": 63230
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.564236349489679e-05,
+      "loss": 1.8768,
+      "step": 63240
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.564007506064351e-05,
+      "loss": 2.0305,
+      "step": 63250
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5637786626390224e-05,
+      "loss": 1.805,
+      "step": 63260
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.563549819213694e-05,
+      "loss": 2.1142,
+      "step": 63270
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.563320975788366e-05,
+      "loss": 1.8948,
+      "step": 63280
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.563092132363037e-05,
+      "loss": 1.8392,
+      "step": 63290
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.562863288937709e-05,
+      "loss": 1.8652,
+      "step": 63300
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.562634445512381e-05,
+      "loss": 1.7448,
+      "step": 63310
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.562405602087052e-05,
+      "loss": 1.9783,
+      "step": 63320
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5621767586617236e-05,
+      "loss": 2.072,
+      "step": 63330
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.561947915236396e-05,
+      "loss": 2.1366,
+      "step": 63340
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.561719071811067e-05,
+      "loss": 1.8177,
+      "step": 63350
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5614902283857385e-05,
+      "loss": 1.7811,
+      "step": 63360
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5612613849604106e-05,
+      "loss": 1.8929,
+      "step": 63370
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.561032541535082e-05,
+      "loss": 2.0066,
+      "step": 63380
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5608036981097535e-05,
+      "loss": 1.9763,
+      "step": 63390
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5605748546844256e-05,
+      "loss": 1.7912,
+      "step": 63400
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.560346011259097e-05,
+      "loss": 1.9465,
+      "step": 63410
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.560117167833768e-05,
+      "loss": 1.8102,
+      "step": 63420
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.55988832440844e-05,
+      "loss": 1.9451,
+      "step": 63430
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.559659480983111e-05,
+      "loss": 1.8815,
+      "step": 63440
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5594306375577826e-05,
+      "loss": 1.9561,
+      "step": 63450
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.559201794132455e-05,
+      "loss": 1.8999,
+      "step": 63460
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.558972950707126e-05,
+      "loss": 1.9295,
+      "step": 63470
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5587441072817975e-05,
+      "loss": 1.9766,
+      "step": 63480
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5585152638564696e-05,
+      "loss": 1.9353,
+      "step": 63490
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.558286420431141e-05,
+      "loss": 1.8066,
+      "step": 63500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5580575770058125e-05,
+      "loss": 2.0879,
+      "step": 63510
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5578287335804845e-05,
+      "loss": 1.8655,
+      "step": 63520
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.557599890155156e-05,
+      "loss": 1.8894,
+      "step": 63530
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5573710467298274e-05,
+      "loss": 2.0968,
+      "step": 63540
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5571422033044995e-05,
+      "loss": 1.9822,
+      "step": 63550
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.556913359879171e-05,
+      "loss": 1.8557,
+      "step": 63560
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.556684516453842e-05,
+      "loss": 1.9945,
+      "step": 63570
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5564556730285144e-05,
+      "loss": 1.9027,
+      "step": 63580
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.556226829603186e-05,
+      "loss": 1.9573,
+      "step": 63590
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.555997986177857e-05,
+      "loss": 1.9271,
+      "step": 63600
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.555769142752529e-05,
+      "loss": 2.0693,
+      "step": 63610
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5555402993272e-05,
+      "loss": 1.9814,
+      "step": 63620
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.555311455901872e-05,
+      "loss": 2.0334,
+      "step": 63630
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5550826124765435e-05,
+      "loss": 1.8852,
+      "step": 63640
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.554853769051215e-05,
+      "loss": 1.5322,
+      "step": 63650
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.554624925625887e-05,
+      "loss": 1.8944,
+      "step": 63660
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5543960822005585e-05,
+      "loss": 2.0169,
+      "step": 63670
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.55416723877523e-05,
+      "loss": 2.0145,
+      "step": 63680
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.553938395349902e-05,
+      "loss": 1.9479,
+      "step": 63690
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5537095519245734e-05,
+      "loss": 1.9101,
+      "step": 63700
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.553480708499245e-05,
+      "loss": 1.7812,
+      "step": 63710
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.553251865073917e-05,
+      "loss": 1.7926,
+      "step": 63720
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.553023021648588e-05,
+      "loss": 1.9802,
+      "step": 63730
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.55279417822326e-05,
+      "loss": 1.9092,
+      "step": 63740
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.552565334797932e-05,
+      "loss": 1.942,
+      "step": 63750
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.552336491372603e-05,
+      "loss": 1.7755,
+      "step": 63760
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5521076479472746e-05,
+      "loss": 1.9491,
+      "step": 63770
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.551878804521947e-05,
+      "loss": 1.9567,
+      "step": 63780
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.551649961096618e-05,
+      "loss": 1.9921,
+      "step": 63790
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5514211176712895e-05,
+      "loss": 2.0343,
+      "step": 63800
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5511922742459616e-05,
+      "loss": 1.9224,
+      "step": 63810
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5509634308206324e-05,
+      "loss": 1.8223,
+      "step": 63820
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.550734587395304e-05,
+      "loss": 1.969,
+      "step": 63830
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.550505743969976e-05,
+      "loss": 2.0363,
+      "step": 63840
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.550276900544647e-05,
+      "loss": 1.886,
+      "step": 63850
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.550048057119319e-05,
+      "loss": 1.9836,
+      "step": 63860
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.549819213693991e-05,
+      "loss": 1.7232,
+      "step": 63870
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.549590370268662e-05,
+      "loss": 1.878,
+      "step": 63880
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5493615268433336e-05,
+      "loss": 1.9552,
+      "step": 63890
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.549132683418006e-05,
+      "loss": 1.7719,
+      "step": 63900
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.548903839992677e-05,
+      "loss": 1.7916,
+      "step": 63910
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5486749965673485e-05,
+      "loss": 2.0048,
+      "step": 63920
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5484461531420206e-05,
+      "loss": 2.0018,
+      "step": 63930
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.548217309716692e-05,
+      "loss": 1.9818,
+      "step": 63940
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5479884662913635e-05,
+      "loss": 1.9747,
+      "step": 63950
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5477596228660355e-05,
+      "loss": 1.9482,
+      "step": 63960
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.547530779440707e-05,
+      "loss": 2.0887,
+      "step": 63970
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5473019360153784e-05,
+      "loss": 1.9064,
+      "step": 63980
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5470730925900505e-05,
+      "loss": 1.8339,
+      "step": 63990
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.546844249164722e-05,
+      "loss": 1.8523,
+      "step": 64000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.546615405739393e-05,
+      "loss": 1.7713,
+      "step": 64010
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.546386562314065e-05,
+      "loss": 1.8322,
+      "step": 64020
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.546157718888736e-05,
+      "loss": 2.1607,
+      "step": 64030
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.545928875463408e-05,
+      "loss": 1.9986,
+      "step": 64040
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5457000320380796e-05,
+      "loss": 2.0054,
+      "step": 64050
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.545471188612751e-05,
+      "loss": 2.0537,
+      "step": 64060
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.545242345187423e-05,
+      "loss": 1.9387,
+      "step": 64070
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5450135017620945e-05,
+      "loss": 1.8831,
+      "step": 64080
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.544784658336766e-05,
+      "loss": 1.9297,
+      "step": 64090
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.544555814911438e-05,
+      "loss": 1.8759,
+      "step": 64100
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5443269714861095e-05,
+      "loss": 2.0205,
+      "step": 64110
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.544098128060781e-05,
+      "loss": 1.8869,
+      "step": 64120
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.543869284635453e-05,
+      "loss": 1.9156,
+      "step": 64130
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5436404412101244e-05,
+      "loss": 1.8772,
+      "step": 64140
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.543411597784796e-05,
+      "loss": 1.9062,
+      "step": 64150
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.543182754359468e-05,
+      "loss": 2.0051,
+      "step": 64160
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.542953910934139e-05,
+      "loss": 2.1032,
+      "step": 64170
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.542725067508811e-05,
+      "loss": 1.8948,
+      "step": 64180
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.542496224083483e-05,
+      "loss": 1.9235,
+      "step": 64190
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.542267380658154e-05,
+      "loss": 1.8876,
+      "step": 64200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.542038537232825e-05,
+      "loss": 1.6432,
+      "step": 64210
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.541809693807497e-05,
+      "loss": 1.8401,
+      "step": 64220
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5415808503821684e-05,
+      "loss": 1.8995,
+      "step": 64230
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.54135200695684e-05,
+      "loss": 1.7003,
+      "step": 64240
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.541123163531512e-05,
+      "loss": 1.8123,
+      "step": 64250
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5408943201061834e-05,
+      "loss": 1.8025,
+      "step": 64260
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.540665476680855e-05,
+      "loss": 1.9615,
+      "step": 64270
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.540436633255527e-05,
+      "loss": 2.1013,
+      "step": 64280
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.540207789830198e-05,
+      "loss": 1.949,
+      "step": 64290
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.53997894640487e-05,
+      "loss": 1.9882,
+      "step": 64300
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.539750102979542e-05,
+      "loss": 1.8805,
+      "step": 64310
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.539521259554213e-05,
+      "loss": 1.8193,
+      "step": 64320
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5392924161288846e-05,
+      "loss": 1.8311,
+      "step": 64330
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.539063572703557e-05,
+      "loss": 1.7831,
+      "step": 64340
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.538834729278228e-05,
+      "loss": 1.9034,
+      "step": 64350
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5386058858528995e-05,
+      "loss": 1.8098,
+      "step": 64360
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5383770424275716e-05,
+      "loss": 1.9074,
+      "step": 64370
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.538148199002243e-05,
+      "loss": 2.1487,
+      "step": 64380
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5379193555769144e-05,
+      "loss": 1.9166,
+      "step": 64390
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5376905121515865e-05,
+      "loss": 1.8936,
+      "step": 64400
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.537461668726257e-05,
+      "loss": 1.8111,
+      "step": 64410
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5372328253009294e-05,
+      "loss": 2.1563,
+      "step": 64420
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.537003981875601e-05,
+      "loss": 2.0387,
+      "step": 64430
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.536775138450272e-05,
+      "loss": 1.9323,
+      "step": 64440
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5365462950249436e-05,
+      "loss": 1.9477,
+      "step": 64450
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.536317451599616e-05,
+      "loss": 1.9427,
+      "step": 64460
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.536088608174287e-05,
+      "loss": 1.7879,
+      "step": 64470
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5358597647489585e-05,
+      "loss": 1.876,
+      "step": 64480
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5356309213236306e-05,
+      "loss": 2.0496,
+      "step": 64490
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.535402077898302e-05,
+      "loss": 1.893,
+      "step": 64500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5351732344729734e-05,
+      "loss": 1.9023,
+      "step": 64510
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5349443910476455e-05,
+      "loss": 2.0654,
+      "step": 64520
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.534715547622317e-05,
+      "loss": 2.1372,
+      "step": 64530
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5344867041969884e-05,
+      "loss": 2.0347,
+      "step": 64540
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5342578607716605e-05,
+      "loss": 1.9693,
+      "step": 64550
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.534029017346332e-05,
+      "loss": 1.7586,
+      "step": 64560
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.533800173921003e-05,
+      "loss": 1.8207,
+      "step": 64570
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5335713304956754e-05,
+      "loss": 1.9588,
+      "step": 64580
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.533342487070347e-05,
+      "loss": 1.6623,
+      "step": 64590
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.533113643645018e-05,
+      "loss": 1.9133,
+      "step": 64600
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5328848002196896e-05,
+      "loss": 1.9299,
+      "step": 64610
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.532655956794361e-05,
+      "loss": 2.0485,
+      "step": 64620
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.532427113369033e-05,
+      "loss": 1.9936,
+      "step": 64630
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5321982699437045e-05,
+      "loss": 1.735,
+      "step": 64640
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.531969426518376e-05,
+      "loss": 1.8677,
+      "step": 64650
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.531740583093048e-05,
+      "loss": 1.9971,
+      "step": 64660
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5315117396677194e-05,
+      "loss": 1.8924,
+      "step": 64670
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.531282896242391e-05,
+      "loss": 1.9155,
+      "step": 64680
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.531054052817063e-05,
+      "loss": 1.9737,
+      "step": 64690
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5308252093917344e-05,
+      "loss": 2.0032,
+      "step": 64700
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.530596365966406e-05,
+      "loss": 2.058,
+      "step": 64710
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.530367522541078e-05,
+      "loss": 1.8235,
+      "step": 64720
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.530138679115749e-05,
+      "loss": 1.8421,
+      "step": 64730
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.529909835690421e-05,
+      "loss": 2.0105,
+      "step": 64740
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.529680992265093e-05,
+      "loss": 1.8812,
+      "step": 64750
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.529452148839764e-05,
+      "loss": 2.0635,
+      "step": 64760
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5292233054144356e-05,
+      "loss": 1.7655,
+      "step": 64770
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.528994461989108e-05,
+      "loss": 1.995,
+      "step": 64780
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.528765618563779e-05,
+      "loss": 1.7921,
+      "step": 64790
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5285367751384505e-05,
+      "loss": 1.8284,
+      "step": 64800
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.528307931713122e-05,
+      "loss": 1.6364,
+      "step": 64810
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5280790882877934e-05,
+      "loss": 1.784,
+      "step": 64820
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.527850244862465e-05,
+      "loss": 1.9723,
+      "step": 64830
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.527621401437137e-05,
+      "loss": 2.0987,
+      "step": 64840
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.527392558011808e-05,
+      "loss": 1.7028,
+      "step": 64850
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.52716371458648e-05,
+      "loss": 1.9571,
+      "step": 64860
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.526934871161152e-05,
+      "loss": 1.8759,
+      "step": 64870
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.526706027735823e-05,
+      "loss": 1.6979,
+      "step": 64880
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5264771843104946e-05,
+      "loss": 1.8347,
+      "step": 64890
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.526248340885167e-05,
+      "loss": 1.7995,
+      "step": 64900
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.526019497459838e-05,
+      "loss": 1.822,
+      "step": 64910
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5257906540345095e-05,
+      "loss": 1.9235,
+      "step": 64920
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5255618106091816e-05,
+      "loss": 1.8268,
+      "step": 64930
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.525332967183853e-05,
+      "loss": 1.8438,
+      "step": 64940
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5251041237585244e-05,
+      "loss": 2.0091,
+      "step": 64950
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5248752803331965e-05,
+      "loss": 1.9503,
+      "step": 64960
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.524646436907868e-05,
+      "loss": 1.7614,
+      "step": 64970
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5244175934825394e-05,
+      "loss": 1.7522,
+      "step": 64980
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5241887500572114e-05,
+      "loss": 1.7531,
+      "step": 64990
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.523959906631882e-05,
+      "loss": 1.8558,
+      "step": 65000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.523731063206554e-05,
+      "loss": 1.791,
+      "step": 65010
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.523502219781226e-05,
+      "loss": 1.827,
+      "step": 65020
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.523273376355897e-05,
+      "loss": 1.7803,
+      "step": 65030
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.523044532930569e-05,
+      "loss": 1.8605,
+      "step": 65040
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5228156895052406e-05,
+      "loss": 1.8823,
+      "step": 65050
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.522586846079912e-05,
+      "loss": 1.8415,
+      "step": 65060
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.522358002654584e-05,
+      "loss": 2.046,
+      "step": 65070
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5221291592292555e-05,
+      "loss": 1.7473,
+      "step": 65080
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.521900315803927e-05,
+      "loss": 1.8824,
+      "step": 65090
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.521671472378599e-05,
+      "loss": 1.8442,
+      "step": 65100
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5214426289532704e-05,
+      "loss": 1.9977,
+      "step": 65110
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.521213785527942e-05,
+      "loss": 1.9237,
+      "step": 65120
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.520984942102614e-05,
+      "loss": 1.7669,
+      "step": 65130
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5207560986772854e-05,
+      "loss": 1.9044,
+      "step": 65140
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.520527255251957e-05,
+      "loss": 2.0043,
+      "step": 65150
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.520298411826629e-05,
+      "loss": 1.8799,
+      "step": 65160
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5200695684013e-05,
+      "loss": 1.6773,
+      "step": 65170
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.519840724975972e-05,
+      "loss": 1.7597,
+      "step": 65180
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.519611881550644e-05,
+      "loss": 1.9074,
+      "step": 65190
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5193830381253145e-05,
+      "loss": 2.0024,
+      "step": 65200
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.519154194699986e-05,
+      "loss": 1.8697,
+      "step": 65210
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.518925351274658e-05,
+      "loss": 1.7033,
+      "step": 65220
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5186965078493294e-05,
+      "loss": 1.8035,
+      "step": 65230
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.518467664424001e-05,
+      "loss": 1.8199,
+      "step": 65240
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.518238820998673e-05,
+      "loss": 1.7095,
+      "step": 65250
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5180099775733444e-05,
+      "loss": 1.8901,
+      "step": 65260
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.517781134148016e-05,
+      "loss": 1.8147,
+      "step": 65270
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.517552290722688e-05,
+      "loss": 1.7665,
+      "step": 65280
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.517323447297359e-05,
+      "loss": 1.7666,
+      "step": 65290
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.517094603872031e-05,
+      "loss": 1.7612,
+      "step": 65300
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.516865760446703e-05,
+      "loss": 1.8007,
+      "step": 65310
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.516636917021374e-05,
+      "loss": 1.7832,
+      "step": 65320
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5164080735960456e-05,
+      "loss": 1.9922,
+      "step": 65330
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.516179230170718e-05,
+      "loss": 1.7989,
+      "step": 65340
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.515950386745389e-05,
+      "loss": 1.8394,
+      "step": 65350
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5157215433200605e-05,
+      "loss": 1.8232,
+      "step": 65360
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5154926998947326e-05,
+      "loss": 1.8581,
+      "step": 65370
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.515263856469404e-05,
+      "loss": 1.8339,
+      "step": 65380
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5150350130440754e-05,
+      "loss": 2.0136,
+      "step": 65390
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.514806169618747e-05,
+      "loss": 1.943,
+      "step": 65400
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.514577326193418e-05,
+      "loss": 1.9084,
+      "step": 65410
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5143484827680904e-05,
+      "loss": 1.9565,
+      "step": 65420
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.514119639342762e-05,
+      "loss": 1.8018,
+      "step": 65430
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.513890795917433e-05,
+      "loss": 1.8721,
+      "step": 65440
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.513661952492105e-05,
+      "loss": 1.7478,
+      "step": 65450
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.513433109066777e-05,
+      "loss": 1.9598,
+      "step": 65460
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.513204265641448e-05,
+      "loss": 1.7203,
+      "step": 65470
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.51297542221612e-05,
+      "loss": 1.6623,
+      "step": 65480
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5127465787907916e-05,
+      "loss": 1.8727,
+      "step": 65490
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.512517735365463e-05,
+      "loss": 1.7237,
+      "step": 65500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.512288891940135e-05,
+      "loss": 1.9115,
+      "step": 65510
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5120600485148065e-05,
+      "loss": 1.7929,
+      "step": 65520
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.511831205089478e-05,
+      "loss": 1.8861,
+      "step": 65530
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.51160236166415e-05,
+      "loss": 2.0061,
+      "step": 65540
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5113735182388214e-05,
+      "loss": 1.8188,
+      "step": 65550
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.511144674813493e-05,
+      "loss": 1.8446,
+      "step": 65560
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.510915831388164e-05,
+      "loss": 1.9129,
+      "step": 65570
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5106869879628364e-05,
+      "loss": 1.9336,
+      "step": 65580
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.510458144537508e-05,
+      "loss": 1.8172,
+      "step": 65590
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.510229301112179e-05,
+      "loss": 1.9895,
+      "step": 65600
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5100004576868506e-05,
+      "loss": 1.7691,
+      "step": 65610
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.509771614261522e-05,
+      "loss": 1.9527,
+      "step": 65620
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.509542770836194e-05,
+      "loss": 1.8455,
+      "step": 65630
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5093139274108655e-05,
+      "loss": 1.7767,
+      "step": 65640
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.509085083985537e-05,
+      "loss": 1.8234,
+      "step": 65650
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.508856240560209e-05,
+      "loss": 1.9327,
+      "step": 65660
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5086273971348804e-05,
+      "loss": 1.9408,
+      "step": 65670
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.508398553709552e-05,
+      "loss": 1.9963,
+      "step": 65680
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.508169710284224e-05,
+      "loss": 1.8424,
+      "step": 65690
+    },
+    {
+      "epoch": 3.0,
+      "eval_cer": 0.6810235444197945,
+      "eval_em": 0.00728476821192053,
+      "eval_f1": 0.00728476821192053,
+      "eval_loss": 1.7074658870697021,
+      "eval_runtime": 2583.2839,
+      "eval_samples_per_second": 4.092,
+      "eval_steps_per_second": 2.046,
+      "eval_wer": 0.9927152317880795,
+      "step": 65699
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5079408668588953e-05,
+      "loss": 2.0082,
+      "step": 65700
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.507712023433567e-05,
+      "loss": 1.9236,
+      "step": 65710
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.507483180008239e-05,
+      "loss": 1.7465,
+      "step": 65720
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.50725433658291e-05,
+      "loss": 1.9264,
+      "step": 65730
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.507025493157582e-05,
+      "loss": 1.779,
+      "step": 65740
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.506796649732254e-05,
+      "loss": 1.7644,
+      "step": 65750
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.506567806306925e-05,
+      "loss": 1.7723,
+      "step": 65760
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5063389628815966e-05,
+      "loss": 1.8406,
+      "step": 65770
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.506110119456269e-05,
+      "loss": 1.7262,
+      "step": 65780
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.50588127603094e-05,
+      "loss": 2.0243,
+      "step": 65790
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.505652432605611e-05,
+      "loss": 1.8153,
+      "step": 65800
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.505423589180283e-05,
+      "loss": 1.7175,
+      "step": 65810
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5051947457549543e-05,
+      "loss": 1.8107,
+      "step": 65820
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.504965902329626e-05,
+      "loss": 1.8826,
+      "step": 65830
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.504737058904298e-05,
+      "loss": 1.8448,
+      "step": 65840
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.504508215478969e-05,
+      "loss": 1.7892,
+      "step": 65850
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.504279372053641e-05,
+      "loss": 1.9239,
+      "step": 65860
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.504050528628313e-05,
+      "loss": 1.8223,
+      "step": 65870
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.503821685202984e-05,
+      "loss": 1.8341,
+      "step": 65880
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5035928417776556e-05,
+      "loss": 1.8385,
+      "step": 65890
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.503363998352328e-05,
+      "loss": 1.8875,
+      "step": 65900
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.503135154926999e-05,
+      "loss": 1.7562,
+      "step": 65910
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5029063115016705e-05,
+      "loss": 1.7359,
+      "step": 65920
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5026774680763426e-05,
+      "loss": 1.6034,
+      "step": 65930
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.502448624651014e-05,
+      "loss": 1.6806,
+      "step": 65940
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5022197812256854e-05,
+      "loss": 1.9916,
+      "step": 65950
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5019909378003575e-05,
+      "loss": 1.6789,
+      "step": 65960
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.501762094375029e-05,
+      "loss": 1.8911,
+      "step": 65970
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5015332509497003e-05,
+      "loss": 1.8196,
+      "step": 65980
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.501304407524372e-05,
+      "loss": 1.7601,
+      "step": 65990
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.501075564099043e-05,
+      "loss": 1.8497,
+      "step": 66000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.500846720673715e-05,
+      "loss": 1.7737,
+      "step": 66010
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.500617877248387e-05,
+      "loss": 1.7045,
+      "step": 66020
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.500389033823058e-05,
+      "loss": 1.8618,
+      "step": 66030
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.50016019039773e-05,
+      "loss": 1.9597,
+      "step": 66040
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4999313469724016e-05,
+      "loss": 1.8738,
+      "step": 66050
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.499702503547073e-05,
+      "loss": 1.6934,
+      "step": 66060
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.499473660121745e-05,
+      "loss": 1.924,
+      "step": 66070
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4992448166964165e-05,
+      "loss": 1.937,
+      "step": 66080
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.499015973271088e-05,
+      "loss": 1.7885,
+      "step": 66090
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.49878712984576e-05,
+      "loss": 1.7927,
+      "step": 66100
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4985582864204314e-05,
+      "loss": 1.7998,
+      "step": 66110
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.498329442995103e-05,
+      "loss": 1.8242,
+      "step": 66120
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.498100599569775e-05,
+      "loss": 1.6451,
+      "step": 66130
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4978717561444463e-05,
+      "loss": 1.8553,
+      "step": 66140
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.497642912719118e-05,
+      "loss": 1.585,
+      "step": 66150
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.49741406929379e-05,
+      "loss": 1.8591,
+      "step": 66160
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.497185225868461e-05,
+      "loss": 1.9516,
+      "step": 66170
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.496956382443133e-05,
+      "loss": 1.7638,
+      "step": 66180
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.496727539017804e-05,
+      "loss": 1.9891,
+      "step": 66190
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4964986955924755e-05,
+      "loss": 1.9083,
+      "step": 66200
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.496269852167147e-05,
+      "loss": 1.8725,
+      "step": 66210
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.496041008741819e-05,
+      "loss": 1.7639,
+      "step": 66220
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4958121653164904e-05,
+      "loss": 1.8529,
+      "step": 66230
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.495583321891162e-05,
+      "loss": 1.972,
+      "step": 66240
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.495354478465834e-05,
+      "loss": 1.8658,
+      "step": 66250
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.495125635040505e-05,
+      "loss": 1.8149,
+      "step": 66260
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.494896791615177e-05,
+      "loss": 1.9893,
+      "step": 66270
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.494667948189849e-05,
+      "loss": 1.9856,
+      "step": 66280
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.49443910476452e-05,
+      "loss": 1.8833,
+      "step": 66290
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.494210261339192e-05,
+      "loss": 1.8682,
+      "step": 66300
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.493981417913864e-05,
+      "loss": 1.7846,
+      "step": 66310
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.493752574488535e-05,
+      "loss": 1.8052,
+      "step": 66320
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4935237310632066e-05,
+      "loss": 1.9652,
+      "step": 66330
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.493294887637879e-05,
+      "loss": 2.1232,
+      "step": 66340
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.49306604421255e-05,
+      "loss": 1.735,
+      "step": 66350
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4928372007872215e-05,
+      "loss": 1.86,
+      "step": 66360
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4926083573618936e-05,
+      "loss": 1.5874,
+      "step": 66370
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.492379513936565e-05,
+      "loss": 1.8253,
+      "step": 66380
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4921506705112364e-05,
+      "loss": 1.7985,
+      "step": 66390
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.491921827085908e-05,
+      "loss": 1.7732,
+      "step": 66400
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.491692983660579e-05,
+      "loss": 1.842,
+      "step": 66410
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4914641402352513e-05,
+      "loss": 1.8042,
+      "step": 66420
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.491235296809923e-05,
+      "loss": 1.8681,
+      "step": 66430
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.491006453384594e-05,
+      "loss": 1.729,
+      "step": 66440
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.490777609959266e-05,
+      "loss": 1.781,
+      "step": 66450
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.490548766533938e-05,
+      "loss": 1.9866,
+      "step": 66460
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.490319923108609e-05,
+      "loss": 1.7896,
+      "step": 66470
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.490091079683281e-05,
+      "loss": 1.8367,
+      "step": 66480
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4898622362579526e-05,
+      "loss": 1.8227,
+      "step": 66490
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.489633392832624e-05,
+      "loss": 1.7682,
+      "step": 66500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.489404549407296e-05,
+      "loss": 1.8766,
+      "step": 66510
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4891757059819675e-05,
+      "loss": 1.9236,
+      "step": 66520
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.488946862556639e-05,
+      "loss": 1.6977,
+      "step": 66530
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.488718019131311e-05,
+      "loss": 1.9414,
+      "step": 66540
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4884891757059824e-05,
+      "loss": 1.7656,
+      "step": 66550
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.488260332280654e-05,
+      "loss": 1.9116,
+      "step": 66560
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.488031488855326e-05,
+      "loss": 1.7284,
+      "step": 66570
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4878026454299973e-05,
+      "loss": 1.8351,
+      "step": 66580
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.487573802004668e-05,
+      "loss": 1.8359,
+      "step": 66590
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.48734495857934e-05,
+      "loss": 1.8834,
+      "step": 66600
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4871161151540116e-05,
+      "loss": 1.9112,
+      "step": 66610
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.486887271728683e-05,
+      "loss": 1.8849,
+      "step": 66620
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.486658428303355e-05,
+      "loss": 1.8758,
+      "step": 66630
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4864295848780265e-05,
+      "loss": 1.7935,
+      "step": 66640
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.486200741452698e-05,
+      "loss": 1.9099,
+      "step": 66650
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.48597189802737e-05,
+      "loss": 1.7949,
+      "step": 66660
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4857430546020414e-05,
+      "loss": 1.6891,
+      "step": 66670
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.485514211176713e-05,
+      "loss": 1.8326,
+      "step": 66680
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.485285367751385e-05,
+      "loss": 1.8512,
+      "step": 66690
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.485056524326056e-05,
+      "loss": 1.6338,
+      "step": 66700
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.484827680900728e-05,
+      "loss": 1.7524,
+      "step": 66710
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4845988374754e-05,
+      "loss": 1.6919,
+      "step": 66720
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.484369994050071e-05,
+      "loss": 1.8179,
+      "step": 66730
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.484141150624743e-05,
+      "loss": 1.7312,
+      "step": 66740
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.483912307199415e-05,
+      "loss": 1.6406,
+      "step": 66750
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.483683463774086e-05,
+      "loss": 1.7079,
+      "step": 66760
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4834546203487576e-05,
+      "loss": 1.8985,
+      "step": 66770
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.483225776923429e-05,
+      "loss": 1.8188,
+      "step": 66780
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4829969334981004e-05,
+      "loss": 1.8982,
+      "step": 66790
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.482768090072772e-05,
+      "loss": 1.9518,
+      "step": 66800
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.482539246647444e-05,
+      "loss": 1.7618,
+      "step": 66810
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.482310403222115e-05,
+      "loss": 2.0259,
+      "step": 66820
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.482081559796787e-05,
+      "loss": 1.9405,
+      "step": 66830
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.481852716371459e-05,
+      "loss": 1.8822,
+      "step": 66840
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.48162387294613e-05,
+      "loss": 1.9203,
+      "step": 66850
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4813950295208017e-05,
+      "loss": 1.8765,
+      "step": 66860
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.481166186095474e-05,
+      "loss": 1.5456,
+      "step": 66870
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.480937342670145e-05,
+      "loss": 1.8903,
+      "step": 66880
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4807084992448166e-05,
+      "loss": 1.7233,
+      "step": 66890
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.480479655819489e-05,
+      "loss": 1.9166,
+      "step": 66900
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.48025081239416e-05,
+      "loss": 1.8762,
+      "step": 66910
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4800219689688315e-05,
+      "loss": 1.7883,
+      "step": 66920
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4797931255435036e-05,
+      "loss": 1.9336,
+      "step": 66930
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.479564282118175e-05,
+      "loss": 1.8858,
+      "step": 66940
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4793354386928464e-05,
+      "loss": 1.7053,
+      "step": 66950
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4791065952675185e-05,
+      "loss": 1.8817,
+      "step": 66960
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.47887775184219e-05,
+      "loss": 1.8366,
+      "step": 66970
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.478648908416861e-05,
+      "loss": 1.5362,
+      "step": 66980
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.478420064991533e-05,
+      "loss": 1.785,
+      "step": 66990
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.478191221566204e-05,
+      "loss": 1.8607,
+      "step": 67000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.477962378140876e-05,
+      "loss": 1.8532,
+      "step": 67010
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4777335347155477e-05,
+      "loss": 1.6712,
+      "step": 67020
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.477504691290219e-05,
+      "loss": 1.9312,
+      "step": 67030
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.477275847864891e-05,
+      "loss": 1.8828,
+      "step": 67040
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4770470044395626e-05,
+      "loss": 1.7811,
+      "step": 67050
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.476818161014234e-05,
+      "loss": 1.8132,
+      "step": 67060
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.476589317588906e-05,
+      "loss": 1.6969,
+      "step": 67070
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4763604741635775e-05,
+      "loss": 1.8673,
+      "step": 67080
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.476131630738249e-05,
+      "loss": 1.912,
+      "step": 67090
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.475902787312921e-05,
+      "loss": 1.8126,
+      "step": 67100
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4756739438875924e-05,
+      "loss": 1.8136,
+      "step": 67110
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.475445100462264e-05,
+      "loss": 1.7597,
+      "step": 67120
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.475216257036936e-05,
+      "loss": 1.8651,
+      "step": 67130
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.474987413611607e-05,
+      "loss": 1.7527,
+      "step": 67140
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.474758570186279e-05,
+      "loss": 1.7671,
+      "step": 67150
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.474529726760951e-05,
+      "loss": 1.8312,
+      "step": 67160
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.474300883335622e-05,
+      "loss": 1.7418,
+      "step": 67170
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.474072039910293e-05,
+      "loss": 1.8404,
+      "step": 67180
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.473843196484965e-05,
+      "loss": 1.8354,
+      "step": 67190
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4736143530596365e-05,
+      "loss": 1.7296,
+      "step": 67200
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.473385509634308e-05,
+      "loss": 1.7168,
+      "step": 67210
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.47315666620898e-05,
+      "loss": 1.8603,
+      "step": 67220
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4729278227836514e-05,
+      "loss": 1.6753,
+      "step": 67230
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.472698979358323e-05,
+      "loss": 1.7897,
+      "step": 67240
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.472470135932995e-05,
+      "loss": 1.8106,
+      "step": 67250
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.472241292507666e-05,
+      "loss": 1.9712,
+      "step": 67260
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.472012449082338e-05,
+      "loss": 1.885,
+      "step": 67270
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.47178360565701e-05,
+      "loss": 1.9207,
+      "step": 67280
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.471554762231681e-05,
+      "loss": 1.8367,
+      "step": 67290
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4713259188063527e-05,
+      "loss": 1.8879,
+      "step": 67300
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.471097075381025e-05,
+      "loss": 1.7377,
+      "step": 67310
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.470868231955696e-05,
+      "loss": 1.8643,
+      "step": 67320
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4706393885303676e-05,
+      "loss": 1.6409,
+      "step": 67330
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.47041054510504e-05,
+      "loss": 1.8309,
+      "step": 67340
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.470181701679711e-05,
+      "loss": 1.8875,
+      "step": 67350
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4699528582543825e-05,
+      "loss": 1.6818,
+      "step": 67360
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4697240148290546e-05,
+      "loss": 1.8811,
+      "step": 67370
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.469495171403725e-05,
+      "loss": 1.7838,
+      "step": 67380
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4692663279783974e-05,
+      "loss": 1.7223,
+      "step": 67390
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.469037484553069e-05,
+      "loss": 1.7655,
+      "step": 67400
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.46880864112774e-05,
+      "loss": 1.746,
+      "step": 67410
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.468579797702412e-05,
+      "loss": 1.7604,
+      "step": 67420
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.468350954277084e-05,
+      "loss": 1.8263,
+      "step": 67430
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.468122110851755e-05,
+      "loss": 1.8382,
+      "step": 67440
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.467893267426427e-05,
+      "loss": 1.7845,
+      "step": 67450
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4676644240010987e-05,
+      "loss": 1.7753,
+      "step": 67460
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.46743558057577e-05,
+      "loss": 1.7773,
+      "step": 67470
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.467206737150442e-05,
+      "loss": 1.9175,
+      "step": 67480
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4669778937251136e-05,
+      "loss": 1.8634,
+      "step": 67490
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.466749050299785e-05,
+      "loss": 1.8878,
+      "step": 67500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.466520206874457e-05,
+      "loss": 1.7003,
+      "step": 67510
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4662913634491285e-05,
+      "loss": 1.8268,
+      "step": 67520
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4660625200238e-05,
+      "loss": 1.8576,
+      "step": 67530
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.465833676598472e-05,
+      "loss": 1.8893,
+      "step": 67540
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4656048331731434e-05,
+      "loss": 1.8668,
+      "step": 67550
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.465375989747815e-05,
+      "loss": 1.8605,
+      "step": 67560
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.465147146322486e-05,
+      "loss": 1.821,
+      "step": 67570
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4649183028971577e-05,
+      "loss": 1.6062,
+      "step": 67580
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.464689459471829e-05,
+      "loss": 1.8769,
+      "step": 67590
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.464460616046501e-05,
+      "loss": 1.945,
+      "step": 67600
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4642317726211726e-05,
+      "loss": 1.82,
+      "step": 67610
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.464002929195844e-05,
+      "loss": 1.7248,
+      "step": 67620
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.463774085770516e-05,
+      "loss": 1.8834,
+      "step": 67630
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4635452423451875e-05,
+      "loss": 1.7171,
+      "step": 67640
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.463316398919859e-05,
+      "loss": 1.776,
+      "step": 67650
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.463087555494531e-05,
+      "loss": 1.7368,
+      "step": 67660
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4628587120692024e-05,
+      "loss": 1.6574,
+      "step": 67670
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.462629868643874e-05,
+      "loss": 1.6888,
+      "step": 67680
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.462401025218546e-05,
+      "loss": 1.6128,
+      "step": 67690
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.462172181793217e-05,
+      "loss": 1.7671,
+      "step": 67700
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.461943338367889e-05,
+      "loss": 1.7726,
+      "step": 67710
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.461714494942561e-05,
+      "loss": 1.6521,
+      "step": 67720
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.461485651517232e-05,
+      "loss": 1.6447,
+      "step": 67730
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4612568080919037e-05,
+      "loss": 1.8365,
+      "step": 67740
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.461027964666576e-05,
+      "loss": 1.7209,
+      "step": 67750
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.460799121241247e-05,
+      "loss": 1.8324,
+      "step": 67760
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4605702778159186e-05,
+      "loss": 1.6994,
+      "step": 67770
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.46034143439059e-05,
+      "loss": 1.7332,
+      "step": 67780
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4601125909652614e-05,
+      "loss": 1.9242,
+      "step": 67790
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4598837475399335e-05,
+      "loss": 1.843,
+      "step": 67800
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.459654904114605e-05,
+      "loss": 1.5988,
+      "step": 67810
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.459426060689276e-05,
+      "loss": 1.8774,
+      "step": 67820
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4591972172639484e-05,
+      "loss": 1.8034,
+      "step": 67830
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.45896837383862e-05,
+      "loss": 1.8662,
+      "step": 67840
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.458739530413291e-05,
+      "loss": 1.7466,
+      "step": 67850
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.458510686987963e-05,
+      "loss": 1.8537,
+      "step": 67860
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.458281843562635e-05,
+      "loss": 1.8736,
+      "step": 67870
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.458053000137306e-05,
+      "loss": 1.7332,
+      "step": 67880
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.457824156711978e-05,
+      "loss": 1.618,
+      "step": 67890
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4575953132866497e-05,
+      "loss": 1.7266,
+      "step": 67900
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.457366469861321e-05,
+      "loss": 1.9774,
+      "step": 67910
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4571376264359925e-05,
+      "loss": 1.8151,
+      "step": 67920
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4569087830106646e-05,
+      "loss": 1.9302,
+      "step": 67930
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.456679939585336e-05,
+      "loss": 1.6073,
+      "step": 67940
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4564510961600074e-05,
+      "loss": 1.7988,
+      "step": 67950
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4562222527346795e-05,
+      "loss": 1.7984,
+      "step": 67960
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.45599340930935e-05,
+      "loss": 1.7269,
+      "step": 67970
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.455764565884022e-05,
+      "loss": 1.7001,
+      "step": 67980
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.455535722458694e-05,
+      "loss": 1.6695,
+      "step": 67990
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.455306879033365e-05,
+      "loss": 1.8122,
+      "step": 68000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.455078035608037e-05,
+      "loss": 1.714,
+      "step": 68010
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4548491921827086e-05,
+      "loss": 1.6645,
+      "step": 68020
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.45462034875738e-05,
+      "loss": 1.78,
+      "step": 68030
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.454391505332052e-05,
+      "loss": 1.9496,
+      "step": 68040
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4541626619067236e-05,
+      "loss": 2.0216,
+      "step": 68050
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.453933818481395e-05,
+      "loss": 1.84,
+      "step": 68060
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.453704975056067e-05,
+      "loss": 1.5849,
+      "step": 68070
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4534761316307385e-05,
+      "loss": 1.8691,
+      "step": 68080
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.45324728820541e-05,
+      "loss": 1.8493,
+      "step": 68090
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.453018444780082e-05,
+      "loss": 1.8456,
+      "step": 68100
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4527896013547534e-05,
+      "loss": 1.6818,
+      "step": 68110
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.452560757929425e-05,
+      "loss": 1.8739,
+      "step": 68120
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.452331914504097e-05,
+      "loss": 1.7854,
+      "step": 68130
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.452103071078768e-05,
+      "loss": 1.7761,
+      "step": 68140
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.45187422765344e-05,
+      "loss": 1.8528,
+      "step": 68150
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.451645384228112e-05,
+      "loss": 1.7587,
+      "step": 68160
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4514165408027826e-05,
+      "loss": 1.7646,
+      "step": 68170
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.451187697377454e-05,
+      "loss": 1.973,
+      "step": 68180
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.450958853952126e-05,
+      "loss": 1.851,
+      "step": 68190
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4507300105267975e-05,
+      "loss": 1.903,
+      "step": 68200
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.450501167101469e-05,
+      "loss": 1.7386,
+      "step": 68210
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.450272323676141e-05,
+      "loss": 1.7277,
+      "step": 68220
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4500434802508124e-05,
+      "loss": 1.7496,
+      "step": 68230
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.449814636825484e-05,
+      "loss": 1.7091,
+      "step": 68240
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.449585793400156e-05,
+      "loss": 1.8363,
+      "step": 68250
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.449356949974827e-05,
+      "loss": 1.7107,
+      "step": 68260
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.449128106549499e-05,
+      "loss": 1.8497,
+      "step": 68270
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.448899263124171e-05,
+      "loss": 1.8232,
+      "step": 68280
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.448670419698842e-05,
+      "loss": 1.7486,
+      "step": 68290
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4484415762735136e-05,
+      "loss": 1.7461,
+      "step": 68300
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.448212732848186e-05,
+      "loss": 1.845,
+      "step": 68310
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.447983889422857e-05,
+      "loss": 1.9282,
+      "step": 68320
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4477550459975286e-05,
+      "loss": 1.7978,
+      "step": 68330
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4475262025722007e-05,
+      "loss": 1.7619,
+      "step": 68340
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.447297359146872e-05,
+      "loss": 1.7569,
+      "step": 68350
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4470685157215435e-05,
+      "loss": 1.7746,
+      "step": 68360
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.446839672296215e-05,
+      "loss": 1.8847,
+      "step": 68370
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.446610828870886e-05,
+      "loss": 1.7351,
+      "step": 68380
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4463819854455584e-05,
+      "loss": 1.5907,
+      "step": 68390
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.44615314202023e-05,
+      "loss": 1.638,
+      "step": 68400
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.445924298594901e-05,
+      "loss": 1.6442,
+      "step": 68410
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.445695455169573e-05,
+      "loss": 1.7777,
+      "step": 68420
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.445466611744245e-05,
+      "loss": 1.9267,
+      "step": 68430
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.445237768318916e-05,
+      "loss": 1.7699,
+      "step": 68440
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.445008924893588e-05,
+      "loss": 1.8349,
+      "step": 68450
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4447800814682596e-05,
+      "loss": 1.7343,
+      "step": 68460
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.444551238042931e-05,
+      "loss": 1.9381,
+      "step": 68470
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.444322394617603e-05,
+      "loss": 1.7458,
+      "step": 68480
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4440935511922746e-05,
+      "loss": 1.713,
+      "step": 68490
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.443864707766946e-05,
+      "loss": 1.8782,
+      "step": 68500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.443635864341618e-05,
+      "loss": 1.8424,
+      "step": 68510
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4434070209162895e-05,
+      "loss": 1.7294,
+      "step": 68520
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.443178177490961e-05,
+      "loss": 1.7328,
+      "step": 68530
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.442949334065633e-05,
+      "loss": 1.7296,
+      "step": 68540
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4427204906403044e-05,
+      "loss": 1.7063,
+      "step": 68550
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.442491647214975e-05,
+      "loss": 1.9927,
+      "step": 68560
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.442262803789647e-05,
+      "loss": 1.9218,
+      "step": 68570
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4420339603643186e-05,
+      "loss": 1.7455,
+      "step": 68580
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.44180511693899e-05,
+      "loss": 1.5765,
+      "step": 68590
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.441576273513662e-05,
+      "loss": 1.6608,
+      "step": 68600
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4413474300883336e-05,
+      "loss": 1.5993,
+      "step": 68610
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.441118586663005e-05,
+      "loss": 1.8218,
+      "step": 68620
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.440889743237677e-05,
+      "loss": 1.7492,
+      "step": 68630
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4406608998123485e-05,
+      "loss": 1.7772,
+      "step": 68640
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.44043205638702e-05,
+      "loss": 1.9091,
+      "step": 68650
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.440203212961692e-05,
+      "loss": 1.738,
+      "step": 68660
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4399743695363634e-05,
+      "loss": 1.7664,
+      "step": 68670
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.439745526111035e-05,
+      "loss": 1.805,
+      "step": 68680
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.439516682685707e-05,
+      "loss": 1.4438,
+      "step": 68690
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.439287839260378e-05,
+      "loss": 1.8538,
+      "step": 68700
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.43905899583505e-05,
+      "loss": 1.7678,
+      "step": 68710
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.438830152409722e-05,
+      "loss": 1.7042,
+      "step": 68720
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.438601308984393e-05,
+      "loss": 1.8081,
+      "step": 68730
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4383724655590646e-05,
+      "loss": 1.75,
+      "step": 68740
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.438143622133737e-05,
+      "loss": 1.823,
+      "step": 68750
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4379147787084075e-05,
+      "loss": 1.6547,
+      "step": 68760
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4376859352830796e-05,
+      "loss": 1.8295,
+      "step": 68770
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.437457091857751e-05,
+      "loss": 1.8547,
+      "step": 68780
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4372282484324224e-05,
+      "loss": 1.7879,
+      "step": 68790
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4369994050070945e-05,
+      "loss": 1.654,
+      "step": 68800
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.436770561581766e-05,
+      "loss": 1.7949,
+      "step": 68810
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.436541718156437e-05,
+      "loss": 1.6431,
+      "step": 68820
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4363128747311094e-05,
+      "loss": 1.7206,
+      "step": 68830
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.436084031305781e-05,
+      "loss": 1.6454,
+      "step": 68840
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.435855187880452e-05,
+      "loss": 1.7674,
+      "step": 68850
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.435626344455124e-05,
+      "loss": 1.8448,
+      "step": 68860
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.435397501029796e-05,
+      "loss": 1.6755,
+      "step": 68870
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.435168657604467e-05,
+      "loss": 1.5661,
+      "step": 68880
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.434939814179139e-05,
+      "loss": 1.6262,
+      "step": 68890
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4347109707538106e-05,
+      "loss": 1.6667,
+      "step": 68900
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.434482127328482e-05,
+      "loss": 1.8109,
+      "step": 68910
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.434253283903154e-05,
+      "loss": 1.6388,
+      "step": 68920
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4340244404778256e-05,
+      "loss": 1.7376,
+      "step": 68930
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.433795597052497e-05,
+      "loss": 1.8557,
+      "step": 68940
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.433566753627169e-05,
+      "loss": 1.7879,
+      "step": 68950
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.43333791020184e-05,
+      "loss": 1.6956,
+      "step": 68960
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.433109066776511e-05,
+      "loss": 1.7786,
+      "step": 68970
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.432880223351183e-05,
+      "loss": 1.7331,
+      "step": 68980
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.432651379925855e-05,
+      "loss": 2.0673,
+      "step": 68990
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.432422536500526e-05,
+      "loss": 1.7565,
+      "step": 69000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.432193693075198e-05,
+      "loss": 1.7224,
+      "step": 69010
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4319648496498696e-05,
+      "loss": 1.6618,
+      "step": 69020
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.431736006224541e-05,
+      "loss": 1.8148,
+      "step": 69030
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.431507162799213e-05,
+      "loss": 1.8164,
+      "step": 69040
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4312783193738846e-05,
+      "loss": 1.7092,
+      "step": 69050
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.431049475948556e-05,
+      "loss": 1.9989,
+      "step": 69060
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.430820632523228e-05,
+      "loss": 1.8269,
+      "step": 69070
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4305917890978995e-05,
+      "loss": 1.7818,
+      "step": 69080
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.430362945672571e-05,
+      "loss": 1.7345,
+      "step": 69090
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.430134102247243e-05,
+      "loss": 1.7732,
+      "step": 69100
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4299052588219144e-05,
+      "loss": 1.9845,
+      "step": 69110
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.429676415396586e-05,
+      "loss": 1.8979,
+      "step": 69120
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.429447571971258e-05,
+      "loss": 1.5787,
+      "step": 69130
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.429218728545929e-05,
+      "loss": 1.7492,
+      "step": 69140
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4289898851206e-05,
+      "loss": 1.7293,
+      "step": 69150
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.428761041695272e-05,
+      "loss": 1.8577,
+      "step": 69160
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4285321982699435e-05,
+      "loss": 1.7158,
+      "step": 69170
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.428303354844615e-05,
+      "loss": 1.6124,
+      "step": 69180
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.428074511419287e-05,
+      "loss": 1.6171,
+      "step": 69190
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4278456679939585e-05,
+      "loss": 1.7737,
+      "step": 69200
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.42761682456863e-05,
+      "loss": 1.8593,
+      "step": 69210
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.427387981143302e-05,
+      "loss": 1.7622,
+      "step": 69220
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4271591377179734e-05,
+      "loss": 1.7793,
+      "step": 69230
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.426930294292645e-05,
+      "loss": 1.7763,
+      "step": 69240
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.426701450867317e-05,
+      "loss": 1.6456,
+      "step": 69250
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.426472607441988e-05,
+      "loss": 1.8166,
+      "step": 69260
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.42624376401666e-05,
+      "loss": 1.7364,
+      "step": 69270
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.426014920591332e-05,
+      "loss": 1.6167,
+      "step": 69280
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.425786077166003e-05,
+      "loss": 1.6206,
+      "step": 69290
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4255572337406746e-05,
+      "loss": 1.9497,
+      "step": 69300
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.425328390315347e-05,
+      "loss": 1.6494,
+      "step": 69310
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.425099546890018e-05,
+      "loss": 2.0014,
+      "step": 69320
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4248707034646895e-05,
+      "loss": 1.7228,
+      "step": 69330
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4246418600393616e-05,
+      "loss": 1.6583,
+      "step": 69340
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4244130166140324e-05,
+      "loss": 1.6359,
+      "step": 69350
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4241841731887045e-05,
+      "loss": 1.8471,
+      "step": 69360
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.423955329763376e-05,
+      "loss": 1.803,
+      "step": 69370
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.423726486338047e-05,
+      "loss": 1.697,
+      "step": 69380
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4234976429127194e-05,
+      "loss": 1.9739,
+      "step": 69390
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.423268799487391e-05,
+      "loss": 1.6234,
+      "step": 69400
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.423039956062062e-05,
+      "loss": 1.7986,
+      "step": 69410
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.422811112636734e-05,
+      "loss": 1.7443,
+      "step": 69420
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.422582269211406e-05,
+      "loss": 1.8075,
+      "step": 69430
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.422353425786077e-05,
+      "loss": 1.7675,
+      "step": 69440
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.422124582360749e-05,
+      "loss": 1.699,
+      "step": 69450
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4218957389354206e-05,
+      "loss": 1.6489,
+      "step": 69460
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.421666895510092e-05,
+      "loss": 1.7365,
+      "step": 69470
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.421438052084764e-05,
+      "loss": 1.8384,
+      "step": 69480
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4212092086594355e-05,
+      "loss": 1.7131,
+      "step": 69490
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.420980365234107e-05,
+      "loss": 1.6463,
+      "step": 69500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.420751521808779e-05,
+      "loss": 1.8019,
+      "step": 69510
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4205226783834505e-05,
+      "loss": 1.7793,
+      "step": 69520
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.420293834958122e-05,
+      "loss": 1.6761,
+      "step": 69530
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.420064991532794e-05,
+      "loss": 1.7003,
+      "step": 69540
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.419836148107465e-05,
+      "loss": 1.7302,
+      "step": 69550
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.419607304682136e-05,
+      "loss": 1.7575,
+      "step": 69560
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.419378461256808e-05,
+      "loss": 1.6915,
+      "step": 69570
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4191496178314796e-05,
+      "loss": 1.7202,
+      "step": 69580
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.418920774406151e-05,
+      "loss": 1.8012,
+      "step": 69590
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.418691930980823e-05,
+      "loss": 1.6556,
+      "step": 69600
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4184630875554945e-05,
+      "loss": 1.7071,
+      "step": 69610
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.418234244130166e-05,
+      "loss": 1.637,
+      "step": 69620
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.418005400704838e-05,
+      "loss": 1.5121,
+      "step": 69630
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4177765572795095e-05,
+      "loss": 1.6649,
+      "step": 69640
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.417547713854181e-05,
+      "loss": 1.7736,
+      "step": 69650
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.417318870428853e-05,
+      "loss": 1.5797,
+      "step": 69660
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4170900270035244e-05,
+      "loss": 1.9296,
+      "step": 69670
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.416861183578196e-05,
+      "loss": 1.839,
+      "step": 69680
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.416632340152868e-05,
+      "loss": 1.655,
+      "step": 69690
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.416403496727539e-05,
+      "loss": 1.8006,
+      "step": 69700
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.416174653302211e-05,
+      "loss": 1.6733,
+      "step": 69710
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.415945809876883e-05,
+      "loss": 1.685,
+      "step": 69720
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.415716966451554e-05,
+      "loss": 1.8863,
+      "step": 69730
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4154881230262256e-05,
+      "loss": 1.6108,
+      "step": 69740
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.415259279600897e-05,
+      "loss": 1.5868,
+      "step": 69750
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4150304361755685e-05,
+      "loss": 1.4357,
+      "step": 69760
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4148015927502405e-05,
+      "loss": 1.7466,
+      "step": 69770
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.414572749324912e-05,
+      "loss": 1.5788,
+      "step": 69780
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4143439058995834e-05,
+      "loss": 1.7572,
+      "step": 69790
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4141150624742555e-05,
+      "loss": 1.8272,
+      "step": 69800
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.413886219048927e-05,
+      "loss": 1.788,
+      "step": 69810
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.413657375623598e-05,
+      "loss": 1.9066,
+      "step": 69820
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4134285321982704e-05,
+      "loss": 1.7627,
+      "step": 69830
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.413199688772942e-05,
+      "loss": 1.635,
+      "step": 69840
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.412970845347613e-05,
+      "loss": 1.8862,
+      "step": 69850
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.412742001922285e-05,
+      "loss": 1.9245,
+      "step": 69860
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.412513158496957e-05,
+      "loss": 1.6997,
+      "step": 69870
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.412284315071628e-05,
+      "loss": 1.8563,
+      "step": 69880
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4120554716463e-05,
+      "loss": 1.675,
+      "step": 69890
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4118266282209716e-05,
+      "loss": 1.7596,
+      "step": 69900
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.411597784795643e-05,
+      "loss": 1.6975,
+      "step": 69910
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.411368941370315e-05,
+      "loss": 1.7418,
+      "step": 69920
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4111400979449865e-05,
+      "loss": 1.9548,
+      "step": 69930
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.410911254519657e-05,
+      "loss": 1.6185,
+      "step": 69940
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4106824110943294e-05,
+      "loss": 1.7948,
+      "step": 69950
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.410453567669001e-05,
+      "loss": 1.6046,
+      "step": 69960
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.410224724243672e-05,
+      "loss": 1.7939,
+      "step": 69970
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.409995880818344e-05,
+      "loss": 1.6951,
+      "step": 69980
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.409767037393016e-05,
+      "loss": 1.634,
+      "step": 69990
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.409538193967687e-05,
+      "loss": 1.7374,
+      "step": 70000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.409309350542359e-05,
+      "loss": 1.671,
+      "step": 70010
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4090805071170306e-05,
+      "loss": 1.5343,
+      "step": 70020
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.408851663691702e-05,
+      "loss": 1.9255,
+      "step": 70030
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.408622820266374e-05,
+      "loss": 1.7532,
+      "step": 70040
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4083939768410455e-05,
+      "loss": 1.6215,
+      "step": 70050
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.408165133415717e-05,
+      "loss": 1.7772,
+      "step": 70060
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.407936289990389e-05,
+      "loss": 1.8844,
+      "step": 70070
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4077074465650605e-05,
+      "loss": 1.7865,
+      "step": 70080
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.407478603139732e-05,
+      "loss": 1.7616,
+      "step": 70090
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.407249759714404e-05,
+      "loss": 1.7968,
+      "step": 70100
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4070209162890754e-05,
+      "loss": 1.8039,
+      "step": 70110
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.406792072863747e-05,
+      "loss": 1.7293,
+      "step": 70120
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.406563229438419e-05,
+      "loss": 1.9113,
+      "step": 70130
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4063343860130896e-05,
+      "loss": 1.6556,
+      "step": 70140
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.406105542587762e-05,
+      "loss": 1.6795,
+      "step": 70150
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.405876699162433e-05,
+      "loss": 1.5458,
+      "step": 70160
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4056478557371045e-05,
+      "loss": 1.8006,
+      "step": 70170
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4054190123117766e-05,
+      "loss": 1.6203,
+      "step": 70180
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.405190168886448e-05,
+      "loss": 1.6752,
+      "step": 70190
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4049613254611194e-05,
+      "loss": 1.7049,
+      "step": 70200
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4047324820357915e-05,
+      "loss": 1.7893,
+      "step": 70210
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.404503638610463e-05,
+      "loss": 1.6008,
+      "step": 70220
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4042747951851344e-05,
+      "loss": 1.7931,
+      "step": 70230
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.404045951759806e-05,
+      "loss": 1.7247,
+      "step": 70240
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.403817108334478e-05,
+      "loss": 1.6238,
+      "step": 70250
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.403588264909149e-05,
+      "loss": 1.83,
+      "step": 70260
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.403359421483821e-05,
+      "loss": 1.7096,
+      "step": 70270
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.403130578058493e-05,
+      "loss": 1.6918,
+      "step": 70280
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.402901734633164e-05,
+      "loss": 1.7864,
+      "step": 70290
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4026728912078356e-05,
+      "loss": 1.8004,
+      "step": 70300
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.402444047782508e-05,
+      "loss": 1.9507,
+      "step": 70310
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.402215204357179e-05,
+      "loss": 1.6949,
+      "step": 70320
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4019863609318505e-05,
+      "loss": 1.716,
+      "step": 70330
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.401757517506522e-05,
+      "loss": 1.8053,
+      "step": 70340
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4015286740811934e-05,
+      "loss": 1.8545,
+      "step": 70350
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4012998306558655e-05,
+      "loss": 1.9178,
+      "step": 70360
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.401070987230537e-05,
+      "loss": 1.8124,
+      "step": 70370
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.400842143805208e-05,
+      "loss": 1.4884,
+      "step": 70380
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4006133003798804e-05,
+      "loss": 1.7417,
+      "step": 70390
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.400384456954552e-05,
+      "loss": 1.6662,
+      "step": 70400
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.400155613529223e-05,
+      "loss": 1.7288,
+      "step": 70410
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.399926770103895e-05,
+      "loss": 1.743,
+      "step": 70420
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.399697926678567e-05,
+      "loss": 1.9328,
+      "step": 70430
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.399469083253238e-05,
+      "loss": 1.7185,
+      "step": 70440
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.39924023982791e-05,
+      "loss": 1.8222,
+      "step": 70450
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3990113964025816e-05,
+      "loss": 1.6273,
+      "step": 70460
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.398782552977253e-05,
+      "loss": 1.762,
+      "step": 70470
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.398553709551925e-05,
+      "loss": 1.8013,
+      "step": 70480
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3983248661265965e-05,
+      "loss": 1.6543,
+      "step": 70490
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.398096022701268e-05,
+      "loss": 1.7236,
+      "step": 70500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.39786717927594e-05,
+      "loss": 1.8145,
+      "step": 70510
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3976383358506115e-05,
+      "loss": 1.8038,
+      "step": 70520
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.397409492425283e-05,
+      "loss": 1.8,
+      "step": 70530
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.397180648999954e-05,
+      "loss": 1.8272,
+      "step": 70540
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.396951805574626e-05,
+      "loss": 1.6069,
+      "step": 70550
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.396722962149297e-05,
+      "loss": 1.6247,
+      "step": 70560
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.396494118723969e-05,
+      "loss": 1.6849,
+      "step": 70570
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3962652752986406e-05,
+      "loss": 1.5509,
+      "step": 70580
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.396036431873312e-05,
+      "loss": 1.7239,
+      "step": 70590
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.395807588447984e-05,
+      "loss": 1.6991,
+      "step": 70600
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3955787450226555e-05,
+      "loss": 1.8279,
+      "step": 70610
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.395349901597327e-05,
+      "loss": 1.5866,
+      "step": 70620
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.395121058171999e-05,
+      "loss": 1.5912,
+      "step": 70630
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3948922147466704e-05,
+      "loss": 1.8636,
+      "step": 70640
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.394663371321342e-05,
+      "loss": 1.7962,
+      "step": 70650
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.394434527896014e-05,
+      "loss": 1.7646,
+      "step": 70660
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3942056844706854e-05,
+      "loss": 1.7311,
+      "step": 70670
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.393976841045357e-05,
+      "loss": 1.7984,
+      "step": 70680
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.393747997620029e-05,
+      "loss": 1.7737,
+      "step": 70690
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3935191541947e-05,
+      "loss": 1.7682,
+      "step": 70700
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.393290310769372e-05,
+      "loss": 1.6015,
+      "step": 70710
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.393061467344044e-05,
+      "loss": 1.8704,
+      "step": 70720
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3928326239187145e-05,
+      "loss": 1.738,
+      "step": 70730
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3926037804933866e-05,
+      "loss": 1.786,
+      "step": 70740
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.392374937068058e-05,
+      "loss": 1.9052,
+      "step": 70750
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3921460936427294e-05,
+      "loss": 1.878,
+      "step": 70760
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3919172502174015e-05,
+      "loss": 1.7626,
+      "step": 70770
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.391688406792073e-05,
+      "loss": 1.6155,
+      "step": 70780
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3914595633667444e-05,
+      "loss": 1.6156,
+      "step": 70790
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3912307199414164e-05,
+      "loss": 1.6679,
+      "step": 70800
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.391001876516088e-05,
+      "loss": 1.7171,
+      "step": 70810
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.390773033090759e-05,
+      "loss": 1.6262,
+      "step": 70820
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3905441896654314e-05,
+      "loss": 1.8246,
+      "step": 70830
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.390315346240103e-05,
+      "loss": 1.9839,
+      "step": 70840
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.390086502814774e-05,
+      "loss": 1.7657,
+      "step": 70850
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.389857659389446e-05,
+      "loss": 1.7509,
+      "step": 70860
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.389628815964118e-05,
+      "loss": 1.6982,
+      "step": 70870
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.389399972538789e-05,
+      "loss": 1.6829,
+      "step": 70880
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.389171129113461e-05,
+      "loss": 1.6245,
+      "step": 70890
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3889422856881326e-05,
+      "loss": 1.629,
+      "step": 70900
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.388713442262804e-05,
+      "loss": 1.8104,
+      "step": 70910
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.388484598837476e-05,
+      "loss": 1.7058,
+      "step": 70920
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.388255755412147e-05,
+      "loss": 1.861,
+      "step": 70930
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.388026911986818e-05,
+      "loss": 1.6086,
+      "step": 70940
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3877980685614904e-05,
+      "loss": 1.6816,
+      "step": 70950
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.387569225136162e-05,
+      "loss": 1.5911,
+      "step": 70960
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.387340381710833e-05,
+      "loss": 1.8934,
+      "step": 70970
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.387111538285505e-05,
+      "loss": 1.6982,
+      "step": 70980
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.386882694860177e-05,
+      "loss": 1.8171,
+      "step": 70990
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.386653851434848e-05,
+      "loss": 1.6876,
+      "step": 71000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.38642500800952e-05,
+      "loss": 1.8643,
+      "step": 71010
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3861961645841916e-05,
+      "loss": 1.8149,
+      "step": 71020
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.385967321158863e-05,
+      "loss": 1.5721,
+      "step": 71030
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.385738477733535e-05,
+      "loss": 1.632,
+      "step": 71040
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3855096343082065e-05,
+      "loss": 1.9215,
+      "step": 71050
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.385280790882878e-05,
+      "loss": 1.8987,
+      "step": 71060
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.38505194745755e-05,
+      "loss": 1.7088,
+      "step": 71070
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3848231040322214e-05,
+      "loss": 1.8151,
+      "step": 71080
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.384594260606893e-05,
+      "loss": 1.8259,
+      "step": 71090
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.384365417181565e-05,
+      "loss": 1.7353,
+      "step": 71100
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3841365737562364e-05,
+      "loss": 1.6368,
+      "step": 71110
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.383907730330908e-05,
+      "loss": 1.8798,
+      "step": 71120
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.383678886905579e-05,
+      "loss": 1.7317,
+      "step": 71130
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3834500434802506e-05,
+      "loss": 1.6765,
+      "step": 71140
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.383221200054923e-05,
+      "loss": 1.8557,
+      "step": 71150
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.382992356629594e-05,
+      "loss": 1.9942,
+      "step": 71160
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3827635132042655e-05,
+      "loss": 1.8569,
+      "step": 71170
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3825346697789376e-05,
+      "loss": 1.7149,
+      "step": 71180
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.382305826353609e-05,
+      "loss": 1.7091,
+      "step": 71190
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3820769829282804e-05,
+      "loss": 1.7081,
+      "step": 71200
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3818481395029525e-05,
+      "loss": 1.5639,
+      "step": 71210
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.381619296077624e-05,
+      "loss": 2.0248,
+      "step": 71220
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3813904526522954e-05,
+      "loss": 1.6295,
+      "step": 71230
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3811616092269674e-05,
+      "loss": 1.5829,
+      "step": 71240
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.380932765801639e-05,
+      "loss": 1.7021,
+      "step": 71250
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.38070392237631e-05,
+      "loss": 1.7825,
+      "step": 71260
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3804750789509824e-05,
+      "loss": 1.7578,
+      "step": 71270
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.380246235525654e-05,
+      "loss": 1.78,
+      "step": 71280
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.380017392100325e-05,
+      "loss": 1.5419,
+      "step": 71290
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.379788548674997e-05,
+      "loss": 1.6828,
+      "step": 71300
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.379559705249669e-05,
+      "loss": 1.775,
+      "step": 71310
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.37933086182434e-05,
+      "loss": 1.7309,
+      "step": 71320
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3791020183990115e-05,
+      "loss": 1.8254,
+      "step": 71330
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.378873174973683e-05,
+      "loss": 1.9168,
+      "step": 71340
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3786443315483543e-05,
+      "loss": 1.843,
+      "step": 71350
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3784154881230264e-05,
+      "loss": 1.7607,
+      "step": 71360
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.378186644697698e-05,
+      "loss": 1.7073,
+      "step": 71370
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.377957801272369e-05,
+      "loss": 1.7599,
+      "step": 71380
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3777289578470414e-05,
+      "loss": 1.7449,
+      "step": 71390
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.377500114421713e-05,
+      "loss": 1.6075,
+      "step": 71400
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.377271270996384e-05,
+      "loss": 1.8447,
+      "step": 71410
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.377042427571056e-05,
+      "loss": 1.7503,
+      "step": 71420
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.376813584145728e-05,
+      "loss": 1.7352,
+      "step": 71430
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.376584740720399e-05,
+      "loss": 1.7258,
+      "step": 71440
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.376355897295071e-05,
+      "loss": 1.8187,
+      "step": 71450
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3761270538697426e-05,
+      "loss": 1.7211,
+      "step": 71460
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.375898210444414e-05,
+      "loss": 1.6795,
+      "step": 71470
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.375669367019086e-05,
+      "loss": 1.7956,
+      "step": 71480
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3754405235937575e-05,
+      "loss": 1.7201,
+      "step": 71490
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.375211680168429e-05,
+      "loss": 1.7036,
+      "step": 71500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.374982836743101e-05,
+      "loss": 1.4896,
+      "step": 71510
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.374753993317772e-05,
+      "loss": 1.6889,
+      "step": 71520
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.374525149892443e-05,
+      "loss": 1.6816,
+      "step": 71530
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.374296306467115e-05,
+      "loss": 1.8322,
+      "step": 71540
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.374067463041787e-05,
+      "loss": 1.7881,
+      "step": 71550
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.373838619616458e-05,
+      "loss": 1.7111,
+      "step": 71560
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.37360977619113e-05,
+      "loss": 1.7394,
+      "step": 71570
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3733809327658016e-05,
+      "loss": 1.8023,
+      "step": 71580
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.373152089340473e-05,
+      "loss": 1.733,
+      "step": 71590
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.372923245915145e-05,
+      "loss": 1.6427,
+      "step": 71600
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3726944024898165e-05,
+      "loss": 1.8419,
+      "step": 71610
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.372465559064488e-05,
+      "loss": 1.6791,
+      "step": 71620
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.37223671563916e-05,
+      "loss": 1.7931,
+      "step": 71630
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3720078722138314e-05,
+      "loss": 1.7717,
+      "step": 71640
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.371779028788503e-05,
+      "loss": 1.7735,
+      "step": 71650
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.371550185363175e-05,
+      "loss": 1.6324,
+      "step": 71660
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3713213419378464e-05,
+      "loss": 1.8313,
+      "step": 71670
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.371092498512518e-05,
+      "loss": 1.9635,
+      "step": 71680
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.37086365508719e-05,
+      "loss": 1.6416,
+      "step": 71690
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.370634811661861e-05,
+      "loss": 1.7442,
+      "step": 71700
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.370405968236533e-05,
+      "loss": 1.7754,
+      "step": 71710
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.370177124811204e-05,
+      "loss": 1.6126,
+      "step": 71720
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3699482813858755e-05,
+      "loss": 1.5657,
+      "step": 71730
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3697194379605476e-05,
+      "loss": 1.5619,
+      "step": 71740
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.369490594535219e-05,
+      "loss": 1.801,
+      "step": 71750
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3692617511098904e-05,
+      "loss": 1.7297,
+      "step": 71760
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3690329076845625e-05,
+      "loss": 1.5865,
+      "step": 71770
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.368804064259234e-05,
+      "loss": 1.6314,
+      "step": 71780
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3685752208339053e-05,
+      "loss": 1.9618,
+      "step": 71790
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3683463774085774e-05,
+      "loss": 1.6084,
+      "step": 71800
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.368117533983249e-05,
+      "loss": 1.8306,
+      "step": 71810
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.36788869055792e-05,
+      "loss": 1.6027,
+      "step": 71820
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3676598471325924e-05,
+      "loss": 1.6296,
+      "step": 71830
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.367431003707264e-05,
+      "loss": 1.9229,
+      "step": 71840
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.367202160281935e-05,
+      "loss": 1.7909,
+      "step": 71850
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.366973316856607e-05,
+      "loss": 1.6445,
+      "step": 71860
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.366744473431279e-05,
+      "loss": 1.5888,
+      "step": 71870
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.36651563000595e-05,
+      "loss": 1.5867,
+      "step": 71880
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.366286786580622e-05,
+      "loss": 1.4593,
+      "step": 71890
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3660579431552936e-05,
+      "loss": 1.6665,
+      "step": 71900
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.365829099729965e-05,
+      "loss": 1.6219,
+      "step": 71910
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3656002563046364e-05,
+      "loss": 1.692,
+      "step": 71920
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.365371412879308e-05,
+      "loss": 1.5343,
+      "step": 71930
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.365142569453979e-05,
+      "loss": 1.8459,
+      "step": 71940
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3649137260286513e-05,
+      "loss": 1.9061,
+      "step": 71950
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.364684882603323e-05,
+      "loss": 1.5404,
+      "step": 71960
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.364456039177994e-05,
+      "loss": 1.639,
+      "step": 71970
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.364227195752666e-05,
+      "loss": 1.5827,
+      "step": 71980
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.363998352327338e-05,
+      "loss": 1.7827,
+      "step": 71990
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.363769508902009e-05,
+      "loss": 1.4901,
+      "step": 72000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.363540665476681e-05,
+      "loss": 1.7251,
+      "step": 72010
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3633118220513526e-05,
+      "loss": 1.6614,
+      "step": 72020
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.363082978626024e-05,
+      "loss": 1.5981,
+      "step": 72030
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.362854135200696e-05,
+      "loss": 1.8068,
+      "step": 72040
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3626252917753675e-05,
+      "loss": 1.7848,
+      "step": 72050
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.362396448350039e-05,
+      "loss": 1.7673,
+      "step": 72060
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.362167604924711e-05,
+      "loss": 1.7002,
+      "step": 72070
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3619387614993824e-05,
+      "loss": 1.8128,
+      "step": 72080
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.361709918074054e-05,
+      "loss": 1.6612,
+      "step": 72090
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.361481074648726e-05,
+      "loss": 1.7565,
+      "step": 72100
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3612522312233973e-05,
+      "loss": 1.5967,
+      "step": 72110
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.361023387798069e-05,
+      "loss": 1.5583,
+      "step": 72120
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.36079454437274e-05,
+      "loss": 1.7509,
+      "step": 72130
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3605657009474116e-05,
+      "loss": 1.7735,
+      "step": 72140
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.360336857522084e-05,
+      "loss": 1.691,
+      "step": 72150
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.360108014096755e-05,
+      "loss": 1.7482,
+      "step": 72160
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3598791706714265e-05,
+      "loss": 1.8768,
+      "step": 72170
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3596503272460986e-05,
+      "loss": 1.7349,
+      "step": 72180
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.35942148382077e-05,
+      "loss": 1.6129,
+      "step": 72190
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3591926403954414e-05,
+      "loss": 1.7478,
+      "step": 72200
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3589637969701135e-05,
+      "loss": 1.8931,
+      "step": 72210
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.358734953544785e-05,
+      "loss": 1.6801,
+      "step": 72220
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3585061101194563e-05,
+      "loss": 1.9876,
+      "step": 72230
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3582772666941284e-05,
+      "loss": 1.655,
+      "step": 72240
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3580484232688e-05,
+      "loss": 1.6467,
+      "step": 72250
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.357819579843471e-05,
+      "loss": 1.7025,
+      "step": 72260
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3575907364181433e-05,
+      "loss": 1.7961,
+      "step": 72270
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.357361892992815e-05,
+      "loss": 1.8771,
+      "step": 72280
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.357133049567486e-05,
+      "loss": 1.7425,
+      "step": 72290
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.356904206142158e-05,
+      "loss": 1.7531,
+      "step": 72300
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.356675362716829e-05,
+      "loss": 1.6957,
+      "step": 72310
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3564465192915004e-05,
+      "loss": 1.5032,
+      "step": 72320
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3562176758661725e-05,
+      "loss": 1.5315,
+      "step": 72330
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.355988832440844e-05,
+      "loss": 1.7406,
+      "step": 72340
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.355759989015515e-05,
+      "loss": 1.5926,
+      "step": 72350
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3555311455901874e-05,
+      "loss": 1.72,
+      "step": 72360
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.355302302164859e-05,
+      "loss": 1.6877,
+      "step": 72370
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.35507345873953e-05,
+      "loss": 1.6804,
+      "step": 72380
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3548446153142023e-05,
+      "loss": 1.8473,
+      "step": 72390
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.354615771888874e-05,
+      "loss": 1.5904,
+      "step": 72400
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.354386928463545e-05,
+      "loss": 1.788,
+      "step": 72410
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.354158085038217e-05,
+      "loss": 1.6255,
+      "step": 72420
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.353929241612889e-05,
+      "loss": 1.6949,
+      "step": 72430
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.35370039818756e-05,
+      "loss": 1.7558,
+      "step": 72440
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.353471554762232e-05,
+      "loss": 1.7421,
+      "step": 72450
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3532427113369036e-05,
+      "loss": 1.8235,
+      "step": 72460
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.353013867911575e-05,
+      "loss": 1.7697,
+      "step": 72470
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.352785024486247e-05,
+      "loss": 1.7298,
+      "step": 72480
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3525561810609185e-05,
+      "loss": 1.5889,
+      "step": 72490
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.35232733763559e-05,
+      "loss": 1.6287,
+      "step": 72500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.352098494210261e-05,
+      "loss": 1.5948,
+      "step": 72510
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.351869650784933e-05,
+      "loss": 1.7892,
+      "step": 72520
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.351640807359605e-05,
+      "loss": 1.8146,
+      "step": 72530
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.351411963934276e-05,
+      "loss": 1.7375,
+      "step": 72540
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.351183120508948e-05,
+      "loss": 1.5989,
+      "step": 72550
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.35095427708362e-05,
+      "loss": 1.7414,
+      "step": 72560
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.350725433658291e-05,
+      "loss": 1.5913,
+      "step": 72570
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3504965902329626e-05,
+      "loss": 1.8189,
+      "step": 72580
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.350267746807634e-05,
+      "loss": 1.6982,
+      "step": 72590
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.350038903382306e-05,
+      "loss": 1.8096,
+      "step": 72600
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3498100599569775e-05,
+      "loss": 1.7897,
+      "step": 72610
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.349581216531649e-05,
+      "loss": 1.6815,
+      "step": 72620
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.349352373106321e-05,
+      "loss": 1.7916,
+      "step": 72630
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3491235296809924e-05,
+      "loss": 1.7736,
+      "step": 72640
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.348894686255664e-05,
+      "loss": 1.623,
+      "step": 72650
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.348665842830336e-05,
+      "loss": 1.6617,
+      "step": 72660
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.348436999405007e-05,
+      "loss": 1.5431,
+      "step": 72670
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.348208155979679e-05,
+      "loss": 1.5446,
+      "step": 72680
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.347979312554351e-05,
+      "loss": 1.663,
+      "step": 72690
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.347750469129022e-05,
+      "loss": 1.628,
+      "step": 72700
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.347521625703694e-05,
+      "loss": 1.6873,
+      "step": 72710
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.347292782278365e-05,
+      "loss": 1.6843,
+      "step": 72720
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3470639388530365e-05,
+      "loss": 1.718,
+      "step": 72730
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3468350954277086e-05,
+      "loss": 1.5531,
+      "step": 72740
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.34660625200238e-05,
+      "loss": 1.7296,
+      "step": 72750
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3463774085770514e-05,
+      "loss": 1.7352,
+      "step": 72760
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3461485651517235e-05,
+      "loss": 1.7843,
+      "step": 72770
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.345919721726395e-05,
+      "loss": 1.7135,
+      "step": 72780
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.345690878301066e-05,
+      "loss": 1.7062,
+      "step": 72790
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3454620348757384e-05,
+      "loss": 1.7414,
+      "step": 72800
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.34523319145041e-05,
+      "loss": 1.7094,
+      "step": 72810
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.345004348025081e-05,
+      "loss": 1.8046,
+      "step": 72820
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.344775504599753e-05,
+      "loss": 1.5912,
+      "step": 72830
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.344546661174425e-05,
+      "loss": 1.748,
+      "step": 72840
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.344317817749096e-05,
+      "loss": 1.6945,
+      "step": 72850
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.344088974323768e-05,
+      "loss": 1.6895,
+      "step": 72860
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.34386013089844e-05,
+      "loss": 1.7009,
+      "step": 72870
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.343631287473111e-05,
+      "loss": 1.772,
+      "step": 72880
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.343402444047783e-05,
+      "loss": 1.652,
+      "step": 72890
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3431736006224546e-05,
+      "loss": 1.5643,
+      "step": 72900
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.342944757197125e-05,
+      "loss": 1.8231,
+      "step": 72910
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3427159137717974e-05,
+      "loss": 1.6068,
+      "step": 72920
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.342487070346469e-05,
+      "loss": 1.6606,
+      "step": 72930
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.34225822692114e-05,
+      "loss": 1.8327,
+      "step": 72940
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.342029383495812e-05,
+      "loss": 1.6429,
+      "step": 72950
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.341800540070484e-05,
+      "loss": 1.7126,
+      "step": 72960
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.341571696645155e-05,
+      "loss": 1.7168,
+      "step": 72970
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.341342853219827e-05,
+      "loss": 1.6211,
+      "step": 72980
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.341114009794499e-05,
+      "loss": 1.7007,
+      "step": 72990
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.34088516636917e-05,
+      "loss": 1.6622,
+      "step": 73000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.340656322943842e-05,
+      "loss": 1.7673,
+      "step": 73010
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3404274795185136e-05,
+      "loss": 1.775,
+      "step": 73020
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.340198636093185e-05,
+      "loss": 1.5305,
+      "step": 73030
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.339969792667857e-05,
+      "loss": 1.7028,
+      "step": 73040
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3397409492425285e-05,
+      "loss": 1.6083,
+      "step": 73050
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3395121058172e-05,
+      "loss": 1.83,
+      "step": 73060
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.339283262391872e-05,
+      "loss": 1.7735,
+      "step": 73070
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3390544189665434e-05,
+      "loss": 1.7527,
+      "step": 73080
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.338825575541215e-05,
+      "loss": 1.7747,
+      "step": 73090
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.338596732115886e-05,
+      "loss": 1.7574,
+      "step": 73100
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3383678886905577e-05,
+      "loss": 1.7566,
+      "step": 73110
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.33813904526523e-05,
+      "loss": 1.6563,
+      "step": 73120
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.337910201839901e-05,
+      "loss": 1.6414,
+      "step": 73130
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3376813584145726e-05,
+      "loss": 1.6734,
+      "step": 73140
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.337452514989245e-05,
+      "loss": 1.6317,
+      "step": 73150
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.337223671563916e-05,
+      "loss": 1.8106,
+      "step": 73160
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3369948281385875e-05,
+      "loss": 1.7077,
+      "step": 73170
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3367659847132596e-05,
+      "loss": 1.6348,
+      "step": 73180
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.336537141287931e-05,
+      "loss": 1.7752,
+      "step": 73190
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3363082978626024e-05,
+      "loss": 1.7839,
+      "step": 73200
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3360794544372745e-05,
+      "loss": 1.6691,
+      "step": 73210
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.335850611011946e-05,
+      "loss": 1.7464,
+      "step": 73220
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.335621767586617e-05,
+      "loss": 1.7459,
+      "step": 73230
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3353929241612894e-05,
+      "loss": 1.7142,
+      "step": 73240
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.335164080735961e-05,
+      "loss": 1.7486,
+      "step": 73250
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.334935237310632e-05,
+      "loss": 1.6691,
+      "step": 73260
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.334706393885304e-05,
+      "loss": 1.7346,
+      "step": 73270
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.334477550459976e-05,
+      "loss": 1.6266,
+      "step": 73280
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.334248707034647e-05,
+      "loss": 1.7474,
+      "step": 73290
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3340198636093186e-05,
+      "loss": 1.536,
+      "step": 73300
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.33379102018399e-05,
+      "loss": 1.733,
+      "step": 73310
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3335621767586614e-05,
+      "loss": 1.742,
+      "step": 73320
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 1.8243,
+      "step": 73330
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.333104489908005e-05,
+      "loss": 1.737,
+      "step": 73340
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.332875646482676e-05,
+      "loss": 1.9515,
+      "step": 73350
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3326468030573484e-05,
+      "loss": 1.6221,
+      "step": 73360
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.33241795963202e-05,
+      "loss": 1.7299,
+      "step": 73370
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.332189116206691e-05,
+      "loss": 1.6029,
+      "step": 73380
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.331960272781363e-05,
+      "loss": 1.6997,
+      "step": 73390
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.331731429356035e-05,
+      "loss": 1.7373,
+      "step": 73400
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.331502585930706e-05,
+      "loss": 1.7565,
+      "step": 73410
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.331273742505378e-05,
+      "loss": 1.701,
+      "step": 73420
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3310448990800497e-05,
+      "loss": 1.8505,
+      "step": 73430
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.330816055654721e-05,
+      "loss": 1.6318,
+      "step": 73440
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.330587212229393e-05,
+      "loss": 1.6245,
+      "step": 73450
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3303583688040646e-05,
+      "loss": 1.7686,
+      "step": 73460
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.330129525378736e-05,
+      "loss": 1.8403,
+      "step": 73470
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.329900681953408e-05,
+      "loss": 1.6763,
+      "step": 73480
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3296718385280795e-05,
+      "loss": 1.6101,
+      "step": 73490
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.329442995102751e-05,
+      "loss": 1.7168,
+      "step": 73500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.329214151677422e-05,
+      "loss": 1.8389,
+      "step": 73510
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.328985308252094e-05,
+      "loss": 1.7571,
+      "step": 73520
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.328756464826766e-05,
+      "loss": 1.7558,
+      "step": 73530
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.328527621401437e-05,
+      "loss": 1.578,
+      "step": 73540
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3282987779761087e-05,
+      "loss": 1.7713,
+      "step": 73550
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.328069934550781e-05,
+      "loss": 1.592,
+      "step": 73560
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.327841091125452e-05,
+      "loss": 1.6054,
+      "step": 73570
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3276122477001236e-05,
+      "loss": 1.6922,
+      "step": 73580
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.327383404274796e-05,
+      "loss": 1.75,
+      "step": 73590
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.327154560849467e-05,
+      "loss": 1.7739,
+      "step": 73600
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3269257174241385e-05,
+      "loss": 1.5421,
+      "step": 73610
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3266968739988106e-05,
+      "loss": 1.9391,
+      "step": 73620
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.326468030573482e-05,
+      "loss": 1.8903,
+      "step": 73630
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3262391871481534e-05,
+      "loss": 1.5589,
+      "step": 73640
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3260103437228255e-05,
+      "loss": 1.5633,
+      "step": 73650
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.325781500297497e-05,
+      "loss": 1.7194,
+      "step": 73660
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.325552656872168e-05,
+      "loss": 1.7366,
+      "step": 73670
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.32532381344684e-05,
+      "loss": 1.822,
+      "step": 73680
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.325094970021512e-05,
+      "loss": 1.8145,
+      "step": 73690
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3248661265961826e-05,
+      "loss": 1.8021,
+      "step": 73700
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3246372831708547e-05,
+      "loss": 1.7816,
+      "step": 73710
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.324408439745526e-05,
+      "loss": 1.5793,
+      "step": 73720
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3241795963201975e-05,
+      "loss": 1.6637,
+      "step": 73730
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3239507528948696e-05,
+      "loss": 1.8112,
+      "step": 73740
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.323721909469541e-05,
+      "loss": 1.6573,
+      "step": 73750
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3234930660442124e-05,
+      "loss": 1.7347,
+      "step": 73760
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3232642226188845e-05,
+      "loss": 1.7813,
+      "step": 73770
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.323035379193556e-05,
+      "loss": 1.6598,
+      "step": 73780
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.322806535768227e-05,
+      "loss": 1.6269,
+      "step": 73790
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3225776923428994e-05,
+      "loss": 1.7914,
+      "step": 73800
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.322348848917571e-05,
+      "loss": 1.7203,
+      "step": 73810
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.322120005492242e-05,
+      "loss": 1.7126,
+      "step": 73820
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.321891162066914e-05,
+      "loss": 1.6242,
+      "step": 73830
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.321662318641586e-05,
+      "loss": 1.6119,
+      "step": 73840
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.321433475216257e-05,
+      "loss": 1.6824,
+      "step": 73850
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.321204631790929e-05,
+      "loss": 1.719,
+      "step": 73860
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3209757883656007e-05,
+      "loss": 1.7358,
+      "step": 73870
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.320746944940272e-05,
+      "loss": 1.53,
+      "step": 73880
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3205181015149435e-05,
+      "loss": 1.6768,
+      "step": 73890
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.320289258089615e-05,
+      "loss": 1.6963,
+      "step": 73900
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.320060414664286e-05,
+      "loss": 1.7954,
+      "step": 73910
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3198315712389584e-05,
+      "loss": 1.8435,
+      "step": 73920
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.31960272781363e-05,
+      "loss": 1.7507,
+      "step": 73930
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.319373884388301e-05,
+      "loss": 1.8189,
+      "step": 73940
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.319145040962973e-05,
+      "loss": 1.5948,
+      "step": 73950
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.318916197537645e-05,
+      "loss": 1.5239,
+      "step": 73960
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.318687354112316e-05,
+      "loss": 1.6711,
+      "step": 73970
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.318458510686988e-05,
+      "loss": 1.8153,
+      "step": 73980
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3182296672616596e-05,
+      "loss": 1.6177,
+      "step": 73990
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.318000823836331e-05,
+      "loss": 1.6931,
+      "step": 74000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.317771980411003e-05,
+      "loss": 1.6481,
+      "step": 74010
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3175431369856746e-05,
+      "loss": 1.7604,
+      "step": 74020
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.317314293560346e-05,
+      "loss": 1.7801,
+      "step": 74030
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.317085450135018e-05,
+      "loss": 1.6482,
+      "step": 74040
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3168566067096895e-05,
+      "loss": 1.6151,
+      "step": 74050
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.316627763284361e-05,
+      "loss": 1.7753,
+      "step": 74060
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.316398919859033e-05,
+      "loss": 1.8455,
+      "step": 74070
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3161700764337044e-05,
+      "loss": 1.813,
+      "step": 74080
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.315941233008376e-05,
+      "loss": 1.7054,
+      "step": 74090
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.315712389583047e-05,
+      "loss": 1.7242,
+      "step": 74100
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3154835461577186e-05,
+      "loss": 1.7422,
+      "step": 74110
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.315254702732391e-05,
+      "loss": 1.6533,
+      "step": 74120
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.315025859307062e-05,
+      "loss": 1.7513,
+      "step": 74130
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3147970158817336e-05,
+      "loss": 1.6613,
+      "step": 74140
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3145681724564057e-05,
+      "loss": 1.5444,
+      "step": 74150
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.314339329031077e-05,
+      "loss": 1.481,
+      "step": 74160
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3141104856057485e-05,
+      "loss": 1.6732,
+      "step": 74170
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3138816421804206e-05,
+      "loss": 1.6983,
+      "step": 74180
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.313652798755092e-05,
+      "loss": 1.7625,
+      "step": 74190
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3134239553297634e-05,
+      "loss": 1.622,
+      "step": 74200
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3131951119044355e-05,
+      "loss": 1.8223,
+      "step": 74210
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.312966268479107e-05,
+      "loss": 1.7289,
+      "step": 74220
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.312737425053778e-05,
+      "loss": 1.848,
+      "step": 74230
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3125085816284504e-05,
+      "loss": 1.6344,
+      "step": 74240
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.312279738203122e-05,
+      "loss": 1.7061,
+      "step": 74250
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.312050894777793e-05,
+      "loss": 1.5567,
+      "step": 74260
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.311822051352465e-05,
+      "loss": 1.7639,
+      "step": 74270
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.311593207927137e-05,
+      "loss": 1.6403,
+      "step": 74280
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3113643645018075e-05,
+      "loss": 1.6674,
+      "step": 74290
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3111355210764796e-05,
+      "loss": 1.537,
+      "step": 74300
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.310906677651151e-05,
+      "loss": 1.7172,
+      "step": 74310
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3106778342258224e-05,
+      "loss": 1.5103,
+      "step": 74320
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3104489908004945e-05,
+      "loss": 1.6533,
+      "step": 74330
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.310220147375166e-05,
+      "loss": 1.7599,
+      "step": 74340
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.309991303949837e-05,
+      "loss": 1.6975,
+      "step": 74350
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3097624605245094e-05,
+      "loss": 1.7432,
+      "step": 74360
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.309533617099181e-05,
+      "loss": 1.6648,
+      "step": 74370
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.309304773673852e-05,
+      "loss": 1.6104,
+      "step": 74380
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.309075930248524e-05,
+      "loss": 1.4753,
+      "step": 74390
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.308847086823196e-05,
+      "loss": 1.7571,
+      "step": 74400
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.308618243397867e-05,
+      "loss": 1.7483,
+      "step": 74410
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.308389399972539e-05,
+      "loss": 1.6847,
+      "step": 74420
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3081605565472106e-05,
+      "loss": 1.6939,
+      "step": 74430
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.307931713121882e-05,
+      "loss": 1.7334,
+      "step": 74440
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.307702869696554e-05,
+      "loss": 1.8699,
+      "step": 74450
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3074740262712256e-05,
+      "loss": 1.9417,
+      "step": 74460
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.307245182845897e-05,
+      "loss": 1.7054,
+      "step": 74470
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.307016339420569e-05,
+      "loss": 1.6967,
+      "step": 74480
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.30678749599524e-05,
+      "loss": 1.8651,
+      "step": 74490
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.306558652569912e-05,
+      "loss": 1.8968,
+      "step": 74500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.306329809144583e-05,
+      "loss": 1.7462,
+      "step": 74510
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.306100965719255e-05,
+      "loss": 1.6672,
+      "step": 74520
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.305872122293927e-05,
+      "loss": 1.5905,
+      "step": 74530
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.305643278868598e-05,
+      "loss": 1.6638,
+      "step": 74540
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3054144354432696e-05,
+      "loss": 1.609,
+      "step": 74550
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.305185592017942e-05,
+      "loss": 1.5991,
+      "step": 74560
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.304956748592613e-05,
+      "loss": 1.6324,
+      "step": 74570
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3047279051672846e-05,
+      "loss": 1.6873,
+      "step": 74580
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3044990617419566e-05,
+      "loss": 1.8176,
+      "step": 74590
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.304270218316628e-05,
+      "loss": 1.7727,
+      "step": 74600
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3040413748912995e-05,
+      "loss": 1.8829,
+      "step": 74610
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3038125314659716e-05,
+      "loss": 1.6417,
+      "step": 74620
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.303583688040643e-05,
+      "loss": 1.6906,
+      "step": 74630
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3033548446153144e-05,
+      "loss": 1.9113,
+      "step": 74640
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3031260011899865e-05,
+      "loss": 1.7408,
+      "step": 74650
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.302897157764658e-05,
+      "loss": 1.5889,
+      "step": 74660
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.302668314339329e-05,
+      "loss": 1.8235,
+      "step": 74670
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.302439470914001e-05,
+      "loss": 1.6361,
+      "step": 74680
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.302210627488672e-05,
+      "loss": 1.5696,
+      "step": 74690
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3019817840633436e-05,
+      "loss": 1.786,
+      "step": 74700
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3017529406380156e-05,
+      "loss": 1.6528,
+      "step": 74710
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.301524097212687e-05,
+      "loss": 1.7462,
+      "step": 74720
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3012952537873585e-05,
+      "loss": 1.7443,
+      "step": 74730
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3010664103620306e-05,
+      "loss": 1.5619,
+      "step": 74740
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.300837566936702e-05,
+      "loss": 1.9888,
+      "step": 74750
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3006087235113734e-05,
+      "loss": 1.6725,
+      "step": 74760
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3003798800860455e-05,
+      "loss": 1.6887,
+      "step": 74770
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.300151036660717e-05,
+      "loss": 1.6146,
+      "step": 74780
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.299922193235388e-05,
+      "loss": 1.7002,
+      "step": 74790
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2996933498100604e-05,
+      "loss": 1.6588,
+      "step": 74800
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.299464506384732e-05,
+      "loss": 1.7134,
+      "step": 74810
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.299235662959403e-05,
+      "loss": 1.6959,
+      "step": 74820
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.299006819534075e-05,
+      "loss": 1.8146,
+      "step": 74830
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.298777976108747e-05,
+      "loss": 1.8,
+      "step": 74840
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.298549132683418e-05,
+      "loss": 1.6556,
+      "step": 74850
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.29832028925809e-05,
+      "loss": 1.8278,
+      "step": 74860
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2980914458327616e-05,
+      "loss": 1.579,
+      "step": 74870
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.297862602407433e-05,
+      "loss": 1.6853,
+      "step": 74880
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2976337589821045e-05,
+      "loss": 1.6707,
+      "step": 74890
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.297404915556776e-05,
+      "loss": 1.6188,
+      "step": 74900
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.297176072131448e-05,
+      "loss": 1.5936,
+      "step": 74910
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2969472287061194e-05,
+      "loss": 1.6282,
+      "step": 74920
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.296718385280791e-05,
+      "loss": 1.7848,
+      "step": 74930
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.296489541855462e-05,
+      "loss": 1.6508,
+      "step": 74940
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.296260698430134e-05,
+      "loss": 1.8478,
+      "step": 74950
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.296031855004806e-05,
+      "loss": 1.6811,
+      "step": 74960
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.295803011579477e-05,
+      "loss": 1.6804,
+      "step": 74970
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.295574168154149e-05,
+      "loss": 1.6858,
+      "step": 74980
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2953453247288206e-05,
+      "loss": 1.6693,
+      "step": 74990
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.295116481303492e-05,
+      "loss": 1.7578,
+      "step": 75000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.294887637878164e-05,
+      "loss": 1.6495,
+      "step": 75010
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2946587944528356e-05,
+      "loss": 1.6666,
+      "step": 75020
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.294429951027507e-05,
+      "loss": 1.5204,
+      "step": 75030
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.294201107602179e-05,
+      "loss": 1.6596,
+      "step": 75040
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2939722641768505e-05,
+      "loss": 1.5945,
+      "step": 75050
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.293743420751522e-05,
+      "loss": 1.7712,
+      "step": 75060
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.293514577326194e-05,
+      "loss": 1.792,
+      "step": 75070
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.293285733900865e-05,
+      "loss": 1.7408,
+      "step": 75080
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.293056890475537e-05,
+      "loss": 1.7591,
+      "step": 75090
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.292828047050208e-05,
+      "loss": 1.7925,
+      "step": 75100
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2925992036248796e-05,
+      "loss": 1.5578,
+      "step": 75110
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.292370360199552e-05,
+      "loss": 1.8059,
+      "step": 75120
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.292141516774223e-05,
+      "loss": 1.6076,
+      "step": 75130
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2919126733488945e-05,
+      "loss": 1.6546,
+      "step": 75140
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2916838299235666e-05,
+      "loss": 1.7152,
+      "step": 75150
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.291454986498238e-05,
+      "loss": 1.562,
+      "step": 75160
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2912261430729095e-05,
+      "loss": 1.7394,
+      "step": 75170
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2909972996475816e-05,
+      "loss": 1.5244,
+      "step": 75180
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.290768456222253e-05,
+      "loss": 1.5927,
+      "step": 75190
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2905396127969244e-05,
+      "loss": 1.6773,
+      "step": 75200
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2903107693715965e-05,
+      "loss": 1.6635,
+      "step": 75210
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.290081925946268e-05,
+      "loss": 1.6701,
+      "step": 75220
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.289853082520939e-05,
+      "loss": 1.6883,
+      "step": 75230
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2896242390956114e-05,
+      "loss": 1.6496,
+      "step": 75240
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.289395395670283e-05,
+      "loss": 1.6142,
+      "step": 75250
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.289166552244954e-05,
+      "loss": 1.7788,
+      "step": 75260
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.288937708819626e-05,
+      "loss": 1.7005,
+      "step": 75270
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.288708865394297e-05,
+      "loss": 1.6255,
+      "step": 75280
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2884800219689685e-05,
+      "loss": 1.6137,
+      "step": 75290
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2882511785436405e-05,
+      "loss": 1.6572,
+      "step": 75300
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.288022335118312e-05,
+      "loss": 1.7105,
+      "step": 75310
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2877934916929834e-05,
+      "loss": 1.7652,
+      "step": 75320
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2875646482676555e-05,
+      "loss": 1.6192,
+      "step": 75330
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.287335804842327e-05,
+      "loss": 1.6413,
+      "step": 75340
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.287106961416998e-05,
+      "loss": 1.6996,
+      "step": 75350
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2868781179916704e-05,
+      "loss": 1.6881,
+      "step": 75360
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.286649274566342e-05,
+      "loss": 1.8135,
+      "step": 75370
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.286420431141013e-05,
+      "loss": 1.9046,
+      "step": 75380
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.286191587715685e-05,
+      "loss": 1.5952,
+      "step": 75390
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.285962744290357e-05,
+      "loss": 1.5994,
+      "step": 75400
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.285733900865028e-05,
+      "loss": 1.748,
+      "step": 75410
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2855050574397e-05,
+      "loss": 1.663,
+      "step": 75420
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2852762140143716e-05,
+      "loss": 1.8038,
+      "step": 75430
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.285047370589043e-05,
+      "loss": 1.9155,
+      "step": 75440
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.284818527163715e-05,
+      "loss": 1.6643,
+      "step": 75450
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2845896837383866e-05,
+      "loss": 1.5925,
+      "step": 75460
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.284360840313058e-05,
+      "loss": 1.6411,
+      "step": 75470
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2841319968877294e-05,
+      "loss": 1.6664,
+      "step": 75480
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.283903153462401e-05,
+      "loss": 1.6516,
+      "step": 75490
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.283674310037073e-05,
+      "loss": 1.6547,
+      "step": 75500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.283445466611744e-05,
+      "loss": 1.7078,
+      "step": 75510
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.283216623186416e-05,
+      "loss": 1.6215,
+      "step": 75520
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.282987779761088e-05,
+      "loss": 1.7134,
+      "step": 75530
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.282758936335759e-05,
+      "loss": 1.8529,
+      "step": 75540
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2825300929104306e-05,
+      "loss": 1.572,
+      "step": 75550
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.282301249485103e-05,
+      "loss": 1.9712,
+      "step": 75560
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.282072406059774e-05,
+      "loss": 1.7543,
+      "step": 75570
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2818435626344455e-05,
+      "loss": 1.5348,
+      "step": 75580
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2816147192091176e-05,
+      "loss": 1.9183,
+      "step": 75590
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.281385875783789e-05,
+      "loss": 1.7568,
+      "step": 75600
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2811570323584605e-05,
+      "loss": 1.6799,
+      "step": 75610
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2809281889331326e-05,
+      "loss": 1.7777,
+      "step": 75620
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.280699345507804e-05,
+      "loss": 1.7359,
+      "step": 75630
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2804705020824754e-05,
+      "loss": 1.6255,
+      "step": 75640
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2802416586571475e-05,
+      "loss": 1.5748,
+      "step": 75650
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.280012815231819e-05,
+      "loss": 1.5939,
+      "step": 75660
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2797839718064896e-05,
+      "loss": 1.6073,
+      "step": 75670
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.279555128381162e-05,
+      "loss": 1.6391,
+      "step": 75680
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.279326284955833e-05,
+      "loss": 1.6295,
+      "step": 75690
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2790974415305045e-05,
+      "loss": 1.7964,
+      "step": 75700
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2788685981051766e-05,
+      "loss": 1.8895,
+      "step": 75710
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.278639754679848e-05,
+      "loss": 1.582,
+      "step": 75720
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2784109112545195e-05,
+      "loss": 1.7056,
+      "step": 75730
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2781820678291915e-05,
+      "loss": 1.681,
+      "step": 75740
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.277953224403863e-05,
+      "loss": 1.4447,
+      "step": 75750
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2777243809785344e-05,
+      "loss": 1.65,
+      "step": 75760
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2774955375532065e-05,
+      "loss": 1.6501,
+      "step": 75770
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.277266694127878e-05,
+      "loss": 1.6505,
+      "step": 75780
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.277037850702549e-05,
+      "loss": 1.7238,
+      "step": 75790
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2768090072772214e-05,
+      "loss": 1.7853,
+      "step": 75800
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.276580163851893e-05,
+      "loss": 1.597,
+      "step": 75810
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.276351320426564e-05,
+      "loss": 1.7719,
+      "step": 75820
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.276122477001236e-05,
+      "loss": 1.839,
+      "step": 75830
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.275893633575908e-05,
+      "loss": 1.7726,
+      "step": 75840
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.275664790150579e-05,
+      "loss": 1.6963,
+      "step": 75850
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.275435946725251e-05,
+      "loss": 1.6878,
+      "step": 75860
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.275207103299922e-05,
+      "loss": 1.8568,
+      "step": 75870
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.274978259874594e-05,
+      "loss": 1.652,
+      "step": 75880
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2747494164492655e-05,
+      "loss": 1.4984,
+      "step": 75890
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.274520573023937e-05,
+      "loss": 1.8262,
+      "step": 75900
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.274291729598609e-05,
+      "loss": 1.6528,
+      "step": 75910
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2740628861732804e-05,
+      "loss": 1.5731,
+      "step": 75920
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.273834042747952e-05,
+      "loss": 1.5589,
+      "step": 75930
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.273605199322624e-05,
+      "loss": 1.6699,
+      "step": 75940
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.273376355897295e-05,
+      "loss": 1.6491,
+      "step": 75950
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.273147512471967e-05,
+      "loss": 1.7113,
+      "step": 75960
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.272918669046639e-05,
+      "loss": 1.5401,
+      "step": 75970
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.27268982562131e-05,
+      "loss": 1.8276,
+      "step": 75980
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2724609821959816e-05,
+      "loss": 1.7859,
+      "step": 75990
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.272232138770654e-05,
+      "loss": 1.7421,
+      "step": 76000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.272003295345325e-05,
+      "loss": 1.6643,
+      "step": 76010
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2717744519199965e-05,
+      "loss": 1.7332,
+      "step": 76020
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.271545608494668e-05,
+      "loss": 1.6165,
+      "step": 76030
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.27131676506934e-05,
+      "loss": 1.7256,
+      "step": 76040
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2710879216440115e-05,
+      "loss": 1.7982,
+      "step": 76050
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.270859078218683e-05,
+      "loss": 1.4954,
+      "step": 76060
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.270630234793354e-05,
+      "loss": 1.6416,
+      "step": 76070
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.270401391368026e-05,
+      "loss": 1.6924,
+      "step": 76080
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.270172547942698e-05,
+      "loss": 1.7071,
+      "step": 76090
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.269943704517369e-05,
+      "loss": 1.4712,
+      "step": 76100
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2697148610920406e-05,
+      "loss": 1.6771,
+      "step": 76110
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.269486017666713e-05,
+      "loss": 1.7919,
+      "step": 76120
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.269257174241384e-05,
+      "loss": 1.5675,
+      "step": 76130
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2690283308160555e-05,
+      "loss": 1.7159,
+      "step": 76140
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2687994873907276e-05,
+      "loss": 1.7838,
+      "step": 76150
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.268570643965399e-05,
+      "loss": 1.7128,
+      "step": 76160
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2683418005400705e-05,
+      "loss": 1.6506,
+      "step": 76170
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2681129571147425e-05,
+      "loss": 1.7172,
+      "step": 76180
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.267884113689414e-05,
+      "loss": 1.6813,
+      "step": 76190
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2676552702640854e-05,
+      "loss": 1.9065,
+      "step": 76200
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2674264268387575e-05,
+      "loss": 1.6612,
+      "step": 76210
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.267197583413429e-05,
+      "loss": 1.6596,
+      "step": 76220
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2669687399881e-05,
+      "loss": 1.7513,
+      "step": 76230
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2667398965627724e-05,
+      "loss": 1.8113,
+      "step": 76240
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.266511053137444e-05,
+      "loss": 1.7065,
+      "step": 76250
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.266282209712115e-05,
+      "loss": 1.6547,
+      "step": 76260
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2660533662867866e-05,
+      "loss": 1.6889,
+      "step": 76270
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.265824522861458e-05,
+      "loss": 1.5447,
+      "step": 76280
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2655956794361294e-05,
+      "loss": 1.6932,
+      "step": 76290
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2653668360108015e-05,
+      "loss": 1.7708,
+      "step": 76300
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.265137992585473e-05,
+      "loss": 1.6732,
+      "step": 76310
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2649091491601444e-05,
+      "loss": 1.7099,
+      "step": 76320
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2646803057348165e-05,
+      "loss": 1.8583,
+      "step": 76330
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.264451462309488e-05,
+      "loss": 1.7078,
+      "step": 76340
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.264222618884159e-05,
+      "loss": 1.5725,
+      "step": 76350
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2639937754588314e-05,
+      "loss": 1.6581,
+      "step": 76360
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.263764932033503e-05,
+      "loss": 1.6479,
+      "step": 76370
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.263536088608174e-05,
+      "loss": 1.6662,
+      "step": 76380
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.263307245182846e-05,
+      "loss": 1.6826,
+      "step": 76390
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.263078401757518e-05,
+      "loss": 1.7121,
+      "step": 76400
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.262849558332189e-05,
+      "loss": 1.7229,
+      "step": 76410
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.262620714906861e-05,
+      "loss": 1.7043,
+      "step": 76420
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2623918714815326e-05,
+      "loss": 1.6629,
+      "step": 76430
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.262163028056204e-05,
+      "loss": 1.6082,
+      "step": 76440
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.261934184630876e-05,
+      "loss": 1.7758,
+      "step": 76450
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.261705341205547e-05,
+      "loss": 1.5285,
+      "step": 76460
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.261476497780219e-05,
+      "loss": 1.6832,
+      "step": 76470
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2612476543548904e-05,
+      "loss": 1.8026,
+      "step": 76480
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.261018810929562e-05,
+      "loss": 1.697,
+      "step": 76490
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.260789967504234e-05,
+      "loss": 1.6911,
+      "step": 76500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.260561124078905e-05,
+      "loss": 1.8949,
+      "step": 76510
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.260332280653577e-05,
+      "loss": 1.6984,
+      "step": 76520
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.260103437228249e-05,
+      "loss": 1.7765,
+      "step": 76530
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.25987459380292e-05,
+      "loss": 1.5611,
+      "step": 76540
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2596457503775916e-05,
+      "loss": 1.7047,
+      "step": 76550
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.259416906952264e-05,
+      "loss": 1.7235,
+      "step": 76560
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.259188063526935e-05,
+      "loss": 1.7394,
+      "step": 76570
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2589592201016065e-05,
+      "loss": 1.6705,
+      "step": 76580
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2587303766762786e-05,
+      "loss": 1.5228,
+      "step": 76590
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.25850153325095e-05,
+      "loss": 1.8227,
+      "step": 76600
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2582726898256214e-05,
+      "loss": 1.8006,
+      "step": 76610
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2580438464002935e-05,
+      "loss": 1.4395,
+      "step": 76620
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.257815002974965e-05,
+      "loss": 1.5796,
+      "step": 76630
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2575861595496364e-05,
+      "loss": 1.7321,
+      "step": 76640
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2573573161243085e-05,
+      "loss": 1.6113,
+      "step": 76650
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.257128472698979e-05,
+      "loss": 1.8255,
+      "step": 76660
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2568996292736506e-05,
+      "loss": 1.8957,
+      "step": 76670
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.256670785848323e-05,
+      "loss": 1.6731,
+      "step": 76680
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.256441942422994e-05,
+      "loss": 1.8003,
+      "step": 76690
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2562130989976655e-05,
+      "loss": 1.7491,
+      "step": 76700
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2559842555723376e-05,
+      "loss": 1.7095,
+      "step": 76710
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.255755412147009e-05,
+      "loss": 1.6674,
+      "step": 76720
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2555265687216804e-05,
+      "loss": 1.9088,
+      "step": 76730
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2552977252963525e-05,
+      "loss": 1.6795,
+      "step": 76740
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.255068881871024e-05,
+      "loss": 1.6476,
+      "step": 76750
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2548400384456954e-05,
+      "loss": 1.8876,
+      "step": 76760
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2546111950203674e-05,
+      "loss": 1.4741,
+      "step": 76770
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.254382351595039e-05,
+      "loss": 1.7159,
+      "step": 76780
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.25415350816971e-05,
+      "loss": 1.687,
+      "step": 76790
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2539246647443824e-05,
+      "loss": 1.775,
+      "step": 76800
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.253695821319054e-05,
+      "loss": 1.5854,
+      "step": 76810
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.253466977893725e-05,
+      "loss": 1.689,
+      "step": 76820
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.253238134468397e-05,
+      "loss": 1.4563,
+      "step": 76830
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.253009291043069e-05,
+      "loss": 1.7187,
+      "step": 76840
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.25278044761774e-05,
+      "loss": 1.6512,
+      "step": 76850
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2525516041924115e-05,
+      "loss": 1.6661,
+      "step": 76860
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.252322760767083e-05,
+      "loss": 1.7632,
+      "step": 76870
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.252093917341755e-05,
+      "loss": 1.6853,
+      "step": 76880
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2518650739164264e-05,
+      "loss": 1.7831,
+      "step": 76890
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.251636230491098e-05,
+      "loss": 1.7112,
+      "step": 76900
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.25140738706577e-05,
+      "loss": 1.6754,
+      "step": 76910
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2511785436404414e-05,
+      "loss": 1.5235,
+      "step": 76920
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.250949700215113e-05,
+      "loss": 1.8453,
+      "step": 76930
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.250720856789785e-05,
+      "loss": 1.8138,
+      "step": 76940
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.250492013364456e-05,
+      "loss": 1.6592,
+      "step": 76950
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.250263169939128e-05,
+      "loss": 1.6883,
+      "step": 76960
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2500343265138e-05,
+      "loss": 1.7781,
+      "step": 76970
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.249805483088471e-05,
+      "loss": 1.5321,
+      "step": 76980
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2495766396631426e-05,
+      "loss": 1.7357,
+      "step": 76990
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.249347796237815e-05,
+      "loss": 1.6853,
+      "step": 77000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.249118952812486e-05,
+      "loss": 1.6704,
+      "step": 77010
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2488901093871575e-05,
+      "loss": 1.7481,
+      "step": 77020
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2486612659618296e-05,
+      "loss": 1.5829,
+      "step": 77030
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.248432422536501e-05,
+      "loss": 1.6948,
+      "step": 77040
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2482035791111724e-05,
+      "loss": 1.6639,
+      "step": 77050
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.247974735685844e-05,
+      "loss": 1.4829,
+      "step": 77060
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.247745892260515e-05,
+      "loss": 1.5416,
+      "step": 77070
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.247517048835187e-05,
+      "loss": 1.5415,
+      "step": 77080
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.247288205409859e-05,
+      "loss": 1.5349,
+      "step": 77090
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.24705936198453e-05,
+      "loss": 1.6664,
+      "step": 77100
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2468305185592016e-05,
+      "loss": 1.7662,
+      "step": 77110
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.246601675133874e-05,
+      "loss": 1.645,
+      "step": 77120
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.246372831708545e-05,
+      "loss": 1.7637,
+      "step": 77130
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2461439882832165e-05,
+      "loss": 1.6318,
+      "step": 77140
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2459151448578886e-05,
+      "loss": 1.6508,
+      "step": 77150
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.24568630143256e-05,
+      "loss": 1.5847,
+      "step": 77160
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2454574580072314e-05,
+      "loss": 1.8042,
+      "step": 77170
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2452286145819035e-05,
+      "loss": 1.7716,
+      "step": 77180
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.244999771156575e-05,
+      "loss": 1.752,
+      "step": 77190
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2447709277312464e-05,
+      "loss": 1.6295,
+      "step": 77200
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2445420843059184e-05,
+      "loss": 1.705,
+      "step": 77210
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.24431324088059e-05,
+      "loss": 1.7088,
+      "step": 77220
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.244084397455261e-05,
+      "loss": 1.7573,
+      "step": 77230
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2438555540299334e-05,
+      "loss": 1.6137,
+      "step": 77240
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.243626710604604e-05,
+      "loss": 1.6023,
+      "step": 77250
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2433978671792755e-05,
+      "loss": 1.6547,
+      "step": 77260
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2431690237539476e-05,
+      "loss": 1.6089,
+      "step": 77270
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.242940180328619e-05,
+      "loss": 1.5887,
+      "step": 77280
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2427113369032904e-05,
+      "loss": 1.7977,
+      "step": 77290
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2424824934779625e-05,
+      "loss": 1.5108,
+      "step": 77300
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.242253650052634e-05,
+      "loss": 1.4521,
+      "step": 77310
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2420248066273053e-05,
+      "loss": 1.6904,
+      "step": 77320
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2417959632019774e-05,
+      "loss": 1.627,
+      "step": 77330
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.241567119776649e-05,
+      "loss": 1.7171,
+      "step": 77340
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.24133827635132e-05,
+      "loss": 1.6028,
+      "step": 77350
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2411094329259924e-05,
+      "loss": 1.5792,
+      "step": 77360
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.240880589500664e-05,
+      "loss": 1.7689,
+      "step": 77370
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.240651746075335e-05,
+      "loss": 1.652,
+      "step": 77380
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.240422902650007e-05,
+      "loss": 1.7295,
+      "step": 77390
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.240194059224679e-05,
+      "loss": 1.781,
+      "step": 77400
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.23996521579935e-05,
+      "loss": 1.6914,
+      "step": 77410
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.239736372374022e-05,
+      "loss": 1.6737,
+      "step": 77420
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2395075289486936e-05,
+      "loss": 1.5848,
+      "step": 77430
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.239278685523365e-05,
+      "loss": 1.8382,
+      "step": 77440
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2390498420980364e-05,
+      "loss": 1.6334,
+      "step": 77450
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.238820998672708e-05,
+      "loss": 1.6681,
+      "step": 77460
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.23859215524738e-05,
+      "loss": 1.5744,
+      "step": 77470
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2383633118220514e-05,
+      "loss": 1.7585,
+      "step": 77480
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.238134468396723e-05,
+      "loss": 1.7157,
+      "step": 77490
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.237905624971395e-05,
+      "loss": 1.7179,
+      "step": 77500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.237676781546066e-05,
+      "loss": 1.8158,
+      "step": 77510
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.237447938120738e-05,
+      "loss": 1.5499,
+      "step": 77520
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.23721909469541e-05,
+      "loss": 1.8336,
+      "step": 77530
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.236990251270081e-05,
+      "loss": 1.6128,
+      "step": 77540
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2367614078447526e-05,
+      "loss": 1.7412,
+      "step": 77550
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.236532564419425e-05,
+      "loss": 1.7701,
+      "step": 77560
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.236303720994096e-05,
+      "loss": 1.6913,
+      "step": 77570
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2360748775687675e-05,
+      "loss": 1.7692,
+      "step": 77580
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2358460341434396e-05,
+      "loss": 1.5486,
+      "step": 77590
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.235617190718111e-05,
+      "loss": 1.753,
+      "step": 77600
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2353883472927824e-05,
+      "loss": 1.6771,
+      "step": 77610
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2351595038674545e-05,
+      "loss": 1.71,
+      "step": 77620
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.234930660442126e-05,
+      "loss": 1.7746,
+      "step": 77630
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2347018170167974e-05,
+      "loss": 1.6932,
+      "step": 77640
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.234472973591469e-05,
+      "loss": 1.6793,
+      "step": 77650
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.23424413016614e-05,
+      "loss": 1.7064,
+      "step": 77660
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2340152867408116e-05,
+      "loss": 1.5605,
+      "step": 77670
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.233786443315484e-05,
+      "loss": 1.8103,
+      "step": 77680
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.233557599890155e-05,
+      "loss": 1.6155,
+      "step": 77690
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2333287564648265e-05,
+      "loss": 1.6922,
+      "step": 77700
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2330999130394986e-05,
+      "loss": 1.6491,
+      "step": 77710
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.23287106961417e-05,
+      "loss": 1.5504,
+      "step": 77720
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2326422261888414e-05,
+      "loss": 1.6911,
+      "step": 77730
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2324133827635135e-05,
+      "loss": 1.8003,
+      "step": 77740
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.232184539338185e-05,
+      "loss": 1.6766,
+      "step": 77750
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2319556959128563e-05,
+      "loss": 1.5878,
+      "step": 77760
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2317268524875284e-05,
+      "loss": 1.6403,
+      "step": 77770
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2314980090622e-05,
+      "loss": 1.6473,
+      "step": 77780
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.231269165636871e-05,
+      "loss": 1.5951,
+      "step": 77790
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2310403222115434e-05,
+      "loss": 1.5374,
+      "step": 77800
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.230811478786215e-05,
+      "loss": 1.6006,
+      "step": 77810
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.230582635360886e-05,
+      "loss": 1.5512,
+      "step": 77820
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.230353791935558e-05,
+      "loss": 1.6216,
+      "step": 77830
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.23012494851023e-05,
+      "loss": 1.5199,
+      "step": 77840
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.229896105084901e-05,
+      "loss": 1.712,
+      "step": 77850
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2296672616595725e-05,
+      "loss": 1.7751,
+      "step": 77860
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.229438418234244e-05,
+      "loss": 1.6454,
+      "step": 77870
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.229209574808916e-05,
+      "loss": 1.5657,
+      "step": 77880
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2289807313835874e-05,
+      "loss": 1.7216,
+      "step": 77890
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.228751887958259e-05,
+      "loss": 1.6073,
+      "step": 77900
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.228523044532931e-05,
+      "loss": 1.5301,
+      "step": 77910
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2282942011076023e-05,
+      "loss": 1.6478,
+      "step": 77920
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.228065357682274e-05,
+      "loss": 1.5391,
+      "step": 77930
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.227836514256946e-05,
+      "loss": 1.7804,
+      "step": 77940
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.227607670831617e-05,
+      "loss": 1.7333,
+      "step": 77950
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.227378827406289e-05,
+      "loss": 1.6855,
+      "step": 77960
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.227149983980961e-05,
+      "loss": 1.768,
+      "step": 77970
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.226921140555632e-05,
+      "loss": 1.5113,
+      "step": 77980
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2266922971303036e-05,
+      "loss": 1.7422,
+      "step": 77990
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.226463453704976e-05,
+      "loss": 1.6449,
+      "step": 78000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.226234610279647e-05,
+      "loss": 1.5107,
+      "step": 78010
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2260057668543185e-05,
+      "loss": 1.5874,
+      "step": 78020
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2257769234289906e-05,
+      "loss": 1.5491,
+      "step": 78030
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2255480800036613e-05,
+      "loss": 1.5895,
+      "step": 78040
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.225319236578333e-05,
+      "loss": 1.7431,
+      "step": 78050
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.225090393153005e-05,
+      "loss": 1.6326,
+      "step": 78060
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.224861549727676e-05,
+      "loss": 1.7783,
+      "step": 78070
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.224632706302348e-05,
+      "loss": 1.6578,
+      "step": 78080
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.22440386287702e-05,
+      "loss": 1.8406,
+      "step": 78090
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.224175019451691e-05,
+      "loss": 1.8701,
+      "step": 78100
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2239461760263626e-05,
+      "loss": 1.6868,
+      "step": 78110
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.223717332601035e-05,
+      "loss": 1.5885,
+      "step": 78120
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.223488489175706e-05,
+      "loss": 1.7367,
+      "step": 78130
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2232596457503775e-05,
+      "loss": 1.5605,
+      "step": 78140
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2230308023250496e-05,
+      "loss": 1.7429,
+      "step": 78150
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.222801958899721e-05,
+      "loss": 1.8632,
+      "step": 78160
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2225731154743924e-05,
+      "loss": 1.7679,
+      "step": 78170
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2223442720490645e-05,
+      "loss": 1.6784,
+      "step": 78180
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.222115428623736e-05,
+      "loss": 1.8321,
+      "step": 78190
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2218865851984073e-05,
+      "loss": 1.6323,
+      "step": 78200
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2216577417730794e-05,
+      "loss": 1.6403,
+      "step": 78210
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.221428898347751e-05,
+      "loss": 1.7277,
+      "step": 78220
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.221200054922422e-05,
+      "loss": 1.7748,
+      "step": 78230
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.220971211497094e-05,
+      "loss": 1.6014,
+      "step": 78240
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.220742368071765e-05,
+      "loss": 1.6125,
+      "step": 78250
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.220513524646437e-05,
+      "loss": 1.7018,
+      "step": 78260
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2202846812211086e-05,
+      "loss": 1.6987,
+      "step": 78270
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.22005583779578e-05,
+      "loss": 1.7886,
+      "step": 78280
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.219826994370452e-05,
+      "loss": 1.5049,
+      "step": 78290
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2195981509451235e-05,
+      "loss": 1.5657,
+      "step": 78300
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.219369307519795e-05,
+      "loss": 1.669,
+      "step": 78310
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.219140464094467e-05,
+      "loss": 1.6048,
+      "step": 78320
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2189116206691384e-05,
+      "loss": 1.695,
+      "step": 78330
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.21868277724381e-05,
+      "loss": 1.7106,
+      "step": 78340
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.218453933818482e-05,
+      "loss": 1.6297,
+      "step": 78350
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2182250903931533e-05,
+      "loss": 1.6915,
+      "step": 78360
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.217996246967825e-05,
+      "loss": 1.5926,
+      "step": 78370
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.217767403542496e-05,
+      "loss": 1.6168,
+      "step": 78380
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.217538560117168e-05,
+      "loss": 1.576,
+      "step": 78390
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.21730971669184e-05,
+      "loss": 1.9817,
+      "step": 78400
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.217080873266511e-05,
+      "loss": 1.5982,
+      "step": 78410
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.216852029841183e-05,
+      "loss": 1.773,
+      "step": 78420
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2166231864158546e-05,
+      "loss": 1.8565,
+      "step": 78430
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.216394342990526e-05,
+      "loss": 1.624,
+      "step": 78440
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2161654995651974e-05,
+      "loss": 1.7994,
+      "step": 78450
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.215936656139869e-05,
+      "loss": 1.6861,
+      "step": 78460
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.215707812714541e-05,
+      "loss": 1.6971,
+      "step": 78470
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.215478969289212e-05,
+      "loss": 1.9033,
+      "step": 78480
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.215250125863884e-05,
+      "loss": 1.4987,
+      "step": 78490
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.215021282438556e-05,
+      "loss": 1.5623,
+      "step": 78500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.214792439013227e-05,
+      "loss": 1.6077,
+      "step": 78510
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.214563595587899e-05,
+      "loss": 1.7211,
+      "step": 78520
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.214334752162571e-05,
+      "loss": 1.6721,
+      "step": 78530
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.214105908737242e-05,
+      "loss": 1.5019,
+      "step": 78540
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2138770653119136e-05,
+      "loss": 1.5181,
+      "step": 78550
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.213648221886586e-05,
+      "loss": 1.8064,
+      "step": 78560
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.213419378461257e-05,
+      "loss": 1.7079,
+      "step": 78570
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2131905350359285e-05,
+      "loss": 1.5535,
+      "step": 78580
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2129616916106006e-05,
+      "loss": 1.712,
+      "step": 78590
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.212732848185272e-05,
+      "loss": 1.6997,
+      "step": 78600
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2125040047599434e-05,
+      "loss": 1.6238,
+      "step": 78610
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2122751613346155e-05,
+      "loss": 1.7036,
+      "step": 78620
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.212046317909287e-05,
+      "loss": 1.6542,
+      "step": 78630
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.211817474483958e-05,
+      "loss": 1.6513,
+      "step": 78640
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.21158863105863e-05,
+      "loss": 1.9484,
+      "step": 78650
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.211359787633301e-05,
+      "loss": 1.7921,
+      "step": 78660
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2111309442079726e-05,
+      "loss": 1.805,
+      "step": 78670
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.210902100782645e-05,
+      "loss": 1.7106,
+      "step": 78680
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.210673257357316e-05,
+      "loss": 1.6199,
+      "step": 78690
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2104444139319875e-05,
+      "loss": 1.6199,
+      "step": 78700
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2102155705066596e-05,
+      "loss": 1.6098,
+      "step": 78710
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.209986727081331e-05,
+      "loss": 1.628,
+      "step": 78720
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2097578836560024e-05,
+      "loss": 1.4949,
+      "step": 78730
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2095290402306745e-05,
+      "loss": 1.6325,
+      "step": 78740
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.209300196805346e-05,
+      "loss": 1.6204,
+      "step": 78750
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.209071353380017e-05,
+      "loss": 1.5225,
+      "step": 78760
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2088425099546894e-05,
+      "loss": 1.8724,
+      "step": 78770
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.208613666529361e-05,
+      "loss": 1.6619,
+      "step": 78780
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.208384823104032e-05,
+      "loss": 1.6951,
+      "step": 78790
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2081559796787043e-05,
+      "loss": 1.7784,
+      "step": 78800
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.207927136253376e-05,
+      "loss": 1.7251,
+      "step": 78810
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.207698292828047e-05,
+      "loss": 1.7584,
+      "step": 78820
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2074694494027186e-05,
+      "loss": 1.7043,
+      "step": 78830
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.20724060597739e-05,
+      "loss": 1.5972,
+      "step": 78840
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.207011762552062e-05,
+      "loss": 1.4774,
+      "step": 78850
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2067829191267335e-05,
+      "loss": 1.6825,
+      "step": 78860
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.206554075701405e-05,
+      "loss": 1.7122,
+      "step": 78870
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.206325232276077e-05,
+      "loss": 1.6941,
+      "step": 78880
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2060963888507484e-05,
+      "loss": 1.5137,
+      "step": 78890
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.20586754542542e-05,
+      "loss": 1.7409,
+      "step": 78900
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.205638702000092e-05,
+      "loss": 1.6512,
+      "step": 78910
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.205409858574763e-05,
+      "loss": 1.5237,
+      "step": 78920
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.205181015149435e-05,
+      "loss": 1.6453,
+      "step": 78930
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.204952171724107e-05,
+      "loss": 1.5692,
+      "step": 78940
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.204723328298778e-05,
+      "loss": 1.7068,
+      "step": 78950
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.20449448487345e-05,
+      "loss": 1.7796,
+      "step": 78960
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.204265641448122e-05,
+      "loss": 1.6762,
+      "step": 78970
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.204036798022793e-05,
+      "loss": 1.8082,
+      "step": 78980
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2038079545974646e-05,
+      "loss": 1.502,
+      "step": 78990
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.203579111172137e-05,
+      "loss": 1.6254,
+      "step": 79000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.203350267746808e-05,
+      "loss": 1.598,
+      "step": 79010
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2031214243214795e-05,
+      "loss": 1.5459,
+      "step": 79020
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.202892580896151e-05,
+      "loss": 1.6756,
+      "step": 79030
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.202663737470822e-05,
+      "loss": 1.7069,
+      "step": 79040
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.202434894045494e-05,
+      "loss": 1.7028,
+      "step": 79050
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.202206050620166e-05,
+      "loss": 1.5658,
+      "step": 79060
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.201977207194837e-05,
+      "loss": 1.6257,
+      "step": 79070
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2017483637695087e-05,
+      "loss": 1.7366,
+      "step": 79080
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.201519520344181e-05,
+      "loss": 1.5959,
+      "step": 79090
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.201290676918852e-05,
+      "loss": 1.6624,
+      "step": 79100
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2010618334935236e-05,
+      "loss": 1.7168,
+      "step": 79110
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.200832990068196e-05,
+      "loss": 1.6676,
+      "step": 79120
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.200604146642867e-05,
+      "loss": 1.7246,
+      "step": 79130
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2003753032175385e-05,
+      "loss": 1.8727,
+      "step": 79140
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2001464597922106e-05,
+      "loss": 1.6817,
+      "step": 79150
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.199917616366882e-05,
+      "loss": 1.6978,
+      "step": 79160
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1996887729415534e-05,
+      "loss": 1.4482,
+      "step": 79170
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1994599295162255e-05,
+      "loss": 1.5454,
+      "step": 79180
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.199231086090897e-05,
+      "loss": 1.6046,
+      "step": 79190
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.199002242665568e-05,
+      "loss": 1.567,
+      "step": 79200
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1987733992402404e-05,
+      "loss": 1.6007,
+      "step": 79210
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.198544555814912e-05,
+      "loss": 1.5486,
+      "step": 79220
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.198315712389583e-05,
+      "loss": 1.5941,
+      "step": 79230
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1980868689642547e-05,
+      "loss": 1.6228,
+      "step": 79240
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.197858025538926e-05,
+      "loss": 1.6974,
+      "step": 79250
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.197629182113598e-05,
+      "loss": 1.5132,
+      "step": 79260
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1974003386882696e-05,
+      "loss": 1.687,
+      "step": 79270
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.197171495262941e-05,
+      "loss": 1.6733,
+      "step": 79280
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.196942651837613e-05,
+      "loss": 1.6182,
+      "step": 79290
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1967138084122845e-05,
+      "loss": 1.6824,
+      "step": 79300
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.196484964986956e-05,
+      "loss": 1.6468,
+      "step": 79310
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.196256121561628e-05,
+      "loss": 1.6849,
+      "step": 79320
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1960272781362994e-05,
+      "loss": 1.6616,
+      "step": 79330
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.195798434710971e-05,
+      "loss": 1.6884,
+      "step": 79340
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.195569591285643e-05,
+      "loss": 1.6277,
+      "step": 79350
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.195340747860314e-05,
+      "loss": 1.7661,
+      "step": 79360
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.195111904434986e-05,
+      "loss": 1.6057,
+      "step": 79370
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.194883061009658e-05,
+      "loss": 1.8014,
+      "step": 79380
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.194654217584329e-05,
+      "loss": 1.6242,
+      "step": 79390
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.194425374159001e-05,
+      "loss": 1.758,
+      "step": 79400
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.194196530733673e-05,
+      "loss": 1.6627,
+      "step": 79410
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.193967687308344e-05,
+      "loss": 1.7027,
+      "step": 79420
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.193738843883015e-05,
+      "loss": 1.6776,
+      "step": 79430
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.193510000457687e-05,
+      "loss": 1.5284,
+      "step": 79440
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1932811570323584e-05,
+      "loss": 1.6886,
+      "step": 79450
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.19305231360703e-05,
+      "loss": 1.6294,
+      "step": 79460
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.192823470181702e-05,
+      "loss": 1.5885,
+      "step": 79470
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.192594626756373e-05,
+      "loss": 1.6837,
+      "step": 79480
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.192365783331045e-05,
+      "loss": 1.4936,
+      "step": 79490
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.192136939905717e-05,
+      "loss": 1.61,
+      "step": 79500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.191908096480388e-05,
+      "loss": 1.7411,
+      "step": 79510
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1916792530550597e-05,
+      "loss": 1.6813,
+      "step": 79520
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.191450409629732e-05,
+      "loss": 1.6126,
+      "step": 79530
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.191221566204403e-05,
+      "loss": 1.9506,
+      "step": 79540
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1909927227790746e-05,
+      "loss": 1.5959,
+      "step": 79550
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.190763879353747e-05,
+      "loss": 1.6521,
+      "step": 79560
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.190535035928418e-05,
+      "loss": 1.5688,
+      "step": 79570
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1903061925030895e-05,
+      "loss": 1.7462,
+      "step": 79580
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1900773490777616e-05,
+      "loss": 1.7627,
+      "step": 79590
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.189848505652433e-05,
+      "loss": 1.3863,
+      "step": 79600
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1896196622271044e-05,
+      "loss": 1.538,
+      "step": 79610
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.189390818801776e-05,
+      "loss": 1.9543,
+      "step": 79620
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.189161975376447e-05,
+      "loss": 1.6517,
+      "step": 79630
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1889331319511186e-05,
+      "loss": 1.5907,
+      "step": 79640
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.188704288525791e-05,
+      "loss": 1.5732,
+      "step": 79650
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.188475445100462e-05,
+      "loss": 1.5969,
+      "step": 79660
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1882466016751336e-05,
+      "loss": 1.4243,
+      "step": 79670
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1880177582498057e-05,
+      "loss": 1.6609,
+      "step": 79680
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.187788914824477e-05,
+      "loss": 1.5267,
+      "step": 79690
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1875600713991485e-05,
+      "loss": 1.7148,
+      "step": 79700
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1873312279738206e-05,
+      "loss": 1.6603,
+      "step": 79710
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.187102384548492e-05,
+      "loss": 1.7487,
+      "step": 79720
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1868735411231634e-05,
+      "loss": 1.6041,
+      "step": 79730
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1866446976978355e-05,
+      "loss": 1.8743,
+      "step": 79740
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.186415854272507e-05,
+      "loss": 1.5824,
+      "step": 79750
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.186187010847178e-05,
+      "loss": 1.5402,
+      "step": 79760
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1859581674218504e-05,
+      "loss": 1.6107,
+      "step": 79770
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.185729323996522e-05,
+      "loss": 1.6749,
+      "step": 79780
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.185500480571193e-05,
+      "loss": 1.4565,
+      "step": 79790
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.185271637145865e-05,
+      "loss": 1.7672,
+      "step": 79800
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.185042793720537e-05,
+      "loss": 1.6241,
+      "step": 79810
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.184813950295208e-05,
+      "loss": 1.6916,
+      "step": 79820
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1845851068698796e-05,
+      "loss": 1.7266,
+      "step": 79830
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.184356263444551e-05,
+      "loss": 1.5944,
+      "step": 79840
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.184127420019223e-05,
+      "loss": 1.5588,
+      "step": 79850
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1838985765938945e-05,
+      "loss": 1.591,
+      "step": 79860
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.183669733168566e-05,
+      "loss": 1.815,
+      "step": 79870
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.183440889743238e-05,
+      "loss": 1.6056,
+      "step": 79880
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1832120463179094e-05,
+      "loss": 1.5727,
+      "step": 79890
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.182983202892581e-05,
+      "loss": 1.7729,
+      "step": 79900
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.182754359467253e-05,
+      "loss": 1.5837,
+      "step": 79910
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.182525516041924e-05,
+      "loss": 1.6584,
+      "step": 79920
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.182296672616596e-05,
+      "loss": 1.7245,
+      "step": 79930
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.182067829191268e-05,
+      "loss": 1.579,
+      "step": 79940
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.181838985765939e-05,
+      "loss": 1.7137,
+      "step": 79950
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1816101423406107e-05,
+      "loss": 1.477,
+      "step": 79960
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.181381298915283e-05,
+      "loss": 1.8135,
+      "step": 79970
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.181152455489954e-05,
+      "loss": 1.7322,
+      "step": 79980
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1809236120646256e-05,
+      "loss": 1.8194,
+      "step": 79990
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1806947686392977e-05,
+      "loss": 1.6555,
+      "step": 80000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.180465925213969e-05,
+      "loss": 1.6452,
+      "step": 80010
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.18023708178864e-05,
+      "loss": 1.6037,
+      "step": 80020
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.180008238363312e-05,
+      "loss": 1.6343,
+      "step": 80030
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.179779394937983e-05,
+      "loss": 1.7176,
+      "step": 80040
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.179550551512655e-05,
+      "loss": 1.6261,
+      "step": 80050
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.179321708087327e-05,
+      "loss": 1.5632,
+      "step": 80060
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.179092864661998e-05,
+      "loss": 1.6769,
+      "step": 80070
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1788640212366696e-05,
+      "loss": 1.622,
+      "step": 80080
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.178635177811342e-05,
+      "loss": 1.658,
+      "step": 80090
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.178406334386013e-05,
+      "loss": 1.5603,
+      "step": 80100
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1781774909606846e-05,
+      "loss": 1.6809,
+      "step": 80110
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1779486475353567e-05,
+      "loss": 1.701,
+      "step": 80120
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.177719804110028e-05,
+      "loss": 1.5526,
+      "step": 80130
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1774909606846995e-05,
+      "loss": 1.7751,
+      "step": 80140
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1772621172593716e-05,
+      "loss": 1.7999,
+      "step": 80150
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.177033273834043e-05,
+      "loss": 1.7246,
+      "step": 80160
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1768044304087144e-05,
+      "loss": 1.612,
+      "step": 80170
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1765755869833865e-05,
+      "loss": 1.4525,
+      "step": 80180
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.176346743558058e-05,
+      "loss": 1.6197,
+      "step": 80190
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.176117900132729e-05,
+      "loss": 1.5733,
+      "step": 80200
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1758890567074014e-05,
+      "loss": 1.7781,
+      "step": 80210
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.175660213282072e-05,
+      "loss": 1.543,
+      "step": 80220
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.175431369856744e-05,
+      "loss": 1.5559,
+      "step": 80230
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1752025264314156e-05,
+      "loss": 1.5848,
+      "step": 80240
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.174973683006087e-05,
+      "loss": 1.6863,
+      "step": 80250
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.174744839580759e-05,
+      "loss": 1.8502,
+      "step": 80260
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1745159961554306e-05,
+      "loss": 1.8257,
+      "step": 80270
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.174287152730102e-05,
+      "loss": 1.7235,
+      "step": 80280
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.174058309304774e-05,
+      "loss": 1.5162,
+      "step": 80290
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1738294658794455e-05,
+      "loss": 1.7164,
+      "step": 80300
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.173600622454117e-05,
+      "loss": 1.8426,
+      "step": 80310
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.173371779028789e-05,
+      "loss": 1.7978,
+      "step": 80320
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1731429356034604e-05,
+      "loss": 1.5888,
+      "step": 80330
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.172914092178132e-05,
+      "loss": 1.5468,
+      "step": 80340
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.172685248752804e-05,
+      "loss": 1.5859,
+      "step": 80350
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.172456405327475e-05,
+      "loss": 1.6549,
+      "step": 80360
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.172227561902147e-05,
+      "loss": 1.6692,
+      "step": 80370
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.171998718476819e-05,
+      "loss": 1.6627,
+      "step": 80380
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.17176987505149e-05,
+      "loss": 1.5038,
+      "step": 80390
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1715410316261616e-05,
+      "loss": 1.5498,
+      "step": 80400
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.171312188200833e-05,
+      "loss": 1.5674,
+      "step": 80410
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1710833447755045e-05,
+      "loss": 1.7757,
+      "step": 80420
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.170854501350176e-05,
+      "loss": 1.7192,
+      "step": 80430
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.170625657924848e-05,
+      "loss": 1.7983,
+      "step": 80440
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1703968144995194e-05,
+      "loss": 1.8378,
+      "step": 80450
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.170167971074191e-05,
+      "loss": 1.6136,
+      "step": 80460
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.169939127648863e-05,
+      "loss": 1.7424,
+      "step": 80470
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.169710284223534e-05,
+      "loss": 1.4895,
+      "step": 80480
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.169481440798206e-05,
+      "loss": 1.7619,
+      "step": 80490
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.169252597372878e-05,
+      "loss": 1.6696,
+      "step": 80500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.169023753947549e-05,
+      "loss": 1.5832,
+      "step": 80510
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1687949105222206e-05,
+      "loss": 1.6422,
+      "step": 80520
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.168566067096893e-05,
+      "loss": 1.5817,
+      "step": 80530
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.168337223671564e-05,
+      "loss": 1.5463,
+      "step": 80540
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1681083802462356e-05,
+      "loss": 1.7336,
+      "step": 80550
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1678795368209077e-05,
+      "loss": 1.6118,
+      "step": 80560
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.167650693395579e-05,
+      "loss": 1.7218,
+      "step": 80570
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1674218499702505e-05,
+      "loss": 1.6078,
+      "step": 80580
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1671930065449226e-05,
+      "loss": 1.8719,
+      "step": 80590
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.166964163119594e-05,
+      "loss": 1.6154,
+      "step": 80600
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1667353196942654e-05,
+      "loss": 1.767,
+      "step": 80610
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.166506476268937e-05,
+      "loss": 1.6806,
+      "step": 80620
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.166277632843608e-05,
+      "loss": 1.5629,
+      "step": 80630
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.16604878941828e-05,
+      "loss": 1.6004,
+      "step": 80640
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.165819945992952e-05,
+      "loss": 1.59,
+      "step": 80650
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.165591102567623e-05,
+      "loss": 1.5849,
+      "step": 80660
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.165362259142295e-05,
+      "loss": 1.6363,
+      "step": 80670
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1651334157169666e-05,
+      "loss": 1.5879,
+      "step": 80680
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.164904572291638e-05,
+      "loss": 1.6486,
+      "step": 80690
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1646757288663095e-05,
+      "loss": 1.7051,
+      "step": 80700
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1644468854409816e-05,
+      "loss": 1.739,
+      "step": 80710
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.164218042015653e-05,
+      "loss": 1.6861,
+      "step": 80720
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1639891985903244e-05,
+      "loss": 1.6649,
+      "step": 80730
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1637603551649965e-05,
+      "loss": 1.392,
+      "step": 80740
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.163531511739668e-05,
+      "loss": 1.8005,
+      "step": 80750
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.163302668314339e-05,
+      "loss": 1.6534,
+      "step": 80760
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1630738248890114e-05,
+      "loss": 1.4926,
+      "step": 80770
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.162844981463683e-05,
+      "loss": 1.5988,
+      "step": 80780
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.162616138038354e-05,
+      "loss": 1.5824,
+      "step": 80790
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.162387294613026e-05,
+      "loss": 1.4357,
+      "step": 80800
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.162158451187697e-05,
+      "loss": 1.6511,
+      "step": 80810
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.161929607762369e-05,
+      "loss": 1.5261,
+      "step": 80820
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1617007643370406e-05,
+      "loss": 1.8589,
+      "step": 80830
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.161471920911712e-05,
+      "loss": 1.5695,
+      "step": 80840
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.161243077486384e-05,
+      "loss": 1.6224,
+      "step": 80850
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1610142340610555e-05,
+      "loss": 1.5713,
+      "step": 80860
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.160785390635727e-05,
+      "loss": 1.4761,
+      "step": 80870
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.160556547210399e-05,
+      "loss": 1.6306,
+      "step": 80880
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1603277037850704e-05,
+      "loss": 1.6891,
+      "step": 80890
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.160098860359742e-05,
+      "loss": 1.4681,
+      "step": 80900
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.159870016934414e-05,
+      "loss": 1.678,
+      "step": 80910
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.159641173509085e-05,
+      "loss": 1.5292,
+      "step": 80920
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.159412330083757e-05,
+      "loss": 1.466,
+      "step": 80930
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.159183486658429e-05,
+      "loss": 1.7025,
+      "step": 80940
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1589546432331e-05,
+      "loss": 1.6771,
+      "step": 80950
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1587257998077716e-05,
+      "loss": 1.7219,
+      "step": 80960
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.158496956382444e-05,
+      "loss": 1.5414,
+      "step": 80970
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.158268112957115e-05,
+      "loss": 1.9016,
+      "step": 80980
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1580392695317866e-05,
+      "loss": 1.5156,
+      "step": 80990
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1578104261064586e-05,
+      "loss": 1.7427,
+      "step": 81000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1575815826811294e-05,
+      "loss": 1.7283,
+      "step": 81010
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.157352739255801e-05,
+      "loss": 1.7029,
+      "step": 81020
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.157123895830473e-05,
+      "loss": 1.4823,
+      "step": 81030
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.156895052405144e-05,
+      "loss": 1.7263,
+      "step": 81040
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.156666208979816e-05,
+      "loss": 1.8082,
+      "step": 81050
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.156437365554488e-05,
+      "loss": 1.4278,
+      "step": 81060
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.156208522129159e-05,
+      "loss": 1.6411,
+      "step": 81070
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1559796787038306e-05,
+      "loss": 1.5808,
+      "step": 81080
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.155750835278503e-05,
+      "loss": 1.689,
+      "step": 81090
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.155521991853174e-05,
+      "loss": 1.5641,
+      "step": 81100
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1552931484278455e-05,
+      "loss": 1.7138,
+      "step": 81110
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1550643050025176e-05,
+      "loss": 1.5405,
+      "step": 81120
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.154835461577189e-05,
+      "loss": 1.6927,
+      "step": 81130
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1546066181518605e-05,
+      "loss": 1.7158,
+      "step": 81140
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1543777747265326e-05,
+      "loss": 1.8111,
+      "step": 81150
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.154148931301204e-05,
+      "loss": 1.6295,
+      "step": 81160
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1539200878758754e-05,
+      "loss": 1.6739,
+      "step": 81170
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1536912444505475e-05,
+      "loss": 1.4465,
+      "step": 81180
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.153462401025219e-05,
+      "loss": 1.8063,
+      "step": 81190
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.15323355759989e-05,
+      "loss": 1.6218,
+      "step": 81200
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.153004714174562e-05,
+      "loss": 1.5621,
+      "step": 81210
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.152775870749233e-05,
+      "loss": 1.5895,
+      "step": 81220
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.152547027323905e-05,
+      "loss": 1.7987,
+      "step": 81230
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1523181838985766e-05,
+      "loss": 1.4608,
+      "step": 81240
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.152089340473248e-05,
+      "loss": 1.681,
+      "step": 81250
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.15186049704792e-05,
+      "loss": 1.6384,
+      "step": 81260
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1516316536225916e-05,
+      "loss": 1.4755,
+      "step": 81270
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.151402810197263e-05,
+      "loss": 1.5936,
+      "step": 81280
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.151173966771935e-05,
+      "loss": 1.5783,
+      "step": 81290
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1509451233466065e-05,
+      "loss": 1.7393,
+      "step": 81300
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.150716279921278e-05,
+      "loss": 1.693,
+      "step": 81310
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.15048743649595e-05,
+      "loss": 1.7479,
+      "step": 81320
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1502585930706214e-05,
+      "loss": 1.4457,
+      "step": 81330
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.150029749645293e-05,
+      "loss": 1.584,
+      "step": 81340
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.149800906219965e-05,
+      "loss": 1.6935,
+      "step": 81350
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.149572062794636e-05,
+      "loss": 1.5169,
+      "step": 81360
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.149343219369308e-05,
+      "loss": 1.6163,
+      "step": 81370
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.14911437594398e-05,
+      "loss": 1.7047,
+      "step": 81380
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.148885532518651e-05,
+      "loss": 1.5909,
+      "step": 81390
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.148656689093322e-05,
+      "loss": 1.5819,
+      "step": 81400
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.148427845667994e-05,
+      "loss": 1.6393,
+      "step": 81410
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1481990022426655e-05,
+      "loss": 1.5301,
+      "step": 81420
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.147970158817337e-05,
+      "loss": 1.6759,
+      "step": 81430
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.147741315392009e-05,
+      "loss": 1.6089,
+      "step": 81440
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1475124719666804e-05,
+      "loss": 1.5644,
+      "step": 81450
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.147283628541352e-05,
+      "loss": 1.6824,
+      "step": 81460
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.147054785116024e-05,
+      "loss": 1.7251,
+      "step": 81470
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.146825941690695e-05,
+      "loss": 1.4724,
+      "step": 81480
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.146597098265367e-05,
+      "loss": 1.6416,
+      "step": 81490
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.146368254840039e-05,
+      "loss": 1.549,
+      "step": 81500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.14613941141471e-05,
+      "loss": 1.7365,
+      "step": 81510
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1459105679893816e-05,
+      "loss": 1.7722,
+      "step": 81520
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.145681724564054e-05,
+      "loss": 1.5734,
+      "step": 81530
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.145452881138725e-05,
+      "loss": 1.6814,
+      "step": 81540
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1452240377133965e-05,
+      "loss": 1.7359,
+      "step": 81550
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1449951942880686e-05,
+      "loss": 1.601,
+      "step": 81560
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.14476635086274e-05,
+      "loss": 1.4886,
+      "step": 81570
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1445375074374115e-05,
+      "loss": 1.731,
+      "step": 81580
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1443086640120836e-05,
+      "loss": 1.7326,
+      "step": 81590
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.144079820586754e-05,
+      "loss": 1.7464,
+      "step": 81600
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1438509771614264e-05,
+      "loss": 1.5897,
+      "step": 81610
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.143622133736098e-05,
+      "loss": 1.5521,
+      "step": 81620
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.143393290310769e-05,
+      "loss": 1.4957,
+      "step": 81630
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.143164446885441e-05,
+      "loss": 1.5359,
+      "step": 81640
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.142935603460113e-05,
+      "loss": 1.7563,
+      "step": 81650
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.142706760034784e-05,
+      "loss": 1.5595,
+      "step": 81660
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.142477916609456e-05,
+      "loss": 1.7068,
+      "step": 81670
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1422490731841276e-05,
+      "loss": 1.7296,
+      "step": 81680
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.142020229758799e-05,
+      "loss": 1.5678,
+      "step": 81690
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.141791386333471e-05,
+      "loss": 1.5711,
+      "step": 81700
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1415625429081425e-05,
+      "loss": 1.9688,
+      "step": 81710
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.141333699482814e-05,
+      "loss": 1.5943,
+      "step": 81720
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.141104856057486e-05,
+      "loss": 1.7062,
+      "step": 81730
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1408760126321575e-05,
+      "loss": 1.5979,
+      "step": 81740
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.140647169206829e-05,
+      "loss": 1.6445,
+      "step": 81750
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.140418325781501e-05,
+      "loss": 1.5907,
+      "step": 81760
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1401894823561724e-05,
+      "loss": 1.7494,
+      "step": 81770
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.139960638930844e-05,
+      "loss": 1.6648,
+      "step": 81780
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.139731795505516e-05,
+      "loss": 1.563,
+      "step": 81790
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1395029520801866e-05,
+      "loss": 1.6204,
+      "step": 81800
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.139274108654858e-05,
+      "loss": 1.691,
+      "step": 81810
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.13904526522953e-05,
+      "loss": 1.7794,
+      "step": 81820
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1388164218042015e-05,
+      "loss": 1.7596,
+      "step": 81830
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.138587578378873e-05,
+      "loss": 1.5341,
+      "step": 81840
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.138358734953545e-05,
+      "loss": 1.7797,
+      "step": 81850
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1381298915282165e-05,
+      "loss": 1.5848,
+      "step": 81860
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.137901048102888e-05,
+      "loss": 1.6393,
+      "step": 81870
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.13767220467756e-05,
+      "loss": 1.7898,
+      "step": 81880
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1374433612522314e-05,
+      "loss": 1.6338,
+      "step": 81890
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.137214517826903e-05,
+      "loss": 1.7461,
+      "step": 81900
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.136985674401575e-05,
+      "loss": 1.4512,
+      "step": 81910
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.136756830976246e-05,
+      "loss": 1.6375,
+      "step": 81920
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.136527987550918e-05,
+      "loss": 1.7403,
+      "step": 81930
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.13629914412559e-05,
+      "loss": 1.637,
+      "step": 81940
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.136070300700261e-05,
+      "loss": 1.7613,
+      "step": 81950
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1358414572749326e-05,
+      "loss": 1.6351,
+      "step": 81960
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.135612613849605e-05,
+      "loss": 1.6075,
+      "step": 81970
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.135383770424276e-05,
+      "loss": 1.7476,
+      "step": 81980
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.135154926998947e-05,
+      "loss": 1.6101,
+      "step": 81990
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.134926083573619e-05,
+      "loss": 1.5954,
+      "step": 82000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1346972401482904e-05,
+      "loss": 1.5261,
+      "step": 82010
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.134468396722962e-05,
+      "loss": 1.4762,
+      "step": 82020
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.134239553297634e-05,
+      "loss": 1.4537,
+      "step": 82030
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.134010709872305e-05,
+      "loss": 1.5286,
+      "step": 82040
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.133781866446977e-05,
+      "loss": 1.5572,
+      "step": 82050
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.133553023021649e-05,
+      "loss": 1.6852,
+      "step": 82060
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.13332417959632e-05,
+      "loss": 1.5971,
+      "step": 82070
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1330953361709916e-05,
+      "loss": 1.6924,
+      "step": 82080
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.132866492745664e-05,
+      "loss": 1.5707,
+      "step": 82090
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.132637649320335e-05,
+      "loss": 1.5937,
+      "step": 82100
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1324088058950065e-05,
+      "loss": 1.5356,
+      "step": 82110
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1321799624696786e-05,
+      "loss": 1.5009,
+      "step": 82120
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.13195111904435e-05,
+      "loss": 1.7457,
+      "step": 82130
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1317222756190215e-05,
+      "loss": 1.7035,
+      "step": 82140
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1314934321936935e-05,
+      "loss": 1.5706,
+      "step": 82150
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.131264588768365e-05,
+      "loss": 1.6343,
+      "step": 82160
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1310357453430364e-05,
+      "loss": 1.4776,
+      "step": 82170
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1308069019177085e-05,
+      "loss": 1.7204,
+      "step": 82180
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.130578058492379e-05,
+      "loss": 1.5522,
+      "step": 82190
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.130349215067051e-05,
+      "loss": 1.5533,
+      "step": 82200
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.130120371641723e-05,
+      "loss": 1.6042,
+      "step": 82210
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.129891528216394e-05,
+      "loss": 1.435,
+      "step": 82220
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.129662684791066e-05,
+      "loss": 1.61,
+      "step": 82230
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1294338413657376e-05,
+      "loss": 1.7631,
+      "step": 82240
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.129204997940409e-05,
+      "loss": 1.5579,
+      "step": 82250
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.128976154515081e-05,
+      "loss": 1.8967,
+      "step": 82260
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1287473110897525e-05,
+      "loss": 1.5835,
+      "step": 82270
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.128518467664424e-05,
+      "loss": 1.6529,
+      "step": 82280
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.128289624239096e-05,
+      "loss": 1.7418,
+      "step": 82290
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1280607808137675e-05,
+      "loss": 1.6163,
+      "step": 82300
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.127831937388439e-05,
+      "loss": 1.575,
+      "step": 82310
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.127603093963111e-05,
+      "loss": 1.4823,
+      "step": 82320
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1273742505377824e-05,
+      "loss": 1.6496,
+      "step": 82330
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.127145407112454e-05,
+      "loss": 1.6012,
+      "step": 82340
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.126916563687126e-05,
+      "loss": 1.6729,
+      "step": 82350
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.126687720261797e-05,
+      "loss": 1.6143,
+      "step": 82360
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.126458876836469e-05,
+      "loss": 1.5555,
+      "step": 82370
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.126230033411141e-05,
+      "loss": 1.576,
+      "step": 82380
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1260011899858115e-05,
+      "loss": 1.5637,
+      "step": 82390
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.125772346560483e-05,
+      "loss": 1.5265,
+      "step": 82400
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.125543503135155e-05,
+      "loss": 1.5636,
+      "step": 82410
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1253146597098264e-05,
+      "loss": 1.7027,
+      "step": 82420
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.125085816284498e-05,
+      "loss": 1.7364,
+      "step": 82430
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.12485697285917e-05,
+      "loss": 1.5814,
+      "step": 82440
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1246281294338414e-05,
+      "loss": 1.348,
+      "step": 82450
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.124399286008513e-05,
+      "loss": 1.4539,
+      "step": 82460
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.124170442583185e-05,
+      "loss": 1.6529,
+      "step": 82470
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.123941599157856e-05,
+      "loss": 1.5324,
+      "step": 82480
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.123712755732528e-05,
+      "loss": 1.6742,
+      "step": 82490
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1234839123072e-05,
+      "loss": 1.6398,
+      "step": 82500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.123255068881871e-05,
+      "loss": 1.6706,
+      "step": 82510
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1230262254565426e-05,
+      "loss": 1.6767,
+      "step": 82520
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.122797382031215e-05,
+      "loss": 1.7095,
+      "step": 82530
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.122568538605886e-05,
+      "loss": 1.4932,
+      "step": 82540
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1223396951805575e-05,
+      "loss": 1.5881,
+      "step": 82550
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1221108517552296e-05,
+      "loss": 1.7692,
+      "step": 82560
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.121882008329901e-05,
+      "loss": 1.5277,
+      "step": 82570
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1216531649045724e-05,
+      "loss": 1.5025,
+      "step": 82580
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.121424321479244e-05,
+      "loss": 1.4976,
+      "step": 82590
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.121195478053915e-05,
+      "loss": 1.7278,
+      "step": 82600
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1209666346285874e-05,
+      "loss": 1.6243,
+      "step": 82610
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.120737791203259e-05,
+      "loss": 1.5031,
+      "step": 82620
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.12050894777793e-05,
+      "loss": 1.6462,
+      "step": 82630
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.120280104352602e-05,
+      "loss": 1.6389,
+      "step": 82640
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.120051260927274e-05,
+      "loss": 1.7302,
+      "step": 82650
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.119822417501945e-05,
+      "loss": 1.4613,
+      "step": 82660
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.119593574076617e-05,
+      "loss": 1.4925,
+      "step": 82670
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1193647306512886e-05,
+      "loss": 1.5663,
+      "step": 82680
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.11913588722596e-05,
+      "loss": 1.5533,
+      "step": 82690
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.118907043800632e-05,
+      "loss": 1.6288,
+      "step": 82700
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1186782003753035e-05,
+      "loss": 1.606,
+      "step": 82710
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.118449356949975e-05,
+      "loss": 1.4834,
+      "step": 82720
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.118220513524647e-05,
+      "loss": 1.4742,
+      "step": 82730
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1179916700993185e-05,
+      "loss": 1.5828,
+      "step": 82740
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.11776282667399e-05,
+      "loss": 1.5192,
+      "step": 82750
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.117533983248662e-05,
+      "loss": 1.5033,
+      "step": 82760
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1173051398233334e-05,
+      "loss": 1.8208,
+      "step": 82770
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.117076296398004e-05,
+      "loss": 1.4756,
+      "step": 82780
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.116847452972676e-05,
+      "loss": 1.6055,
+      "step": 82790
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1166186095473476e-05,
+      "loss": 1.502,
+      "step": 82800
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.116389766122019e-05,
+      "loss": 1.5371,
+      "step": 82810
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.116160922696691e-05,
+      "loss": 1.5651,
+      "step": 82820
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1159320792713625e-05,
+      "loss": 1.6732,
+      "step": 82830
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.115703235846034e-05,
+      "loss": 1.5495,
+      "step": 82840
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.115474392420706e-05,
+      "loss": 1.4596,
+      "step": 82850
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1152455489953774e-05,
+      "loss": 1.8159,
+      "step": 82860
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.115016705570049e-05,
+      "loss": 1.4239,
+      "step": 82870
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.114787862144721e-05,
+      "loss": 1.7639,
+      "step": 82880
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1145590187193924e-05,
+      "loss": 1.4666,
+      "step": 82890
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.114330175294064e-05,
+      "loss": 1.4336,
+      "step": 82900
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.114101331868736e-05,
+      "loss": 1.4775,
+      "step": 82910
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.113872488443407e-05,
+      "loss": 1.6717,
+      "step": 82920
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.113643645018079e-05,
+      "loss": 1.4684,
+      "step": 82930
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.113414801592751e-05,
+      "loss": 1.7577,
+      "step": 82940
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.113185958167422e-05,
+      "loss": 1.6017,
+      "step": 82950
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1129571147420936e-05,
+      "loss": 1.7331,
+      "step": 82960
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.112728271316766e-05,
+      "loss": 1.4865,
+      "step": 82970
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1124994278914364e-05,
+      "loss": 1.4221,
+      "step": 82980
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1122705844661085e-05,
+      "loss": 1.7241,
+      "step": 82990
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.11204174104078e-05,
+      "loss": 1.4911,
+      "step": 83000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1118128976154514e-05,
+      "loss": 1.7685,
+      "step": 83010
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1115840541901234e-05,
+      "loss": 1.5216,
+      "step": 83020
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.111355210764795e-05,
+      "loss": 1.6569,
+      "step": 83030
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.111126367339466e-05,
+      "loss": 1.5753,
+      "step": 83040
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.110897523914138e-05,
+      "loss": 1.4629,
+      "step": 83050
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.11066868048881e-05,
+      "loss": 1.7177,
+      "step": 83060
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.110439837063481e-05,
+      "loss": 1.5228,
+      "step": 83070
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1102109936381526e-05,
+      "loss": 1.6546,
+      "step": 83080
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.109982150212825e-05,
+      "loss": 1.7095,
+      "step": 83090
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.109753306787496e-05,
+      "loss": 1.7334,
+      "step": 83100
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1095244633621675e-05,
+      "loss": 1.6563,
+      "step": 83110
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1092956199368396e-05,
+      "loss": 1.5691,
+      "step": 83120
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.109066776511511e-05,
+      "loss": 1.4792,
+      "step": 83130
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1088379330861824e-05,
+      "loss": 1.5484,
+      "step": 83140
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1086090896608545e-05,
+      "loss": 1.7247,
+      "step": 83150
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.108380246235526e-05,
+      "loss": 1.5339,
+      "step": 83160
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1081514028101974e-05,
+      "loss": 1.7184,
+      "step": 83170
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.107922559384869e-05,
+      "loss": 1.517,
+      "step": 83180
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.10769371595954e-05,
+      "loss": 1.7171,
+      "step": 83190
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.107464872534212e-05,
+      "loss": 1.6156,
+      "step": 83200
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.107236029108884e-05,
+      "loss": 1.6078,
+      "step": 83210
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.107007185683555e-05,
+      "loss": 1.5589,
+      "step": 83220
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.106778342258227e-05,
+      "loss": 1.6529,
+      "step": 83230
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1065494988328986e-05,
+      "loss": 1.534,
+      "step": 83240
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.10632065540757e-05,
+      "loss": 1.4449,
+      "step": 83250
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.106091811982242e-05,
+      "loss": 1.5232,
+      "step": 83260
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1058629685569135e-05,
+      "loss": 1.7441,
+      "step": 83270
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.105634125131585e-05,
+      "loss": 1.5882,
+      "step": 83280
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.105405281706257e-05,
+      "loss": 1.659,
+      "step": 83290
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1051764382809284e-05,
+      "loss": 1.6254,
+      "step": 83300
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1049475948556e-05,
+      "loss": 1.6167,
+      "step": 83310
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.104718751430272e-05,
+      "loss": 1.6211,
+      "step": 83320
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.1044899080049434e-05,
+      "loss": 1.6076,
+      "step": 83330
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.104261064579615e-05,
+      "loss": 1.6271,
+      "step": 83340
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.104032221154287e-05,
+      "loss": 1.532,
+      "step": 83350
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.103803377728958e-05,
+      "loss": 1.8276,
+      "step": 83360
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.10357453430363e-05,
+      "loss": 1.5781,
+      "step": 83370
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.103345690878301e-05,
+      "loss": 1.5825,
+      "step": 83380
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.1031168474529725e-05,
+      "loss": 1.4904,
+      "step": 83390
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.102888004027644e-05,
+      "loss": 1.8152,
+      "step": 83400
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.102659160602316e-05,
+      "loss": 1.6764,
+      "step": 83410
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.1024303171769874e-05,
+      "loss": 1.612,
+      "step": 83420
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.102201473751659e-05,
+      "loss": 1.5652,
+      "step": 83430
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.101972630326331e-05,
+      "loss": 1.6965,
+      "step": 83440
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.1017437869010024e-05,
+      "loss": 1.4836,
+      "step": 83450
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.101514943475674e-05,
+      "loss": 1.6865,
+      "step": 83460
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.101286100050346e-05,
+      "loss": 1.5092,
+      "step": 83470
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.101057256625017e-05,
+      "loss": 1.6151,
+      "step": 83480
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.100828413199689e-05,
+      "loss": 1.681,
+      "step": 83490
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.100599569774361e-05,
+      "loss": 1.649,
+      "step": 83500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.100370726349032e-05,
+      "loss": 1.7172,
+      "step": 83510
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.1001418829237036e-05,
+      "loss": 1.5593,
+      "step": 83520
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.099913039498376e-05,
+      "loss": 1.589,
+      "step": 83530
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.099684196073047e-05,
+      "loss": 1.621,
+      "step": 83540
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0994553526477185e-05,
+      "loss": 1.6756,
+      "step": 83550
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0992265092223906e-05,
+      "loss": 1.5866,
+      "step": 83560
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0989976657970613e-05,
+      "loss": 1.6631,
+      "step": 83570
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0987688223717334e-05,
+      "loss": 1.7372,
+      "step": 83580
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.098539978946405e-05,
+      "loss": 1.6724,
+      "step": 83590
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.098311135521076e-05,
+      "loss": 1.6043,
+      "step": 83600
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0980822920957484e-05,
+      "loss": 1.6385,
+      "step": 83610
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.09785344867042e-05,
+      "loss": 1.6577,
+      "step": 83620
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.097624605245091e-05,
+      "loss": 1.6022,
+      "step": 83630
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.097395761819763e-05,
+      "loss": 1.602,
+      "step": 83640
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.097166918394435e-05,
+      "loss": 1.5621,
+      "step": 83650
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.096938074969106e-05,
+      "loss": 1.5461,
+      "step": 83660
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.096709231543778e-05,
+      "loss": 1.3763,
+      "step": 83670
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0964803881184496e-05,
+      "loss": 1.8366,
+      "step": 83680
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.096251544693121e-05,
+      "loss": 1.6344,
+      "step": 83690
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.096022701267793e-05,
+      "loss": 1.4759,
+      "step": 83700
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0957938578424645e-05,
+      "loss": 1.627,
+      "step": 83710
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.095565014417136e-05,
+      "loss": 1.7154,
+      "step": 83720
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.095336170991808e-05,
+      "loss": 1.7742,
+      "step": 83730
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0951073275664794e-05,
+      "loss": 1.5504,
+      "step": 83740
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.094878484141151e-05,
+      "loss": 1.6742,
+      "step": 83750
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.094649640715823e-05,
+      "loss": 1.5718,
+      "step": 83760
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.094420797290494e-05,
+      "loss": 1.6407,
+      "step": 83770
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.094191953865165e-05,
+      "loss": 1.5559,
+      "step": 83780
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.093963110439837e-05,
+      "loss": 1.4874,
+      "step": 83790
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0937342670145086e-05,
+      "loss": 1.6785,
+      "step": 83800
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.09350542358918e-05,
+      "loss": 1.5608,
+      "step": 83810
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.093276580163852e-05,
+      "loss": 1.4864,
+      "step": 83820
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0930477367385235e-05,
+      "loss": 1.549,
+      "step": 83830
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.092818893313195e-05,
+      "loss": 1.6117,
+      "step": 83840
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.092590049887867e-05,
+      "loss": 1.7273,
+      "step": 83850
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0923612064625384e-05,
+      "loss": 1.6636,
+      "step": 83860
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.09213236303721e-05,
+      "loss": 1.7017,
+      "step": 83870
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.091903519611882e-05,
+      "loss": 1.5771,
+      "step": 83880
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0916746761865533e-05,
+      "loss": 1.6924,
+      "step": 83890
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.091445832761225e-05,
+      "loss": 1.7748,
+      "step": 83900
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.091216989335897e-05,
+      "loss": 1.6417,
+      "step": 83910
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.090988145910568e-05,
+      "loss": 1.7568,
+      "step": 83920
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.09075930248524e-05,
+      "loss": 1.6695,
+      "step": 83930
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.090530459059912e-05,
+      "loss": 1.5873,
+      "step": 83940
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.090301615634583e-05,
+      "loss": 1.6036,
+      "step": 83950
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0900727722092546e-05,
+      "loss": 1.6423,
+      "step": 83960
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.089843928783926e-05,
+      "loss": 1.573,
+      "step": 83970
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0896150853585974e-05,
+      "loss": 1.6002,
+      "step": 83980
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0893862419332695e-05,
+      "loss": 1.432,
+      "step": 83990
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.089157398507941e-05,
+      "loss": 1.6062,
+      "step": 84000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0889285550826123e-05,
+      "loss": 1.6239,
+      "step": 84010
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0886997116572844e-05,
+      "loss": 1.6884,
+      "step": 84020
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.088470868231956e-05,
+      "loss": 1.7076,
+      "step": 84030
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.088242024806627e-05,
+      "loss": 1.7639,
+      "step": 84040
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0880131813812994e-05,
+      "loss": 1.6165,
+      "step": 84050
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.087784337955971e-05,
+      "loss": 1.6982,
+      "step": 84060
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.087555494530642e-05,
+      "loss": 1.7319,
+      "step": 84070
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.087326651105314e-05,
+      "loss": 1.5727,
+      "step": 84080
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.087097807679986e-05,
+      "loss": 1.6315,
+      "step": 84090
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.086868964254657e-05,
+      "loss": 1.5255,
+      "step": 84100
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.086640120829329e-05,
+      "loss": 1.6061,
+      "step": 84110
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0864112774040006e-05,
+      "loss": 1.6309,
+      "step": 84120
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.086182433978672e-05,
+      "loss": 1.3869,
+      "step": 84130
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0859535905533434e-05,
+      "loss": 1.5892,
+      "step": 84140
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0857247471280155e-05,
+      "loss": 1.8163,
+      "step": 84150
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.085495903702687e-05,
+      "loss": 1.6678,
+      "step": 84160
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0852670602773583e-05,
+      "loss": 1.4402,
+      "step": 84170
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.08503821685203e-05,
+      "loss": 1.471,
+      "step": 84180
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.084809373426701e-05,
+      "loss": 1.6117,
+      "step": 84190
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.084580530001373e-05,
+      "loss": 1.5127,
+      "step": 84200
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.084351686576045e-05,
+      "loss": 2.0005,
+      "step": 84210
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.084122843150716e-05,
+      "loss": 1.455,
+      "step": 84220
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.083893999725388e-05,
+      "loss": 1.5731,
+      "step": 84230
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0836651563000596e-05,
+      "loss": 1.688,
+      "step": 84240
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.083436312874731e-05,
+      "loss": 1.6978,
+      "step": 84250
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.083207469449403e-05,
+      "loss": 1.4224,
+      "step": 84260
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0829786260240745e-05,
+      "loss": 1.6079,
+      "step": 84270
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.082749782598746e-05,
+      "loss": 1.6621,
+      "step": 84280
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.082520939173418e-05,
+      "loss": 1.5306,
+      "step": 84290
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0822920957480894e-05,
+      "loss": 1.7794,
+      "step": 84300
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.082063252322761e-05,
+      "loss": 1.5952,
+      "step": 84310
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.081834408897433e-05,
+      "loss": 1.6005,
+      "step": 84320
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0816055654721043e-05,
+      "loss": 1.5737,
+      "step": 84330
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.081376722046776e-05,
+      "loss": 1.5151,
+      "step": 84340
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.081147878621448e-05,
+      "loss": 1.4976,
+      "step": 84350
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0809190351961186e-05,
+      "loss": 1.7481,
+      "step": 84360
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.08069019177079e-05,
+      "loss": 1.7055,
+      "step": 84370
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.080461348345462e-05,
+      "loss": 1.7097,
+      "step": 84380
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0802325049201335e-05,
+      "loss": 1.6621,
+      "step": 84390
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.080003661494805e-05,
+      "loss": 1.5151,
+      "step": 84400
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.079774818069477e-05,
+      "loss": 1.6614,
+      "step": 84410
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0795459746441484e-05,
+      "loss": 1.5269,
+      "step": 84420
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.07931713121882e-05,
+      "loss": 1.6609,
+      "step": 84430
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.079088287793492e-05,
+      "loss": 1.6405,
+      "step": 84440
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0788594443681633e-05,
+      "loss": 1.4441,
+      "step": 84450
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.078630600942835e-05,
+      "loss": 1.6238,
+      "step": 84460
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.078401757517507e-05,
+      "loss": 1.5869,
+      "step": 84470
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.078172914092178e-05,
+      "loss": 1.5629,
+      "step": 84480
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.07794407066685e-05,
+      "loss": 1.5791,
+      "step": 84490
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.077715227241522e-05,
+      "loss": 1.4857,
+      "step": 84500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.077486383816193e-05,
+      "loss": 1.6876,
+      "step": 84510
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0772575403908646e-05,
+      "loss": 1.5809,
+      "step": 84520
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.077028696965537e-05,
+      "loss": 1.5322,
+      "step": 84530
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.076799853540208e-05,
+      "loss": 1.6707,
+      "step": 84540
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0765710101148795e-05,
+      "loss": 1.5677,
+      "step": 84550
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.076342166689551e-05,
+      "loss": 1.6818,
+      "step": 84560
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.076113323264222e-05,
+      "loss": 1.5952,
+      "step": 84570
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0758844798388944e-05,
+      "loss": 1.8476,
+      "step": 84580
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.075655636413566e-05,
+      "loss": 1.6762,
+      "step": 84590
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.075426792988237e-05,
+      "loss": 1.5706,
+      "step": 84600
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0751979495629093e-05,
+      "loss": 1.651,
+      "step": 84610
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.074969106137581e-05,
+      "loss": 1.6692,
+      "step": 84620
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.074740262712252e-05,
+      "loss": 1.728,
+      "step": 84630
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.074511419286924e-05,
+      "loss": 1.6344,
+      "step": 84640
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.074282575861596e-05,
+      "loss": 1.5855,
+      "step": 84650
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.074053732436267e-05,
+      "loss": 1.797,
+      "step": 84660
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.073824889010939e-05,
+      "loss": 1.6284,
+      "step": 84670
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0735960455856106e-05,
+      "loss": 1.6258,
+      "step": 84680
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.073367202160282e-05,
+      "loss": 1.6311,
+      "step": 84690
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.073138358734954e-05,
+      "loss": 1.6753,
+      "step": 84700
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0729095153096255e-05,
+      "loss": 1.6434,
+      "step": 84710
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.072680671884297e-05,
+      "loss": 1.6004,
+      "step": 84720
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.072451828458969e-05,
+      "loss": 1.6652,
+      "step": 84730
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0722229850336404e-05,
+      "loss": 1.6171,
+      "step": 84740
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.071994141608312e-05,
+      "loss": 1.4992,
+      "step": 84750
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.071765298182983e-05,
+      "loss": 1.6437,
+      "step": 84760
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.071536454757655e-05,
+      "loss": 1.6233,
+      "step": 84770
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.071307611332326e-05,
+      "loss": 1.552,
+      "step": 84780
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.071078767906998e-05,
+      "loss": 1.7486,
+      "step": 84790
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0708499244816696e-05,
+      "loss": 1.6856,
+      "step": 84800
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.070621081056341e-05,
+      "loss": 1.5624,
+      "step": 84810
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.070392237631013e-05,
+      "loss": 1.5479,
+      "step": 84820
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0701633942056845e-05,
+      "loss": 1.7122,
+      "step": 84830
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.069934550780356e-05,
+      "loss": 1.5116,
+      "step": 84840
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.069705707355028e-05,
+      "loss": 1.6726,
+      "step": 84850
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0694768639296994e-05,
+      "loss": 1.8607,
+      "step": 84860
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.069248020504371e-05,
+      "loss": 1.6634,
+      "step": 84870
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.069019177079043e-05,
+      "loss": 1.6995,
+      "step": 84880
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.068790333653714e-05,
+      "loss": 1.741,
+      "step": 84890
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.068561490228386e-05,
+      "loss": 1.7259,
+      "step": 84900
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.068332646803058e-05,
+      "loss": 1.6651,
+      "step": 84910
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.068103803377729e-05,
+      "loss": 1.5109,
+      "step": 84920
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.067874959952401e-05,
+      "loss": 1.499,
+      "step": 84930
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.067646116527073e-05,
+      "loss": 1.4333,
+      "step": 84940
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.067417273101744e-05,
+      "loss": 1.5071,
+      "step": 84950
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0671884296764156e-05,
+      "loss": 1.7555,
+      "step": 84960
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.066959586251087e-05,
+      "loss": 1.8698,
+      "step": 84970
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0667307428257584e-05,
+      "loss": 1.5477,
+      "step": 84980
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0665018994004305e-05,
+      "loss": 1.4526,
+      "step": 84990
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.066273055975102e-05,
+      "loss": 1.5442,
+      "step": 85000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.066044212549773e-05,
+      "loss": 1.5944,
+      "step": 85010
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0658153691244454e-05,
+      "loss": 1.8074,
+      "step": 85020
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.065586525699117e-05,
+      "loss": 1.4864,
+      "step": 85030
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.065357682273788e-05,
+      "loss": 1.591,
+      "step": 85040
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.06512883884846e-05,
+      "loss": 1.6127,
+      "step": 85050
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.064899995423132e-05,
+      "loss": 1.665,
+      "step": 85060
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.064671151997803e-05,
+      "loss": 1.7187,
+      "step": 85070
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.064442308572475e-05,
+      "loss": 1.5117,
+      "step": 85080
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.064213465147147e-05,
+      "loss": 1.6607,
+      "step": 85090
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.063984621721818e-05,
+      "loss": 1.593,
+      "step": 85100
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.06375577829649e-05,
+      "loss": 1.7184,
+      "step": 85110
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0635269348711616e-05,
+      "loss": 1.5801,
+      "step": 85120
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.063298091445833e-05,
+      "loss": 1.7346,
+      "step": 85130
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.063069248020505e-05,
+      "loss": 1.5783,
+      "step": 85140
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0628404045951765e-05,
+      "loss": 1.6232,
+      "step": 85150
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.062611561169847e-05,
+      "loss": 1.7688,
+      "step": 85160
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.062382717744519e-05,
+      "loss": 1.5577,
+      "step": 85170
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.062153874319191e-05,
+      "loss": 1.5336,
+      "step": 85180
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.061925030893862e-05,
+      "loss": 1.6763,
+      "step": 85190
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.061696187468534e-05,
+      "loss": 1.6364,
+      "step": 85200
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.061467344043206e-05,
+      "loss": 1.6471,
+      "step": 85210
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.061238500617877e-05,
+      "loss": 1.4962,
+      "step": 85220
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.061009657192549e-05,
+      "loss": 1.757,
+      "step": 85230
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0607808137672206e-05,
+      "loss": 1.6505,
+      "step": 85240
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.060551970341892e-05,
+      "loss": 1.6898,
+      "step": 85250
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.060323126916564e-05,
+      "loss": 1.5833,
+      "step": 85260
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0600942834912355e-05,
+      "loss": 1.5637,
+      "step": 85270
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.059865440065907e-05,
+      "loss": 1.6042,
+      "step": 85280
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.059636596640579e-05,
+      "loss": 1.666,
+      "step": 85290
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0594077532152504e-05,
+      "loss": 1.5845,
+      "step": 85300
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.059178909789922e-05,
+      "loss": 1.4397,
+      "step": 85310
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.058950066364594e-05,
+      "loss": 1.6032,
+      "step": 85320
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.058721222939265e-05,
+      "loss": 1.7798,
+      "step": 85330
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.058492379513937e-05,
+      "loss": 1.4187,
+      "step": 85340
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.058263536088608e-05,
+      "loss": 1.593,
+      "step": 85350
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0580346926632796e-05,
+      "loss": 1.5852,
+      "step": 85360
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.057805849237952e-05,
+      "loss": 1.4764,
+      "step": 85370
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.057577005812623e-05,
+      "loss": 1.5485,
+      "step": 85380
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0573481623872945e-05,
+      "loss": 1.6224,
+      "step": 85390
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.057119318961966e-05,
+      "loss": 1.5244,
+      "step": 85400
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.056890475536638e-05,
+      "loss": 1.8151,
+      "step": 85410
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0566616321113094e-05,
+      "loss": 1.5723,
+      "step": 85420
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.056432788685981e-05,
+      "loss": 1.737,
+      "step": 85430
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.056203945260653e-05,
+      "loss": 1.7104,
+      "step": 85440
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.055975101835324e-05,
+      "loss": 1.6359,
+      "step": 85450
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.055746258409996e-05,
+      "loss": 1.5728,
+      "step": 85460
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.055517414984668e-05,
+      "loss": 1.5713,
+      "step": 85470
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.055288571559339e-05,
+      "loss": 1.6029,
+      "step": 85480
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0550597281340107e-05,
+      "loss": 1.506,
+      "step": 85490
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.054830884708683e-05,
+      "loss": 1.5984,
+      "step": 85500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.054602041283354e-05,
+      "loss": 1.7362,
+      "step": 85510
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0543731978580256e-05,
+      "loss": 1.3713,
+      "step": 85520
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.054144354432698e-05,
+      "loss": 1.6454,
+      "step": 85530
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.053915511007369e-05,
+      "loss": 1.6977,
+      "step": 85540
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0536866675820405e-05,
+      "loss": 1.6779,
+      "step": 85550
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.053457824156712e-05,
+      "loss": 1.5937,
+      "step": 85560
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.053228980731383e-05,
+      "loss": 1.663,
+      "step": 85570
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0530001373060554e-05,
+      "loss": 1.8179,
+      "step": 85580
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.052771293880727e-05,
+      "loss": 1.4837,
+      "step": 85590
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.052542450455398e-05,
+      "loss": 1.6373,
+      "step": 85600
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.05231360703007e-05,
+      "loss": 1.5309,
+      "step": 85610
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.052084763604742e-05,
+      "loss": 1.571,
+      "step": 85620
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.051855920179413e-05,
+      "loss": 1.5837,
+      "step": 85630
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.051627076754085e-05,
+      "loss": 1.7315,
+      "step": 85640
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0513982333287567e-05,
+      "loss": 1.7759,
+      "step": 85650
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0511693899034284e-05,
+      "loss": 1.7038,
+      "step": 85660
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0509405464780998e-05,
+      "loss": 1.7181,
+      "step": 85670
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0507117030527716e-05,
+      "loss": 1.5641,
+      "step": 85680
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0504828596274433e-05,
+      "loss": 1.6964,
+      "step": 85690
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0502540162021147e-05,
+      "loss": 1.5317,
+      "step": 85700
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0500251727767865e-05,
+      "loss": 1.663,
+      "step": 85710
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0497963293514582e-05,
+      "loss": 1.9764,
+      "step": 85720
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0495674859261297e-05,
+      "loss": 1.6075,
+      "step": 85730
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0493386425008014e-05,
+      "loss": 1.5536,
+      "step": 85740
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0491097990754725e-05,
+      "loss": 1.5449,
+      "step": 85750
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0488809556501442e-05,
+      "loss": 1.5377,
+      "step": 85760
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0486521122248157e-05,
+      "loss": 1.5975,
+      "step": 85770
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0484232687994874e-05,
+      "loss": 1.5776,
+      "step": 85780
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.048194425374159e-05,
+      "loss": 1.4256,
+      "step": 85790
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0479655819488306e-05,
+      "loss": 1.6049,
+      "step": 85800
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0477367385235023e-05,
+      "loss": 1.4992,
+      "step": 85810
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.047507895098174e-05,
+      "loss": 1.6766,
+      "step": 85820
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0472790516728455e-05,
+      "loss": 1.6841,
+      "step": 85830
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0470502082475172e-05,
+      "loss": 1.7554,
+      "step": 85840
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.046821364822189e-05,
+      "loss": 1.626,
+      "step": 85850
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0465925213968604e-05,
+      "loss": 1.6005,
+      "step": 85860
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.046363677971532e-05,
+      "loss": 1.4839,
+      "step": 85870
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.046134834546204e-05,
+      "loss": 1.5267,
+      "step": 85880
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0459059911208753e-05,
+      "loss": 1.7146,
+      "step": 85890
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.045677147695547e-05,
+      "loss": 1.3962,
+      "step": 85900
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0454483042702185e-05,
+      "loss": 1.667,
+      "step": 85910
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0452194608448902e-05,
+      "loss": 1.598,
+      "step": 85920
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.044990617419562e-05,
+      "loss": 1.6719,
+      "step": 85930
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0447617739942334e-05,
+      "loss": 1.4993,
+      "step": 85940
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0445329305689048e-05,
+      "loss": 1.5419,
+      "step": 85950
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0443040871435762e-05,
+      "loss": 1.7088,
+      "step": 85960
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.044075243718248e-05,
+      "loss": 1.5784,
+      "step": 85970
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0438464002929194e-05,
+      "loss": 1.704,
+      "step": 85980
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.043617556867591e-05,
+      "loss": 1.5229,
+      "step": 85990
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.043388713442263e-05,
+      "loss": 1.4923,
+      "step": 86000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0431598700169343e-05,
+      "loss": 1.6978,
+      "step": 86010
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.042931026591606e-05,
+      "loss": 1.6283,
+      "step": 86020
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0427021831662778e-05,
+      "loss": 1.6685,
+      "step": 86030
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0424733397409492e-05,
+      "loss": 1.4577,
+      "step": 86040
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.042244496315621e-05,
+      "loss": 1.6445,
+      "step": 86050
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0420156528902927e-05,
+      "loss": 1.5081,
+      "step": 86060
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.041786809464964e-05,
+      "loss": 1.6452,
+      "step": 86070
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.041557966039636e-05,
+      "loss": 1.4488,
+      "step": 86080
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0413291226143077e-05,
+      "loss": 1.6632,
+      "step": 86090
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.041100279188979e-05,
+      "loss": 1.7079,
+      "step": 86100
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0408714357636508e-05,
+      "loss": 1.5663,
+      "step": 86110
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0406425923383226e-05,
+      "loss": 1.736,
+      "step": 86120
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.040413748912994e-05,
+      "loss": 1.4512,
+      "step": 86130
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.040184905487665e-05,
+      "loss": 1.4784,
+      "step": 86140
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0399560620623368e-05,
+      "loss": 1.5619,
+      "step": 86150
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0397272186370086e-05,
+      "loss": 1.6542,
+      "step": 86160
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.03949837521168e-05,
+      "loss": 1.679,
+      "step": 86170
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0392695317863517e-05,
+      "loss": 1.7283,
+      "step": 86180
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0390406883610235e-05,
+      "loss": 1.6497,
+      "step": 86190
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.038811844935695e-05,
+      "loss": 1.6923,
+      "step": 86200
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0385830015103666e-05,
+      "loss": 1.5696,
+      "step": 86210
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0383541580850384e-05,
+      "loss": 1.7236,
+      "step": 86220
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0381253146597098e-05,
+      "loss": 1.6175,
+      "step": 86230
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0378964712343816e-05,
+      "loss": 1.4883,
+      "step": 86240
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0376676278090533e-05,
+      "loss": 1.543,
+      "step": 86250
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0374387843837247e-05,
+      "loss": 1.6781,
+      "step": 86260
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0372099409583965e-05,
+      "loss": 1.5844,
+      "step": 86270
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0369810975330682e-05,
+      "loss": 1.6251,
+      "step": 86280
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0367522541077396e-05,
+      "loss": 1.5491,
+      "step": 86290
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0365234106824114e-05,
+      "loss": 1.4639,
+      "step": 86300
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.036294567257083e-05,
+      "loss": 1.3997,
+      "step": 86310
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0360657238317546e-05,
+      "loss": 1.4729,
+      "step": 86320
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0358368804064263e-05,
+      "loss": 1.5635,
+      "step": 86330
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0356080369810974e-05,
+      "loss": 1.5176,
+      "step": 86340
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.035379193555769e-05,
+      "loss": 1.6474,
+      "step": 86350
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0351503501304406e-05,
+      "loss": 1.4921,
+      "step": 86360
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0349215067051123e-05,
+      "loss": 1.5851,
+      "step": 86370
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.034692663279784e-05,
+      "loss": 1.6899,
+      "step": 86380
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0344638198544555e-05,
+      "loss": 1.5119,
+      "step": 86390
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0342349764291272e-05,
+      "loss": 1.5777,
+      "step": 86400
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.034006133003799e-05,
+      "loss": 1.6306,
+      "step": 86410
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0337772895784704e-05,
+      "loss": 1.7365,
+      "step": 86420
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.033548446153142e-05,
+      "loss": 1.5093,
+      "step": 86430
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.033319602727814e-05,
+      "loss": 1.921,
+      "step": 86440
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0330907593024853e-05,
+      "loss": 1.5856,
+      "step": 86450
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.032861915877157e-05,
+      "loss": 1.6389,
+      "step": 86460
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0326330724518288e-05,
+      "loss": 1.6246,
+      "step": 86470
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0324042290265002e-05,
+      "loss": 1.5516,
+      "step": 86480
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.032175385601172e-05,
+      "loss": 1.3934,
+      "step": 86490
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0319465421758437e-05,
+      "loss": 1.4802,
+      "step": 86500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.031717698750515e-05,
+      "loss": 1.7516,
+      "step": 86510
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.031488855325187e-05,
+      "loss": 1.5777,
+      "step": 86520
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0312600118998587e-05,
+      "loss": 1.5848,
+      "step": 86530
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0310311684745297e-05,
+      "loss": 1.7381,
+      "step": 86540
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.030802325049201e-05,
+      "loss": 1.6427,
+      "step": 86550
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.030573481623873e-05,
+      "loss": 1.5542,
+      "step": 86560
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0303446381985446e-05,
+      "loss": 1.5026,
+      "step": 86570
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.030115794773216e-05,
+      "loss": 1.7823,
+      "step": 86580
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0298869513478878e-05,
+      "loss": 1.7878,
+      "step": 86590
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0296581079225596e-05,
+      "loss": 1.558,
+      "step": 86600
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.029429264497231e-05,
+      "loss": 1.5653,
+      "step": 86610
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0292004210719027e-05,
+      "loss": 1.6988,
+      "step": 86620
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0289715776465745e-05,
+      "loss": 1.7033,
+      "step": 86630
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.028742734221246e-05,
+      "loss": 1.6915,
+      "step": 86640
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0285138907959176e-05,
+      "loss": 1.4103,
+      "step": 86650
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0282850473705894e-05,
+      "loss": 1.5359,
+      "step": 86660
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0280562039452608e-05,
+      "loss": 1.488,
+      "step": 86670
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0278273605199326e-05,
+      "loss": 1.6393,
+      "step": 86680
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0275985170946043e-05,
+      "loss": 1.7662,
+      "step": 86690
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0273696736692757e-05,
+      "loss": 1.5899,
+      "step": 86700
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0271408302439475e-05,
+      "loss": 1.6061,
+      "step": 86710
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0269119868186192e-05,
+      "loss": 1.6198,
+      "step": 86720
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0266831433932906e-05,
+      "loss": 1.5571,
+      "step": 86730
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0264542999679617e-05,
+      "loss": 1.5931,
+      "step": 86740
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0262254565426335e-05,
+      "loss": 1.7164,
+      "step": 86750
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0259966131173052e-05,
+      "loss": 1.5721,
+      "step": 86760
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0257677696919766e-05,
+      "loss": 1.6537,
+      "step": 86770
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0255389262666484e-05,
+      "loss": 1.5761,
+      "step": 86780
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.02531008284132e-05,
+      "loss": 1.5712,
+      "step": 86790
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0250812394159916e-05,
+      "loss": 1.5024,
+      "step": 86800
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0248523959906633e-05,
+      "loss": 1.5378,
+      "step": 86810
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.024623552565335e-05,
+      "loss": 1.4378,
+      "step": 86820
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0243947091400065e-05,
+      "loss": 1.7725,
+      "step": 86830
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0241658657146782e-05,
+      "loss": 1.5737,
+      "step": 86840
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.02393702228935e-05,
+      "loss": 1.5655,
+      "step": 86850
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0237081788640214e-05,
+      "loss": 1.717,
+      "step": 86860
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.023479335438693e-05,
+      "loss": 1.5798,
+      "step": 86870
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.023250492013365e-05,
+      "loss": 1.516,
+      "step": 86880
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0230216485880363e-05,
+      "loss": 1.6399,
+      "step": 86890
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.022792805162708e-05,
+      "loss": 1.5596,
+      "step": 86900
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0225639617373798e-05,
+      "loss": 1.4641,
+      "step": 86910
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0223351183120512e-05,
+      "loss": 1.7695,
+      "step": 86920
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0221062748867223e-05,
+      "loss": 1.6226,
+      "step": 86930
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.021877431461394e-05,
+      "loss": 1.7209,
+      "step": 86940
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0216485880360658e-05,
+      "loss": 1.7297,
+      "step": 86950
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0214197446107372e-05,
+      "loss": 1.465,
+      "step": 86960
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.021190901185409e-05,
+      "loss": 1.6571,
+      "step": 86970
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0209620577600807e-05,
+      "loss": 1.5508,
+      "step": 86980
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.020733214334752e-05,
+      "loss": 1.5637,
+      "step": 86990
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.020504370909424e-05,
+      "loss": 1.6452,
+      "step": 87000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0202755274840956e-05,
+      "loss": 1.5943,
+      "step": 87010
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.020046684058767e-05,
+      "loss": 1.8165,
+      "step": 87020
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0198178406334388e-05,
+      "loss": 1.4194,
+      "step": 87030
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0195889972081106e-05,
+      "loss": 1.6807,
+      "step": 87040
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.019360153782782e-05,
+      "loss": 1.6564,
+      "step": 87050
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0191313103574537e-05,
+      "loss": 1.6347,
+      "step": 87060
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.018902466932125e-05,
+      "loss": 1.6582,
+      "step": 87070
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.018673623506797e-05,
+      "loss": 1.5135,
+      "step": 87080
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0184447800814686e-05,
+      "loss": 1.6074,
+      "step": 87090
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.01821593665614e-05,
+      "loss": 1.5956,
+      "step": 87100
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0179870932308118e-05,
+      "loss": 1.9152,
+      "step": 87110
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0177582498054836e-05,
+      "loss": 1.5889,
+      "step": 87120
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0175294063801546e-05,
+      "loss": 1.6569,
+      "step": 87130
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.017300562954826e-05,
+      "loss": 1.5528,
+      "step": 87140
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0170717195294978e-05,
+      "loss": 1.5454,
+      "step": 87150
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0168428761041696e-05,
+      "loss": 1.5983,
+      "step": 87160
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.016614032678841e-05,
+      "loss": 1.645,
+      "step": 87170
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0163851892535127e-05,
+      "loss": 1.5025,
+      "step": 87180
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0161563458281845e-05,
+      "loss": 1.711,
+      "step": 87190
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.015927502402856e-05,
+      "loss": 1.6259,
+      "step": 87200
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0156986589775276e-05,
+      "loss": 1.4857,
+      "step": 87210
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0154698155521994e-05,
+      "loss": 1.4839,
+      "step": 87220
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0152409721268708e-05,
+      "loss": 1.507,
+      "step": 87230
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0150121287015426e-05,
+      "loss": 1.5343,
+      "step": 87240
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0147832852762143e-05,
+      "loss": 1.5668,
+      "step": 87250
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0145544418508857e-05,
+      "loss": 1.6742,
+      "step": 87260
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0143255984255575e-05,
+      "loss": 1.649,
+      "step": 87270
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0140967550002292e-05,
+      "loss": 1.6219,
+      "step": 87280
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0138679115749006e-05,
+      "loss": 1.7857,
+      "step": 87290
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0136390681495724e-05,
+      "loss": 1.6297,
+      "step": 87300
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.013410224724244e-05,
+      "loss": 1.5559,
+      "step": 87310
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0131813812989156e-05,
+      "loss": 1.6515,
+      "step": 87320
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0129525378735866e-05,
+      "loss": 1.5934,
+      "step": 87330
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0127236944482584e-05,
+      "loss": 1.729,
+      "step": 87340
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.01249485102293e-05,
+      "loss": 1.5493,
+      "step": 87350
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0122660075976015e-05,
+      "loss": 1.5888,
+      "step": 87360
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0120371641722733e-05,
+      "loss": 1.7753,
+      "step": 87370
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.011808320746945e-05,
+      "loss": 1.501,
+      "step": 87380
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0115794773216165e-05,
+      "loss": 1.8703,
+      "step": 87390
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0113506338962882e-05,
+      "loss": 1.8427,
+      "step": 87400
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.01112179047096e-05,
+      "loss": 1.6393,
+      "step": 87410
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0108929470456314e-05,
+      "loss": 1.7296,
+      "step": 87420
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.010664103620303e-05,
+      "loss": 1.465,
+      "step": 87430
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.010435260194975e-05,
+      "loss": 1.4827,
+      "step": 87440
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0102064167696463e-05,
+      "loss": 1.492,
+      "step": 87450
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.009977573344318e-05,
+      "loss": 1.5468,
+      "step": 87460
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0097487299189898e-05,
+      "loss": 1.5962,
+      "step": 87470
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0095198864936612e-05,
+      "loss": 1.5442,
+      "step": 87480
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.009291043068333e-05,
+      "loss": 1.4423,
+      "step": 87490
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0090621996430047e-05,
+      "loss": 1.4339,
+      "step": 87500
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.008833356217676e-05,
+      "loss": 1.5483,
+      "step": 87510
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.008604512792348e-05,
+      "loss": 1.5705,
+      "step": 87520
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.008375669367019e-05,
+      "loss": 1.4546,
+      "step": 87530
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0081468259416907e-05,
+      "loss": 1.5455,
+      "step": 87540
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.007917982516362e-05,
+      "loss": 1.6061,
+      "step": 87550
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.007689139091034e-05,
+      "loss": 1.5896,
+      "step": 87560
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0074602956657056e-05,
+      "loss": 1.4837,
+      "step": 87570
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.007231452240377e-05,
+      "loss": 1.5007,
+      "step": 87580
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0070026088150488e-05,
+      "loss": 1.67,
+      "step": 87590
+    },
+    {
+      "epoch": 4.0,
+      "eval_cer": 0.6792978756320028,
+      "eval_em": 0.00728476821192053,
+      "eval_f1": 0.00728476821192053,
+      "eval_loss": 1.4676569700241089,
+      "eval_runtime": 2607.9631,
+      "eval_samples_per_second": 4.053,
+      "eval_steps_per_second": 2.026,
+      "eval_wer": 0.9927152317880795,
+      "step": 87599
+    }
+  ],
+  "max_steps": 218990,
+  "num_train_epochs": 10,
+  "total_flos": 8.265838015310193e+17,
+  "trial_name": null,
+  "trial_params": null
+}