diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14442 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.99943704259711,
+  "eval_steps": 500,
+  "global_step": 11988,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6680567139282736e-08,
+      "loss": 1.578,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.340283569641369e-08,
+      "loss": 1.6286,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6680567139282737e-07,
+      "loss": 1.5795,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5020850708924106e-07,
+      "loss": 1.6039,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3361134278565475e-07,
+      "loss": 1.6443,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.170141784820684e-07,
+      "loss": 1.6004,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.004170141784821e-07,
+      "loss": 1.5867,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.838198498748958e-07,
+      "loss": 1.5424,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.672226855713095e-07,
+      "loss": 1.5549,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.506255212677231e-07,
+      "loss": 1.5572,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.340283569641368e-07,
+      "loss": 1.5759,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.174311926605506e-07,
+      "loss": 1.502,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0008340283569642e-06,
+      "loss": 1.4912,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.084236864053378e-06,
+      "loss": 1.4926,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1676396997497916e-06,
+      "loss": 1.4766,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2510425354462053e-06,
+      "loss": 1.4288,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.334445371142619e-06,
+      "loss": 1.3864,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4178482068390325e-06,
+      "loss": 1.3985,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5012510425354462e-06,
+      "loss": 1.3879,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5846538782318598e-06,
+      "loss": 1.4315,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6680567139282735e-06,
+      "loss": 1.4185,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7514595496246874e-06,
+      "loss": 1.3851,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8348623853211011e-06,
+      "loss": 1.3849,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.918265221017515e-06,
+      "loss": 1.3234,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0016680567139285e-06,
+      "loss": 1.3462,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.085070892410342e-06,
+      "loss": 1.3774,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.168473728106756e-06,
+      "loss": 1.3447,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.2518765638031695e-06,
+      "loss": 1.3171,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.3352793994995832e-06,
+      "loss": 1.3173,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.418682235195997e-06,
+      "loss": 1.3209,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.5020850708924106e-06,
+      "loss": 1.3433,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.5854879065888243e-06,
+      "loss": 1.3505,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.668890742285238e-06,
+      "loss": 1.3261,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.7522935779816517e-06,
+      "loss": 1.2974,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.835696413678065e-06,
+      "loss": 1.3373,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.919099249374479e-06,
+      "loss": 1.2952,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.0025020850708923e-06,
+      "loss": 1.2941,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.0859049207673064e-06,
+      "loss": 1.2932,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.1693077564637197e-06,
+      "loss": 1.2645,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.252710592160134e-06,
+      "loss": 1.3114,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.336113427856547e-06,
+      "loss": 1.2888,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.419516263552961e-06,
+      "loss": 1.2905,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.502919099249375e-06,
+      "loss": 1.3203,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.5863219349457885e-06,
+      "loss": 1.2881,
+      "step": 215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6697247706422022e-06,
+      "loss": 1.3219,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.753127606338616e-06,
+      "loss": 1.3338,
+      "step": 225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.83653044203503e-06,
+      "loss": 1.3147,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.919933277731443e-06,
+      "loss": 1.2808,
+      "step": 235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.003336113427857e-06,
+      "loss": 1.255,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.086738949124271e-06,
+      "loss": 1.2718,
+      "step": 245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.170141784820684e-06,
+      "loss": 1.2933,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.253544620517098e-06,
+      "loss": 1.295,
+      "step": 255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.336947456213512e-06,
+      "loss": 1.2717,
+      "step": 260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.420350291909925e-06,
+      "loss": 1.2838,
+      "step": 265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.503753127606339e-06,
+      "loss": 1.2425,
+      "step": 270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.587155963302753e-06,
+      "loss": 1.2784,
+      "step": 275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6705587989991665e-06,
+      "loss": 1.2492,
+      "step": 280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.75396163469558e-06,
+      "loss": 1.3248,
+      "step": 285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.837364470391994e-06,
+      "loss": 1.2567,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9207673060884075e-06,
+      "loss": 1.2975,
+      "step": 295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.004170141784821e-06,
+      "loss": 1.2844,
+      "step": 300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.087572977481234e-06,
+      "loss": 1.2544,
+      "step": 305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.170975813177649e-06,
+      "loss": 1.2318,
+      "step": 310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.254378648874062e-06,
+      "loss": 1.2705,
+      "step": 315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.337781484570476e-06,
+      "loss": 1.2621,
+      "step": 320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.421184320266889e-06,
+      "loss": 1.283,
+      "step": 325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.504587155963303e-06,
+      "loss": 1.2959,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.587989991659717e-06,
+      "loss": 1.2653,
+      "step": 335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.67139282735613e-06,
+      "loss": 1.2703,
+      "step": 340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.754795663052544e-06,
+      "loss": 1.2955,
+      "step": 345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.838198498748958e-06,
+      "loss": 1.2815,
+      "step": 350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.921601334445372e-06,
+      "loss": 1.2385,
+      "step": 355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.005004170141785e-06,
+      "loss": 1.3024,
+      "step": 360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.088407005838199e-06,
+      "loss": 1.2379,
+      "step": 365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.171809841534613e-06,
+      "loss": 1.2891,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.2552126772310265e-06,
+      "loss": 1.2406,
+      "step": 375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.338615512927439e-06,
+      "loss": 1.291,
+      "step": 380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.422018348623854e-06,
+      "loss": 1.2885,
+      "step": 385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.505421184320268e-06,
+      "loss": 1.3061,
+      "step": 390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.588824020016681e-06,
+      "loss": 1.2573,
+      "step": 395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.672226855713094e-06,
+      "loss": 1.2933,
+      "step": 400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.755629691409509e-06,
+      "loss": 1.2596,
+      "step": 405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.839032527105922e-06,
+      "loss": 1.289,
+      "step": 410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.922435362802335e-06,
+      "loss": 1.2711,
+      "step": 415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.00583819849875e-06,
+      "loss": 1.2436,
+      "step": 420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.089241034195163e-06,
+      "loss": 1.2932,
+      "step": 425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.172643869891577e-06,
+      "loss": 1.3038,
+      "step": 430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.25604670558799e-06,
+      "loss": 1.2597,
+      "step": 435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.3394495412844045e-06,
+      "loss": 1.2626,
+      "step": 440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.422852376980818e-06,
+      "loss": 1.2772,
+      "step": 445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.506255212677232e-06,
+      "loss": 1.3038,
+      "step": 450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.589658048373645e-06,
+      "loss": 1.2797,
+      "step": 455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.67306088407006e-06,
+      "loss": 1.2389,
+      "step": 460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.756463719766472e-06,
+      "loss": 1.252,
+      "step": 465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.839866555462887e-06,
+      "loss": 1.2813,
+      "step": 470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.923269391159301e-06,
+      "loss": 1.254,
+      "step": 475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.006672226855714e-06,
+      "loss": 1.2902,
+      "step": 480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.090075062552127e-06,
+      "loss": 1.2797,
+      "step": 485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.173477898248541e-06,
+      "loss": 1.2853,
+      "step": 490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.256880733944956e-06,
+      "loss": 1.2919,
+      "step": 495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.340283569641369e-06,
+      "loss": 1.262,
+      "step": 500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.423686405337782e-06,
+      "loss": 1.2363,
+      "step": 505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.507089241034196e-06,
+      "loss": 1.2967,
+      "step": 510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.59049207673061e-06,
+      "loss": 1.2669,
+      "step": 515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.673894912427023e-06,
+      "loss": 1.2518,
+      "step": 520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.757297748123436e-06,
+      "loss": 1.2893,
+      "step": 525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.84070058381985e-06,
+      "loss": 1.26,
+      "step": 530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.924103419516265e-06,
+      "loss": 1.3198,
+      "step": 535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.007506255212678e-06,
+      "loss": 1.3074,
+      "step": 540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 1.2815,
+      "step": 545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.174311926605506e-06,
+      "loss": 1.2765,
+      "step": 550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.257714762301918e-06,
+      "loss": 1.3063,
+      "step": 555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.341117597998333e-06,
+      "loss": 1.3181,
+      "step": 560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.424520433694746e-06,
+      "loss": 1.2279,
+      "step": 565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.50792326939116e-06,
+      "loss": 1.2588,
+      "step": 570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.591326105087573e-06,
+      "loss": 1.3062,
+      "step": 575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.674728940783988e-06,
+      "loss": 1.3182,
+      "step": 580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.7581317764804e-06,
+      "loss": 1.2482,
+      "step": 585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.841534612176815e-06,
+      "loss": 1.2463,
+      "step": 590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.924937447873228e-06,
+      "loss": 1.2707,
+      "step": 595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.0008340283569642e-05,
+      "loss": 1.2961,
+      "step": 600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.0091743119266055e-05,
+      "loss": 1.2361,
+      "step": 605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.0175145954962468e-05,
+      "loss": 1.293,
+      "step": 610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.0258548790658884e-05,
+      "loss": 1.2488,
+      "step": 615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.0341951626355297e-05,
+      "loss": 1.2848,
+      "step": 620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.0425354462051712e-05,
+      "loss": 1.2926,
+      "step": 625
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.0508757297748125e-05,
+      "loss": 1.276,
+      "step": 630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.0592160133444537e-05,
+      "loss": 1.3104,
+      "step": 635
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.0675562969140952e-05,
+      "loss": 1.2973,
+      "step": 640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.0758965804837365e-05,
+      "loss": 1.2913,
+      "step": 645
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.0842368640533778e-05,
+      "loss": 1.2913,
+      "step": 650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.0925771476230194e-05,
+      "loss": 1.2994,
+      "step": 655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1009174311926607e-05,
+      "loss": 1.2896,
+      "step": 660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1092577147623021e-05,
+      "loss": 1.257,
+      "step": 665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1175979983319434e-05,
+      "loss": 1.2842,
+      "step": 670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1259382819015847e-05,
+      "loss": 1.2806,
+      "step": 675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.134278565471226e-05,
+      "loss": 1.3073,
+      "step": 680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1426188490408674e-05,
+      "loss": 1.2742,
+      "step": 685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1509591326105089e-05,
+      "loss": 1.2711,
+      "step": 690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1592994161801503e-05,
+      "loss": 1.2741,
+      "step": 695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1676396997497916e-05,
+      "loss": 1.2972,
+      "step": 700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1759799833194329e-05,
+      "loss": 1.2743,
+      "step": 705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1843202668890744e-05,
+      "loss": 1.2384,
+      "step": 710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1926605504587156e-05,
+      "loss": 1.2695,
+      "step": 715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.201000834028357e-05,
+      "loss": 1.2595,
+      "step": 720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.2093411175979984e-05,
+      "loss": 1.2629,
+      "step": 725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.2176814011676398e-05,
+      "loss": 1.2633,
+      "step": 730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.2260216847372813e-05,
+      "loss": 1.3081,
+      "step": 735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2343619683069226e-05,
+      "loss": 1.2676,
+      "step": 740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2427022518765639e-05,
+      "loss": 1.2731,
+      "step": 745
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2510425354462053e-05,
+      "loss": 1.3074,
+      "step": 750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2593828190158466e-05,
+      "loss": 1.2863,
+      "step": 755
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2677231025854879e-05,
+      "loss": 1.2688,
+      "step": 760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2760633861551295e-05,
+      "loss": 1.2657,
+      "step": 765
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2844036697247708e-05,
+      "loss": 1.2646,
+      "step": 770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2927439532944122e-05,
+      "loss": 1.2945,
+      "step": 775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3010842368640535e-05,
+      "loss": 1.2604,
+      "step": 780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3094245204336948e-05,
+      "loss": 1.2907,
+      "step": 785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3177648040033363e-05,
+      "loss": 1.2658,
+      "step": 790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3261050875729775e-05,
+      "loss": 1.2633,
+      "step": 795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3344453711426188e-05,
+      "loss": 1.2772,
+      "step": 800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3427856547122604e-05,
+      "loss": 1.2477,
+      "step": 805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3511259382819017e-05,
+      "loss": 1.2962,
+      "step": 810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.359466221851543e-05,
+      "loss": 1.2864,
+      "step": 815
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3678065054211845e-05,
+      "loss": 1.306,
+      "step": 820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3761467889908258e-05,
+      "loss": 1.2709,
+      "step": 825
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.384487072560467e-05,
+      "loss": 1.3011,
+      "step": 830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3928273561301085e-05,
+      "loss": 1.286,
+      "step": 835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.40116763969975e-05,
+      "loss": 1.2927,
+      "step": 840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.4095079232693914e-05,
+      "loss": 1.3209,
+      "step": 845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.4178482068390327e-05,
+      "loss": 1.3281,
+      "step": 850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.426188490408674e-05,
+      "loss": 1.263,
+      "step": 855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.4345287739783154e-05,
+      "loss": 1.2634,
+      "step": 860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.4428690575479567e-05,
+      "loss": 1.2543,
+      "step": 865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.451209341117598e-05,
+      "loss": 1.3057,
+      "step": 870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.4595496246872394e-05,
+      "loss": 1.2786,
+      "step": 875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.4678899082568809e-05,
+      "loss": 1.2607,
+      "step": 880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.4762301918265223e-05,
+      "loss": 1.2684,
+      "step": 885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.4845704753961636e-05,
+      "loss": 1.3002,
+      "step": 890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.4929107589658049e-05,
+      "loss": 1.2781,
+      "step": 895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5012510425354464e-05,
+      "loss": 1.286,
+      "step": 900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5095913261050877e-05,
+      "loss": 1.2695,
+      "step": 905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.517931609674729e-05,
+      "loss": 1.2862,
+      "step": 910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5262718932443706e-05,
+      "loss": 1.2598,
+      "step": 915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.534612176814012e-05,
+      "loss": 1.259,
+      "step": 920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.542952460383653e-05,
+      "loss": 1.3079,
+      "step": 925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5512927439532944e-05,
+      "loss": 1.2558,
+      "step": 930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.559633027522936e-05,
+      "loss": 1.3157,
+      "step": 935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5679733110925773e-05,
+      "loss": 1.2892,
+      "step": 940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5763135946622186e-05,
+      "loss": 1.2793,
+      "step": 945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5846538782318602e-05,
+      "loss": 1.256,
+      "step": 950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5929941618015015e-05,
+      "loss": 1.273,
+      "step": 955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.6013344453711428e-05,
+      "loss": 1.2675,
+      "step": 960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.609674728940784e-05,
+      "loss": 1.2804,
+      "step": 965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.6180150125104254e-05,
+      "loss": 1.2599,
+      "step": 970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.626355296080067e-05,
+      "loss": 1.2912,
+      "step": 975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6346955796497083e-05,
+      "loss": 1.303,
+      "step": 980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6430358632193495e-05,
+      "loss": 1.2892,
+      "step": 985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6513761467889912e-05,
+      "loss": 1.2741,
+      "step": 990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6597164303586325e-05,
+      "loss": 1.2953,
+      "step": 995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6680567139282737e-05,
+      "loss": 1.2881,
+      "step": 1000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.676396997497915e-05,
+      "loss": 1.2925,
+      "step": 1005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6847372810675563e-05,
+      "loss": 1.2606,
+      "step": 1010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6930775646371976e-05,
+      "loss": 1.247,
+      "step": 1015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7014178482068392e-05,
+      "loss": 1.2693,
+      "step": 1020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7097581317764805e-05,
+      "loss": 1.2882,
+      "step": 1025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.718098415346122e-05,
+      "loss": 1.2447,
+      "step": 1030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7264386989157634e-05,
+      "loss": 1.2877,
+      "step": 1035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7347789824854047e-05,
+      "loss": 1.235,
+      "step": 1040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743119266055046e-05,
+      "loss": 1.2529,
+      "step": 1045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7514595496246873e-05,
+      "loss": 1.2925,
+      "step": 1050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7597998331943285e-05,
+      "loss": 1.2483,
+      "step": 1055
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.76814011676397e-05,
+      "loss": 1.2635,
+      "step": 1060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7764804003336114e-05,
+      "loss": 1.3163,
+      "step": 1065
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.784820683903253e-05,
+      "loss": 1.2778,
+      "step": 1070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7931609674728944e-05,
+      "loss": 1.2481,
+      "step": 1075
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8015012510425356e-05,
+      "loss": 1.267,
+      "step": 1080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.809841534612177e-05,
+      "loss": 1.2677,
+      "step": 1085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 1.2766,
+      "step": 1090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8265221017514595e-05,
+      "loss": 1.233,
+      "step": 1095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.834862385321101e-05,
+      "loss": 1.2755,
+      "step": 1100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8432026688907424e-05,
+      "loss": 1.2431,
+      "step": 1105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8515429524603837e-05,
+      "loss": 1.309,
+      "step": 1110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8598832360300253e-05,
+      "loss": 1.2463,
+      "step": 1115
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8682235195996666e-05,
+      "loss": 1.2716,
+      "step": 1120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.876563803169308e-05,
+      "loss": 1.2906,
+      "step": 1125
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.884904086738949e-05,
+      "loss": 1.2527,
+      "step": 1130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8932443703085904e-05,
+      "loss": 1.2586,
+      "step": 1135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.901584653878232e-05,
+      "loss": 1.3044,
+      "step": 1140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9099249374478733e-05,
+      "loss": 1.2671,
+      "step": 1145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9182652210175146e-05,
+      "loss": 1.305,
+      "step": 1150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9266055045871563e-05,
+      "loss": 1.2718,
+      "step": 1155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9349457881567975e-05,
+      "loss": 1.2714,
+      "step": 1160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9432860717264388e-05,
+      "loss": 1.289,
+      "step": 1165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.95162635529608e-05,
+      "loss": 1.2489,
+      "step": 1170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9599666388657217e-05,
+      "loss": 1.2056,
+      "step": 1175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.968306922435363e-05,
+      "loss": 1.2593,
+      "step": 1180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9766472060050043e-05,
+      "loss": 1.2361,
+      "step": 1185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9849874895746456e-05,
+      "loss": 1.2835,
+      "step": 1190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9933277731442872e-05,
+      "loss": 1.2852,
+      "step": 1195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9999999576057108e-05,
+      "loss": 1.266,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.999998473805958e-05,
+      "loss": 1.2503,
+      "step": 1205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9999948702953286e-05,
+      "loss": 1.2862,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9999891470814604e-05,
+      "loss": 1.257,
+      "step": 1215
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9999813041764854e-05,
+      "loss": 1.2569,
+      "step": 1220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9999713415970277e-05,
+      "loss": 1.2703,
+      "step": 1225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.999959259364206e-05,
+      "loss": 1.2321,
+      "step": 1230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9999450575036306e-05,
+      "loss": 1.2441,
+      "step": 1235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.999928736045406e-05,
+      "loss": 1.2817,
+      "step": 1240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9999102950241278e-05,
+      "loss": 1.231,
+      "step": 1245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.999889734478887e-05,
+      "loss": 1.2513,
+      "step": 1250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9998670544532654e-05,
+      "loss": 1.237,
+      "step": 1255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.999842254995338e-05,
+      "loss": 1.2669,
+      "step": 1260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.999815336157673e-05,
+      "loss": 1.2249,
+      "step": 1265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9997862979973308e-05,
+      "loss": 1.2739,
+      "step": 1270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9997551405758634e-05,
+      "loss": 1.228,
+      "step": 1275
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.999721863959316e-05,
+      "loss": 1.2604,
+      "step": 1280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9996864682182253e-05,
+      "loss": 1.2771,
+      "step": 1285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9996489534276207e-05,
+      "loss": 1.2686,
+      "step": 1290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.999609319667022e-05,
+      "loss": 1.2523,
+      "step": 1295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.999567567020442e-05,
+      "loss": 1.2274,
+      "step": 1300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9995236955763842e-05,
+      "loss": 1.2692,
+      "step": 1305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9994777054278435e-05,
+      "loss": 1.283,
+      "step": 1310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9994295966723062e-05,
+      "loss": 1.2546,
+      "step": 1315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.999379369411749e-05,
+      "loss": 1.2631,
+      "step": 1320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9993270237526384e-05,
+      "loss": 1.2633,
+      "step": 1325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9992725598059333e-05,
+      "loss": 1.3234,
+      "step": 1330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9992159776870815e-05,
+      "loss": 1.2766,
+      "step": 1335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.999157277516021e-05,
+      "loss": 1.2636,
+      "step": 1340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.999096459417179e-05,
+      "loss": 1.2942,
+      "step": 1345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9990335235194727e-05,
+      "loss": 1.2558,
+      "step": 1350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.998968469956308e-05,
+      "loss": 1.2666,
+      "step": 1355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.99890129886558e-05,
+      "loss": 1.308,
+      "step": 1360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9988320103896727e-05,
+      "loss": 1.2606,
+      "step": 1365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.998760604675457e-05,
+      "loss": 1.2628,
+      "step": 1370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9986870818742932e-05,
+      "loss": 1.2668,
+      "step": 1375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9986114421420284e-05,
+      "loss": 1.2605,
+      "step": 1380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.998533685638997e-05,
+      "loss": 1.2889,
+      "step": 1385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.998453812530021e-05,
+      "loss": 1.2386,
+      "step": 1390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9983718229844083e-05,
+      "loss": 1.2597,
+      "step": 1395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9982877171759534e-05,
+      "loss": 1.284,
+      "step": 1400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9982014952829366e-05,
+      "loss": 1.2902,
+      "step": 1405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9981131574881233e-05,
+      "loss": 1.2825,
+      "step": 1410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.998022703978765e-05,
+      "loss": 1.2795,
+      "step": 1415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9979301349465966e-05,
+      "loss": 1.2722,
+      "step": 1420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9978354505878382e-05,
+      "loss": 1.274,
+      "step": 1425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9977386511031943e-05,
+      "loss": 1.2358,
+      "step": 1430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9976397366978508e-05,
+      "loss": 1.234,
+      "step": 1435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.997538707581479e-05,
+      "loss": 1.2799,
+      "step": 1440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9974355639682317e-05,
+      "loss": 1.3058,
+      "step": 1445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.997330306076743e-05,
+      "loss": 1.307,
+      "step": 1450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9972229341301305e-05,
+      "loss": 1.3175,
+      "step": 1455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9971134483559918e-05,
+      "loss": 1.3189,
+      "step": 1460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.997001848986405e-05,
+      "loss": 1.2476,
+      "step": 1465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9968881362579293e-05,
+      "loss": 1.2943,
+      "step": 1470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.996772310411603e-05,
+      "loss": 1.2615,
+      "step": 1475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.996654371692944e-05,
+      "loss": 1.2476,
+      "step": 1480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9965343203519484e-05,
+      "loss": 1.2674,
+      "step": 1485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9964121566430907e-05,
+      "loss": 1.2766,
+      "step": 1490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.996287880825323e-05,
+      "loss": 1.2587,
+      "step": 1495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9961614931620748e-05,
+      "loss": 1.3106,
+      "step": 1500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9960329939212516e-05,
+      "loss": 1.3115,
+      "step": 1505
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.995902383375235e-05,
+      "loss": 1.2558,
+      "step": 1510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9957696618008824e-05,
+      "loss": 1.2626,
+      "step": 1515
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.995634829479525e-05,
+      "loss": 1.2639,
+      "step": 1520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9954978866969695e-05,
+      "loss": 1.3283,
+      "step": 1525
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9953588337434947e-05,
+      "loss": 1.296,
+      "step": 1530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9952176709138538e-05,
+      "loss": 1.2943,
+      "step": 1535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.995074398507271e-05,
+      "loss": 1.2565,
+      "step": 1540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9949290168274437e-05,
+      "loss": 1.2449,
+      "step": 1545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9947815261825382e-05,
+      "loss": 1.2943,
+      "step": 1550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9946319268851938e-05,
+      "loss": 1.3077,
+      "step": 1555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9944802192525176e-05,
+      "loss": 1.3481,
+      "step": 1560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.994326403606086e-05,
+      "loss": 1.4384,
+      "step": 1565
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.994170480271945e-05,
+      "loss": 1.384,
+      "step": 1570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.994012449580607e-05,
+      "loss": 1.4798,
+      "step": 1575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9938523118670524e-05,
+      "loss": 1.4327,
+      "step": 1580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9936900674707268e-05,
+      "loss": 1.3699,
+      "step": 1585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9935257167355426e-05,
+      "loss": 1.3978,
+      "step": 1590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9933592600098753e-05,
+      "loss": 1.4105,
+      "step": 1595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9931906976465668e-05,
+      "loss": 1.3851,
+      "step": 1600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9930200300029205e-05,
+      "loss": 1.4052,
+      "step": 1605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9928472574407025e-05,
+      "loss": 1.3334,
+      "step": 1610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.992672380326142e-05,
+      "loss": 1.329,
+      "step": 1615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9924953990299285e-05,
+      "loss": 1.3553,
+      "step": 1620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9923163139272113e-05,
+      "loss": 1.3293,
+      "step": 1625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9921351253976004e-05,
+      "loss": 1.349,
+      "step": 1630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9919518338251624e-05,
+      "loss": 1.3387,
+      "step": 1635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.991766439598424e-05,
+      "loss": 1.3197,
+      "step": 1640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.991578943110368e-05,
+      "loss": 1.3008,
+      "step": 1645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.991389344758433e-05,
+      "loss": 1.2596,
+      "step": 1650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9911976449445127e-05,
+      "loss": 1.3375,
+      "step": 1655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9910038440749574e-05,
+      "loss": 1.3173,
+      "step": 1660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9908079425605683e-05,
+      "loss": 1.2858,
+      "step": 1665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9906099408166014e-05,
+      "loss": 1.324,
+      "step": 1670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9904098392627628e-05,
+      "loss": 1.3155,
+      "step": 1675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9902076383232117e-05,
+      "loss": 1.2974,
+      "step": 1680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9900033384265556e-05,
+      "loss": 1.2839,
+      "step": 1685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.989796940005852e-05,
+      "loss": 1.2931,
+      "step": 1690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.989588443498607e-05,
+      "loss": 1.2984,
+      "step": 1695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9893778493467726e-05,
+      "loss": 1.2697,
+      "step": 1700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.989165157996749e-05,
+      "loss": 1.3366,
+      "step": 1705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9889503698993812e-05,
+      "loss": 1.2425,
+      "step": 1710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9887334855099576e-05,
+      "loss": 1.3049,
+      "step": 1715
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9885145052882124e-05,
+      "loss": 1.2747,
+      "step": 1720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9882934296983197e-05,
+      "loss": 1.2771,
+      "step": 1725
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9880702592088976e-05,
+      "loss": 1.2838,
+      "step": 1730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9878449942930033e-05,
+      "loss": 1.2996,
+      "step": 1735
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.987617635428134e-05,
+      "loss": 1.2707,
+      "step": 1740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9873881830962256e-05,
+      "loss": 1.2646,
+      "step": 1745
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9871566377836514e-05,
+      "loss": 1.3083,
+      "step": 1750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.986922999981221e-05,
+      "loss": 1.3295,
+      "step": 1755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9866872701841805e-05,
+      "loss": 1.3082,
+      "step": 1760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.986449448892209e-05,
+      "loss": 1.2793,
+      "step": 1765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.98620953660942e-05,
+      "loss": 1.26,
+      "step": 1770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.985967533844359e-05,
+      "loss": 1.2887,
+      "step": 1775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9857234411100017e-05,
+      "loss": 1.2969,
+      "step": 1780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9854772589237564e-05,
+      "loss": 1.2947,
+      "step": 1785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.985228987807458e-05,
+      "loss": 1.2514,
+      "step": 1790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9849786282873706e-05,
+      "loss": 1.2994,
+      "step": 1795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9847261808941847e-05,
+      "loss": 1.3095,
+      "step": 1800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9844716461630168e-05,
+      "loss": 1.3104,
+      "step": 1805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9842150246334072e-05,
+      "loss": 1.3379,
+      "step": 1810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.983956316849321e-05,
+      "loss": 1.2999,
+      "step": 1815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9836955233591443e-05,
+      "loss": 1.3002,
+      "step": 1820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9834326447156847e-05,
+      "loss": 1.298,
+      "step": 1825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9831676814761696e-05,
+      "loss": 1.2815,
+      "step": 1830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9829006342022457e-05,
+      "loss": 1.3062,
+      "step": 1835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.982631503459977e-05,
+      "loss": 1.3009,
+      "step": 1840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9823602898198433e-05,
+      "loss": 1.3136,
+      "step": 1845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.98208699385674e-05,
+      "loss": 1.284,
+      "step": 1850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9818116161499767e-05,
+      "loss": 1.2927,
+      "step": 1855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.981534157283275e-05,
+      "loss": 1.2967,
+      "step": 1860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.981254617844769e-05,
+      "loss": 1.2909,
+      "step": 1865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9809729984270022e-05,
+      "loss": 1.277,
+      "step": 1870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9806892996269266e-05,
+      "loss": 1.2653,
+      "step": 1875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.980403522045903e-05,
+      "loss": 1.3175,
+      "step": 1880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.980115666289699e-05,
+      "loss": 1.2982,
+      "step": 1885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979825732968485e-05,
+      "loss": 1.2532,
+      "step": 1890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9795337226968375e-05,
+      "loss": 1.2995,
+      "step": 1895
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.979239636093735e-05,
+      "loss": 1.2815,
+      "step": 1900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9789434737825566e-05,
+      "loss": 1.335,
+      "step": 1905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9786452363910822e-05,
+      "loss": 1.2426,
+      "step": 1910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9783449245514894e-05,
+      "loss": 1.2865,
+      "step": 1915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9780425389003533e-05,
+      "loss": 1.2814,
+      "step": 1920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9777380800786456e-05,
+      "loss": 1.2615,
+      "step": 1925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977431548731732e-05,
+      "loss": 1.2455,
+      "step": 1930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9771229455093703e-05,
+      "loss": 1.2523,
+      "step": 1935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976812271065712e-05,
+      "loss": 1.2099,
+      "step": 1940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976499526059298e-05,
+      "loss": 1.2894,
+      "step": 1945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9761847111530583e-05,
+      "loss": 1.2304,
+      "step": 1950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.97586782701431e-05,
+      "loss": 1.2943,
+      "step": 1955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9755488743147576e-05,
+      "loss": 1.2906,
+      "step": 1960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9752278537304895e-05,
+      "loss": 1.2717,
+      "step": 1965
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.974904765941977e-05,
+      "loss": 1.25,
+      "step": 1970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9745796116340747e-05,
+      "loss": 1.2799,
+      "step": 1975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9742523914960157e-05,
+      "loss": 1.2312,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973923106221414e-05,
+      "loss": 1.2884,
+      "step": 1985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.97359175650826e-05,
+      "loss": 1.2681,
+      "step": 1990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9732583430589204e-05,
+      "loss": 1.2906,
+      "step": 1995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9729228665801362e-05,
+      "loss": 1.246,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9725853277830217e-05,
+      "loss": 1.2747,
+      "step": 2005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9722457273830633e-05,
+      "loss": 1.3106,
+      "step": 2010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9719040661001156e-05,
+      "loss": 1.2576,
+      "step": 2015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9715603446584037e-05,
+      "loss": 1.2693,
+      "step": 2020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9712145637865185e-05,
+      "loss": 1.2589,
+      "step": 2025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9708667242174163e-05,
+      "loss": 1.2469,
+      "step": 2030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9705168266884183e-05,
+      "loss": 1.2635,
+      "step": 2035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9701648719412064e-05,
+      "loss": 1.2555,
+      "step": 2040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9698108607218244e-05,
+      "loss": 1.2397,
+      "step": 2045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9694547937806752e-05,
+      "loss": 1.2403,
+      "step": 2050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9690966718725188e-05,
+      "loss": 1.2864,
+      "step": 2055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9687364957564705e-05,
+      "loss": 1.2413,
+      "step": 2060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9683742661960017e-05,
+      "loss": 1.2501,
+      "step": 2065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.968009983958935e-05,
+      "loss": 1.2706,
+      "step": 2070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9676436498174448e-05,
+      "loss": 1.2911,
+      "step": 2075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.967275264548054e-05,
+      "loss": 1.2653,
+      "step": 2080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9669048289316353e-05,
+      "loss": 1.2822,
+      "step": 2085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9665323437534058e-05,
+      "loss": 1.318,
+      "step": 2090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9661578098029273e-05,
+      "loss": 1.2557,
+      "step": 2095
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.965781227874105e-05,
+      "loss": 1.2961,
+      "step": 2100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9654025987651845e-05,
+      "loss": 1.2769,
+      "step": 2105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.965021923278752e-05,
+      "loss": 1.293,
+      "step": 2110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.96463920222173e-05,
+      "loss": 1.2305,
+      "step": 2115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9642544364053782e-05,
+      "loss": 1.2827,
+      "step": 2120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9638676266452896e-05,
+      "loss": 1.2815,
+      "step": 2125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.963478773761391e-05,
+      "loss": 1.2395,
+      "step": 2130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.963087878577939e-05,
+      "loss": 1.2816,
+      "step": 2135
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9626949419235194e-05,
+      "loss": 1.2256,
+      "step": 2140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9622999646310458e-05,
+      "loss": 1.2822,
+      "step": 2145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9619029475377573e-05,
+      "loss": 1.2774,
+      "step": 2150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9615038914852163e-05,
+      "loss": 1.2733,
+      "step": 2155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.961102797319308e-05,
+      "loss": 1.2998,
+      "step": 2160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.960699665890237e-05,
+      "loss": 1.2837,
+      "step": 2165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9602944980525267e-05,
+      "loss": 1.2506,
+      "step": 2170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.959887294665017e-05,
+      "loss": 1.2368,
+      "step": 2175
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.959478056590863e-05,
+      "loss": 1.2752,
+      "step": 2180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9590667846975324e-05,
+      "loss": 1.2696,
+      "step": 2185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9586534798568032e-05,
+      "loss": 1.2564,
+      "step": 2190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9582381429447648e-05,
+      "loss": 1.269,
+      "step": 2195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.957820774841812e-05,
+      "loss": 1.2848,
+      "step": 2200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.957401376432646e-05,
+      "loss": 1.2823,
+      "step": 2205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9569799486062712e-05,
+      "loss": 1.2451,
+      "step": 2210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9565564922559947e-05,
+      "loss": 1.2443,
+      "step": 2215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9561310082794224e-05,
+      "loss": 1.2492,
+      "step": 2220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.955703497578459e-05,
+      "loss": 1.2946,
+      "step": 2225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9552739610593048e-05,
+      "loss": 1.2667,
+      "step": 2230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9548423996324544e-05,
+      "loss": 1.2519,
+      "step": 2235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9544088142126947e-05,
+      "loss": 1.2239,
+      "step": 2240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9539732057191027e-05,
+      "loss": 1.2566,
+      "step": 2245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9535355750750444e-05,
+      "loss": 1.2681,
+      "step": 2250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9530959232081713e-05,
+      "loss": 1.2845,
+      "step": 2255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.952654251050419e-05,
+      "loss": 1.2385,
+      "step": 2260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9522105595380073e-05,
+      "loss": 1.2225,
+      "step": 2265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.951764849611435e-05,
+      "loss": 1.264,
+      "step": 2270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9513171222154796e-05,
+      "loss": 1.2981,
+      "step": 2275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.950867378299195e-05,
+      "loss": 1.2757,
+      "step": 2280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9504156188159098e-05,
+      "loss": 1.2719,
+      "step": 2285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9499618447232252e-05,
+      "loss": 1.2896,
+      "step": 2290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9495060569830126e-05,
+      "loss": 1.2691,
+      "step": 2295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9490482565614118e-05,
+      "loss": 1.2827,
+      "step": 2300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9485884444288282e-05,
+      "loss": 1.251,
+      "step": 2305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.948126621559932e-05,
+      "loss": 1.283,
+      "step": 2310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9476627889336564e-05,
+      "loss": 1.2337,
+      "step": 2315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.947196947533194e-05,
+      "loss": 1.2514,
+      "step": 2320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9467290983459945e-05,
+      "loss": 1.2583,
+      "step": 2325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.946259242363765e-05,
+      "loss": 1.2779,
+      "step": 2330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9457873805824664e-05,
+      "loss": 1.2621,
+      "step": 2335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9453135140023095e-05,
+      "loss": 1.2799,
+      "step": 2340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.944837643627757e-05,
+      "loss": 1.2895,
+      "step": 2345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9443597704675176e-05,
+      "loss": 1.2657,
+      "step": 2350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9438798955345458e-05,
+      "loss": 1.2654,
+      "step": 2355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9433980198460395e-05,
+      "loss": 1.2616,
+      "step": 2360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9429141444234373e-05,
+      "loss": 1.2972,
+      "step": 2365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9424282702924163e-05,
+      "loss": 1.2421,
+      "step": 2370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9419403984828915e-05,
+      "loss": 1.2655,
+      "step": 2375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9414505300290113e-05,
+      "loss": 1.2678,
+      "step": 2380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.940958665969157e-05,
+      "loss": 1.2602,
+      "step": 2385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.940464807345939e-05,
+      "loss": 1.2533,
+      "step": 2390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9399689552061977e-05,
+      "loss": 1.2565,
+      "step": 2395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9394711106009967e-05,
+      "loss": 1.2586,
+      "step": 2400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9389712745856245e-05,
+      "loss": 1.2651,
+      "step": 2405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9384694482195912e-05,
+      "loss": 1.2328,
+      "step": 2410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9379656325666248e-05,
+      "loss": 1.2737,
+      "step": 2415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.93745982869467e-05,
+      "loss": 1.2939,
+      "step": 2420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9369520376758872e-05,
+      "loss": 1.2772,
+      "step": 2425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9364422605866476e-05,
+      "loss": 1.29,
+      "step": 2430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.935930498507533e-05,
+      "loss": 1.2747,
+      "step": 2435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.935416752523333e-05,
+      "loss": 1.2517,
+      "step": 2440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9349010237230423e-05,
+      "loss": 1.3069,
+      "step": 2445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.934383313199858e-05,
+      "loss": 1.2483,
+      "step": 2450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9338636220511784e-05,
+      "loss": 1.2619,
+      "step": 2455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.933341951378601e-05,
+      "loss": 1.2417,
+      "step": 2460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9328183022879172e-05,
+      "loss": 1.2442,
+      "step": 2465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9322926758891145e-05,
+      "loss": 1.2252,
+      "step": 2470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.93176507329637e-05,
+      "loss": 1.224,
+      "step": 2475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9312354956280505e-05,
+      "loss": 1.2628,
+      "step": 2480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.930703944006709e-05,
+      "loss": 1.246,
+      "step": 2485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.930170419559084e-05,
+      "loss": 1.2641,
+      "step": 2490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9296349234160934e-05,
+      "loss": 1.2418,
+      "step": 2495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.929097456712837e-05,
+      "loss": 1.2798,
+      "step": 2500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.92855802058859e-05,
+      "loss": 1.2591,
+      "step": 2505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9280166161868026e-05,
+      "loss": 1.23,
+      "step": 2510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9274732446550977e-05,
+      "loss": 1.2631,
+      "step": 2515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.926927907145268e-05,
+      "loss": 1.2553,
+      "step": 2520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.926380604813272e-05,
+      "loss": 1.2565,
+      "step": 2525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9258313388192354e-05,
+      "loss": 1.2436,
+      "step": 2530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9252801103274444e-05,
+      "loss": 1.2247,
+      "step": 2535
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9247269205063458e-05,
+      "loss": 1.2535,
+      "step": 2540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9241717705285442e-05,
+      "loss": 1.2245,
+      "step": 2545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9236146615707985e-05,
+      "loss": 1.2645,
+      "step": 2550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9230555948140206e-05,
+      "loss": 1.2581,
+      "step": 2555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.922494571443272e-05,
+      "loss": 1.2468,
+      "step": 2560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9219315926477623e-05,
+      "loss": 1.2478,
+      "step": 2565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9213666596208452e-05,
+      "loss": 1.2706,
+      "step": 2570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.920799773560017e-05,
+      "loss": 1.2564,
+      "step": 2575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9202309356669147e-05,
+      "loss": 1.2638,
+      "step": 2580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9196601471473122e-05,
+      "loss": 1.2491,
+      "step": 2585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.919087409211117e-05,
+      "loss": 1.2776,
+      "step": 2590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9185127230723705e-05,
+      "loss": 1.2722,
+      "step": 2595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9179360899492424e-05,
+      "loss": 1.2535,
+      "step": 2600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9173575110640308e-05,
+      "loss": 1.2544,
+      "step": 2605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.916776987643157e-05,
+      "loss": 1.2477,
+      "step": 2610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9161945209171652e-05,
+      "loss": 1.2718,
+      "step": 2615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9156101121207176e-05,
+      "loss": 1.2541,
+      "step": 2620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9150237624925946e-05,
+      "loss": 1.2296,
+      "step": 2625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.914435473275689e-05,
+      "loss": 1.2627,
+      "step": 2630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9138452457170063e-05,
+      "loss": 1.269,
+      "step": 2635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.91325308106766e-05,
+      "loss": 1.2314,
+      "step": 2640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.91265898058287e-05,
+      "loss": 1.2473,
+      "step": 2645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9120629455219593e-05,
+      "loss": 1.2697,
+      "step": 2650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.911464977148352e-05,
+      "loss": 1.2415,
+      "step": 2655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9108650767295697e-05,
+      "loss": 1.2501,
+      "step": 2660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9102632455372302e-05,
+      "loss": 1.208,
+      "step": 2665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9096594848470436e-05,
+      "loss": 1.2179,
+      "step": 2670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9090537959388098e-05,
+      "loss": 1.2554,
+      "step": 2675
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9084461800964164e-05,
+      "loss": 1.2672,
+      "step": 2680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9078366386078343e-05,
+      "loss": 1.2932,
+      "step": 2685
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9072251727651185e-05,
+      "loss": 1.2643,
+      "step": 2690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.906611783864401e-05,
+      "loss": 1.2124,
+      "step": 2695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.905996473205891e-05,
+      "loss": 1.2361,
+      "step": 2700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9053792420938714e-05,
+      "loss": 1.2619,
+      "step": 2705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9047600918366952e-05,
+      "loss": 1.2366,
+      "step": 2710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9041390237467845e-05,
+      "loss": 1.2755,
+      "step": 2715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9035160391406262e-05,
+      "loss": 1.27,
+      "step": 2720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.902891139338769e-05,
+      "loss": 1.2391,
+      "step": 2725
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.902264325665822e-05,
+      "loss": 1.2563,
+      "step": 2730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9016355994504514e-05,
+      "loss": 1.2728,
+      "step": 2735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9010049620253767e-05,
+      "loss": 1.2541,
+      "step": 2740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9003724147273688e-05,
+      "loss": 1.2591,
+      "step": 2745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8997379588972472e-05,
+      "loss": 1.2492,
+      "step": 2750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.899101595879877e-05,
+      "loss": 1.2303,
+      "step": 2755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8984633270241662e-05,
+      "loss": 1.2625,
+      "step": 2760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8978231536830616e-05,
+      "loss": 1.2255,
+      "step": 2765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.897181077213548e-05,
+      "loss": 1.2308,
+      "step": 2770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8965370989766443e-05,
+      "loss": 1.2286,
+      "step": 2775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8958912203373995e-05,
+      "loss": 1.2831,
+      "step": 2780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.895243442664892e-05,
+      "loss": 1.2525,
+      "step": 2785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.894593767332226e-05,
+      "loss": 1.2579,
+      "step": 2790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8939421957165263e-05,
+      "loss": 1.2803,
+      "step": 2795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.893288729198939e-05,
+      "loss": 1.2096,
+      "step": 2800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8926333691646267e-05,
+      "loss": 1.2152,
+      "step": 2805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8919761170027646e-05,
+      "loss": 1.2426,
+      "step": 2810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8913169741065394e-05,
+      "loss": 1.2495,
+      "step": 2815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8906559418731463e-05,
+      "loss": 1.2752,
+      "step": 2820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.889993021703784e-05,
+      "loss": 1.2167,
+      "step": 2825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.889328215003654e-05,
+      "loss": 1.2462,
+      "step": 2830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8886615231819566e-05,
+      "loss": 1.2549,
+      "step": 2835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8879929476518874e-05,
+      "loss": 1.2588,
+      "step": 2840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.887322489830636e-05,
+      "loss": 1.2532,
+      "step": 2845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8866501511393807e-05,
+      "loss": 1.2476,
+      "step": 2850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8859759330032872e-05,
+      "loss": 1.2312,
+      "step": 2855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8852998368515062e-05,
+      "loss": 1.2752,
+      "step": 2860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8846218641171674e-05,
+      "loss": 1.2374,
+      "step": 2865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8839420162373796e-05,
+      "loss": 1.2732,
+      "step": 2870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8832602946532256e-05,
+      "loss": 1.2276,
+      "step": 2875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8825767008097603e-05,
+      "loss": 1.2522,
+      "step": 2880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8818912361560072e-05,
+      "loss": 1.2646,
+      "step": 2885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.881203902144956e-05,
+      "loss": 1.1994,
+      "step": 2890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8805147002335574e-05,
+      "loss": 1.2944,
+      "step": 2895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.879823631882723e-05,
+      "loss": 1.278,
+      "step": 2900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8791306985573203e-05,
+      "loss": 1.2472,
+      "step": 2905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.878435901726169e-05,
+      "loss": 1.2617,
+      "step": 2910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8777392428620405e-05,
+      "loss": 1.2299,
+      "step": 2915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8770407234416522e-05,
+      "loss": 1.3102,
+      "step": 2920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8763403449456653e-05,
+      "loss": 1.2551,
+      "step": 2925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8756381088586826e-05,
+      "loss": 1.235,
+      "step": 2930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8749340166692435e-05,
+      "loss": 1.262,
+      "step": 2935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.874228069869822e-05,
+      "loss": 1.2342,
+      "step": 2940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8735202699568237e-05,
+      "loss": 1.2244,
+      "step": 2945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.872810618430582e-05,
+      "loss": 1.2785,
+      "step": 2950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8720991167953553e-05,
+      "loss": 1.2856,
+      "step": 2955
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8713857665593235e-05,
+      "loss": 1.2387,
+      "step": 2960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8706705692345854e-05,
+      "loss": 1.2779,
+      "step": 2965
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8699535263371548e-05,
+      "loss": 1.223,
+      "step": 2970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8692346393869575e-05,
+      "loss": 1.2449,
+      "step": 2975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8685139099078286e-05,
+      "loss": 1.2284,
+      "step": 2980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.867791339427508e-05,
+      "loss": 1.2487,
+      "step": 2985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.867066929477639e-05,
+      "loss": 1.2484,
+      "step": 2990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8663406815937634e-05,
+      "loss": 1.2728,
+      "step": 2995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8656125973153193e-05,
+      "loss": 1.2563,
+      "step": 3000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8648826781856372e-05,
+      "loss": 1.2601,
+      "step": 3005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8641509257519367e-05,
+      "loss": 1.2851,
+      "step": 3010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.863417341565324e-05,
+      "loss": 1.2724,
+      "step": 3015
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.862681927180788e-05,
+      "loss": 1.2425,
+      "step": 3020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8619446841571966e-05,
+      "loss": 1.2554,
+      "step": 3025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.861205614057294e-05,
+      "loss": 1.1974,
+      "step": 3030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8604647184476986e-05,
+      "loss": 1.2691,
+      "step": 3035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.859721998898896e-05,
+      "loss": 1.2761,
+      "step": 3040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8589774569852405e-05,
+      "loss": 1.2319,
+      "step": 3045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.858231094284947e-05,
+      "loss": 1.2425,
+      "step": 3050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8574829123800916e-05,
+      "loss": 1.252,
+      "step": 3055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.856732912856606e-05,
+      "loss": 1.2377,
+      "step": 3060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8559810973042748e-05,
+      "loss": 1.2729,
+      "step": 3065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.855227467316732e-05,
+      "loss": 1.2705,
+      "step": 3070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.854472024491458e-05,
+      "loss": 1.303,
+      "step": 3075
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.853714770429775e-05,
+      "loss": 1.2707,
+      "step": 3080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8529557067368452e-05,
+      "loss": 1.264,
+      "step": 3085
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8521948350216673e-05,
+      "loss": 1.2247,
+      "step": 3090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8514321568970714e-05,
+      "loss": 1.277,
+      "step": 3095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8506676739797168e-05,
+      "loss": 1.2416,
+      "step": 3100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.849901387890089e-05,
+      "loss": 1.2639,
+      "step": 3105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8491333002524955e-05,
+      "loss": 1.2943,
+      "step": 3110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.848363412695063e-05,
+      "loss": 1.2739,
+      "step": 3115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8475917268497315e-05,
+      "loss": 1.2777,
+      "step": 3120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8468182443522556e-05,
+      "loss": 1.2357,
+      "step": 3125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8460429668421972e-05,
+      "loss": 1.2598,
+      "step": 3130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8452658959629225e-05,
+      "loss": 1.2673,
+      "step": 3135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8444870333615993e-05,
+      "loss": 1.2599,
+      "step": 3140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8437063806891946e-05,
+      "loss": 1.2581,
+      "step": 3145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.842923939600468e-05,
+      "loss": 1.2423,
+      "step": 3150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.842139711753971e-05,
+      "loss": 1.2551,
+      "step": 3155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8413536988120434e-05,
+      "loss": 1.2583,
+      "step": 3160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8405659024408064e-05,
+      "loss": 1.2046,
+      "step": 3165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.839776324310164e-05,
+      "loss": 1.2457,
+      "step": 3170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8389849660937968e-05,
+      "loss": 1.2223,
+      "step": 3175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.838191829469156e-05,
+      "loss": 1.1858,
+      "step": 3180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8373969161174665e-05,
+      "loss": 1.2131,
+      "step": 3185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8366002277237162e-05,
+      "loss": 1.2404,
+      "step": 3190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8358017659766572e-05,
+      "loss": 1.2425,
+      "step": 3195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8350015325688e-05,
+      "loss": 1.2444,
+      "step": 3200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8341995291964103e-05,
+      "loss": 1.2192,
+      "step": 3205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8333957575595066e-05,
+      "loss": 1.2746,
+      "step": 3210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8325902193618547e-05,
+      "loss": 1.2216,
+      "step": 3215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.831782916310965e-05,
+      "loss": 1.2779,
+      "step": 3220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8309738501180897e-05,
+      "loss": 1.312,
+      "step": 3225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8301630224982173e-05,
+      "loss": 1.238,
+      "step": 3230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8293504351700712e-05,
+      "loss": 1.2619,
+      "step": 3235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8285360898561035e-05,
+      "loss": 1.2462,
+      "step": 3240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.827719988282494e-05,
+      "loss": 1.2367,
+      "step": 3245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.826902132179144e-05,
+      "loss": 1.2649,
+      "step": 3250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8260825232796758e-05,
+      "loss": 1.258,
+      "step": 3255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8252611633214247e-05,
+      "loss": 1.228,
+      "step": 3260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.82443805404544e-05,
+      "loss": 1.208,
+      "step": 3265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8236131971964775e-05,
+      "loss": 1.2569,
+      "step": 3270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8227865945229978e-05,
+      "loss": 1.2435,
+      "step": 3275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.821958247777163e-05,
+      "loss": 1.2539,
+      "step": 3280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8211281587148305e-05,
+      "loss": 1.2749,
+      "step": 3285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8202963290955523e-05,
+      "loss": 1.2343,
+      "step": 3290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.819462760682569e-05,
+      "loss": 1.2426,
+      "step": 3295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.818627455242808e-05,
+      "loss": 1.2714,
+      "step": 3300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.817790414546877e-05,
+      "loss": 1.2755,
+      "step": 3305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.816951640369064e-05,
+      "loss": 1.2361,
+      "step": 3310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.81611113448733e-05,
+      "loss": 1.2403,
+      "step": 3315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8152688986833073e-05,
+      "loss": 1.2117,
+      "step": 3320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8144249347422946e-05,
+      "loss": 1.2105,
+      "step": 3325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.813579244453255e-05,
+      "loss": 1.224,
+      "step": 3330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8127318296088093e-05,
+      "loss": 1.257,
+      "step": 3335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8118826920052352e-05,
+      "loss": 1.26,
+      "step": 3340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8110318334424617e-05,
+      "loss": 1.2564,
+      "step": 3345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8101792557240653e-05,
+      "loss": 1.2597,
+      "step": 3350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8093249606572673e-05,
+      "loss": 1.273,
+      "step": 3355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8084689500529288e-05,
+      "loss": 1.2615,
+      "step": 3360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8076112257255477e-05,
+      "loss": 1.221,
+      "step": 3365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8067517894932548e-05,
+      "loss": 1.2582,
+      "step": 3370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8058906431778085e-05,
+      "loss": 1.2325,
+      "step": 3375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8050277886045932e-05,
+      "loss": 1.2652,
+      "step": 3380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8041632276026138e-05,
+      "loss": 1.2887,
+      "step": 3385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8032969620044923e-05,
+      "loss": 1.2641,
+      "step": 3390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8024289936464644e-05,
+      "loss": 1.2526,
+      "step": 3395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8015593243683747e-05,
+      "loss": 1.2613,
+      "step": 3400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8006879560136733e-05,
+      "loss": 1.2615,
+      "step": 3405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7998148904294124e-05,
+      "loss": 1.2371,
+      "step": 3410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7989401294662414e-05,
+      "loss": 1.2711,
+      "step": 3415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7980636749784028e-05,
+      "loss": 1.2724,
+      "step": 3420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7971855288237302e-05,
+      "loss": 1.2814,
+      "step": 3425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7963056928636424e-05,
+      "loss": 1.2762,
+      "step": 3430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7954241689631397e-05,
+      "loss": 1.2564,
+      "step": 3435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7945409589908013e-05,
+      "loss": 1.2392,
+      "step": 3440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7936560648187793e-05,
+      "loss": 1.2382,
+      "step": 3445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7927694883227968e-05,
+      "loss": 1.2411,
+      "step": 3450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7918812313821422e-05,
+      "loss": 1.2909,
+      "step": 3455
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7909912958796663e-05,
+      "loss": 1.2482,
+      "step": 3460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7900996837017778e-05,
+      "loss": 1.346,
+      "step": 3465
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7892063967384403e-05,
+      "loss": 1.3127,
+      "step": 3470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.788311436883166e-05,
+      "loss": 1.2986,
+      "step": 3475
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7874148060330142e-05,
+      "loss": 1.3083,
+      "step": 3480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.786516506088586e-05,
+      "loss": 1.3185,
+      "step": 3485
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7856165389540208e-05,
+      "loss": 1.3787,
+      "step": 3490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.784714906536991e-05,
+      "loss": 1.3987,
+      "step": 3495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7838116107487e-05,
+      "loss": 1.3606,
+      "step": 3500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7829066535038765e-05,
+      "loss": 1.3729,
+      "step": 3505
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.782000036720771e-05,
+      "loss": 1.3351,
+      "step": 3510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7810917623211524e-05,
+      "loss": 1.3338,
+      "step": 3515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7801818322303018e-05,
+      "loss": 1.33,
+      "step": 3520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.779270248377012e-05,
+      "loss": 1.2902,
+      "step": 3525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7783570126935793e-05,
+      "loss": 1.3373,
+      "step": 3530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.777442127115803e-05,
+      "loss": 1.2832,
+      "step": 3535
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7765255935829784e-05,
+      "loss": 1.2954,
+      "step": 3540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7756074140378943e-05,
+      "loss": 1.3217,
+      "step": 3545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7746875904268294e-05,
+      "loss": 1.2886,
+      "step": 3550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.773766124699547e-05,
+      "loss": 1.2852,
+      "step": 3555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7728430188092902e-05,
+      "loss": 1.3117,
+      "step": 3560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7719182747127805e-05,
+      "loss": 1.295,
+      "step": 3565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.77099189437021e-05,
+      "loss": 1.3147,
+      "step": 3570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.770063879745241e-05,
+      "loss": 1.2886,
+      "step": 3575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.769134232804999e-05,
+      "loss": 1.2906,
+      "step": 3580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.76820295552007e-05,
+      "loss": 1.2706,
+      "step": 3585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7672700498644946e-05,
+      "loss": 1.3159,
+      "step": 3590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.766335517815767e-05,
+      "loss": 1.3029,
+      "step": 3595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7653993613548276e-05,
+      "loss": 1.27,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7644615824660606e-05,
+      "loss": 1.2542,
+      "step": 3605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.763522183137289e-05,
+      "loss": 1.2603,
+      "step": 3610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7625811653597714e-05,
+      "loss": 1.2725,
+      "step": 3615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7616385311281957e-05,
+      "loss": 1.2753,
+      "step": 3620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7606942824406773e-05,
+      "loss": 1.271,
+      "step": 3625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7597484212987534e-05,
+      "loss": 1.3,
+      "step": 3630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7588009497073794e-05,
+      "loss": 1.278,
+      "step": 3635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7578518696749242e-05,
+      "loss": 1.2646,
+      "step": 3640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7569011832131656e-05,
+      "loss": 1.3017,
+      "step": 3645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.755948892337288e-05,
+      "loss": 1.294,
+      "step": 3650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.754994999065875e-05,
+      "loss": 1.2619,
+      "step": 3655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.754039505420908e-05,
+      "loss": 1.2736,
+      "step": 3660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7530824134277603e-05,
+      "loss": 1.2723,
+      "step": 3665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7521237251151932e-05,
+      "loss": 1.2883,
+      "step": 3670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.751163442515352e-05,
+      "loss": 1.3067,
+      "step": 3675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7502015676637617e-05,
+      "loss": 1.284,
+      "step": 3680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.749238102599321e-05,
+      "loss": 1.2915,
+      "step": 3685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.748273049364301e-05,
+      "loss": 1.2955,
+      "step": 3690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7473064100043385e-05,
+      "loss": 1.3036,
+      "step": 3695
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.746338186568433e-05,
+      "loss": 1.2864,
+      "step": 3700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.745368381108941e-05,
+      "loss": 1.3031,
+      "step": 3705
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.744396995681573e-05,
+      "loss": 1.3082,
+      "step": 3710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.743424032345388e-05,
+      "loss": 1.2701,
+      "step": 3715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7424494931627906e-05,
+      "loss": 1.3224,
+      "step": 3720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7414733801995248e-05,
+      "loss": 1.3205,
+      "step": 3725
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7404956955246715e-05,
+      "loss": 1.2921,
+      "step": 3730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7395164412106425e-05,
+      "loss": 1.282,
+      "step": 3735
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7385356193331768e-05,
+      "loss": 1.3137,
+      "step": 3740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7375532319713366e-05,
+      "loss": 1.2913,
+      "step": 3745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7365692812075024e-05,
+      "loss": 1.2841,
+      "step": 3750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.735583769127368e-05,
+      "loss": 1.2971,
+      "step": 3755
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7345966978199377e-05,
+      "loss": 1.2422,
+      "step": 3760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7336080693775207e-05,
+      "loss": 1.3036,
+      "step": 3765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7326178858957257e-05,
+      "loss": 1.3398,
+      "step": 3770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.73162614947346e-05,
+      "loss": 1.2325,
+      "step": 3775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7306328622129197e-05,
+      "loss": 1.2482,
+      "step": 3780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.729638026219591e-05,
+      "loss": 1.321,
+      "step": 3785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7286416436022415e-05,
+      "loss": 1.2955,
+      "step": 3790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7276437164729173e-05,
+      "loss": 1.2773,
+      "step": 3795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.726644246946939e-05,
+      "loss": 1.2761,
+      "step": 3800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7256432371428956e-05,
+      "loss": 1.2698,
+      "step": 3805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7246406891826422e-05,
+      "loss": 1.2834,
+      "step": 3810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7236366051912942e-05,
+      "loss": 1.2817,
+      "step": 3815
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.722630987297222e-05,
+      "loss": 1.2878,
+      "step": 3820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.721623837632048e-05,
+      "loss": 1.2825,
+      "step": 3825
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7206151583306417e-05,
+      "loss": 1.2239,
+      "step": 3830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7196049515311152e-05,
+      "loss": 1.2266,
+      "step": 3835
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7185932193748177e-05,
+      "loss": 1.2509,
+      "step": 3840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7175799640063323e-05,
+      "loss": 1.2736,
+      "step": 3845
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7165651875734706e-05,
+      "loss": 1.245,
+      "step": 3850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7155488922272688e-05,
+      "loss": 1.2409,
+      "step": 3855
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7145310801219823e-05,
+      "loss": 1.2821,
+      "step": 3860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7135117534150812e-05,
+      "loss": 1.2679,
+      "step": 3865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7124909142672475e-05,
+      "loss": 1.2854,
+      "step": 3870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7114685648423687e-05,
+      "loss": 1.2393,
+      "step": 3875
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.710444707307532e-05,
+      "loss": 1.231,
+      "step": 3880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7094193438330237e-05,
+      "loss": 1.1979,
+      "step": 3885
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.708392476592321e-05,
+      "loss": 1.2652,
+      "step": 3890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7073641077620887e-05,
+      "loss": 1.3002,
+      "step": 3895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7063342395221746e-05,
+      "loss": 1.2957,
+      "step": 3900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7053028740556058e-05,
+      "loss": 1.2535,
+      "step": 3905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.704270013548581e-05,
+      "loss": 1.2737,
+      "step": 3910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7032356601904698e-05,
+      "loss": 1.2354,
+      "step": 3915
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7021998161738056e-05,
+      "loss": 1.2608,
+      "step": 3920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.701162483694282e-05,
+      "loss": 1.2307,
+      "step": 3925
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7001236649507467e-05,
+      "loss": 1.2165,
+      "step": 3930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6990833621451983e-05,
+      "loss": 1.27,
+      "step": 3935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.698041577482782e-05,
+      "loss": 1.227,
+      "step": 3940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.696998313171783e-05,
+      "loss": 1.2661,
+      "step": 3945
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6959535714236235e-05,
+      "loss": 1.2685,
+      "step": 3950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.694907354452857e-05,
+      "loss": 1.2117,
+      "step": 3955
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.693859664477165e-05,
+      "loss": 1.2506,
+      "step": 3960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6928105037173506e-05,
+      "loss": 1.2371,
+      "step": 3965
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6917598743973344e-05,
+      "loss": 1.2391,
+      "step": 3970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.69070777874415e-05,
+      "loss": 1.2493,
+      "step": 3975
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6896542189879398e-05,
+      "loss": 1.2521,
+      "step": 3980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6885991973619493e-05,
+      "loss": 1.2184,
+      "step": 3985
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6875427161025227e-05,
+      "loss": 1.2865,
+      "step": 3990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.686484777449098e-05,
+      "loss": 1.2491,
+      "step": 3995
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.234572172164917,
+      "eval_runtime": 1569.4882,
+      "eval_samples_per_second": 18.036,
+      "eval_steps_per_second": 1.128,
+      "step": 3996
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6854253836442027e-05,
+      "loss": 1.254,
+      "step": 4000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.684364536933449e-05,
+      "loss": 1.2453,
+      "step": 4005
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6833022395655286e-05,
+      "loss": 1.2569,
+      "step": 4010
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6822384937922086e-05,
+      "loss": 1.2134,
+      "step": 4015
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.681173301868325e-05,
+      "loss": 1.2483,
+      "step": 4020
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6801066660517808e-05,
+      "loss": 1.2755,
+      "step": 4025
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.679038588603539e-05,
+      "loss": 1.1966,
+      "step": 4030
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.677969071787618e-05,
+      "loss": 1.2462,
+      "step": 4035
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.676898117871088e-05,
+      "loss": 1.2295,
+      "step": 4040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6758257291240655e-05,
+      "loss": 1.2179,
+      "step": 4045
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6747519078197076e-05,
+      "loss": 1.2564,
+      "step": 4050
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6736766562342083e-05,
+      "loss": 1.2066,
+      "step": 4055
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6725999766467943e-05,
+      "loss": 1.2465,
+      "step": 4060
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6715218713397174e-05,
+      "loss": 1.2091,
+      "step": 4065
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6704423425982537e-05,
+      "loss": 1.2519,
+      "step": 4070
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.669361392710695e-05,
+      "loss": 1.2317,
+      "step": 4075
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.668279023968346e-05,
+      "loss": 1.2555,
+      "step": 4080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.667195238665519e-05,
+      "loss": 1.2344,
+      "step": 4085
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6661100390995296e-05,
+      "loss": 1.2233,
+      "step": 4090
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.66502342757069e-05,
+      "loss": 1.2687,
+      "step": 4095
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.6639354063823058e-05,
+      "loss": 1.2618,
+      "step": 4100
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.662845977840671e-05,
+      "loss": 1.2134,
+      "step": 4105
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.6617551442550633e-05,
+      "loss": 1.2157,
+      "step": 4110
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.6606629079377376e-05,
+      "loss": 1.1967,
+      "step": 4115
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.6595692712039225e-05,
+      "loss": 1.2417,
+      "step": 4120
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.6584742363718152e-05,
+      "loss": 1.1993,
+      "step": 4125
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.6573778057625773e-05,
+      "loss": 1.2363,
+      "step": 4130
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.656279981700327e-05,
+      "loss": 1.2077,
+      "step": 4135
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.655180766512138e-05,
+      "loss": 1.2203,
+      "step": 4140
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6540801625280323e-05,
+      "loss": 1.2353,
+      "step": 4145
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6529781720809758e-05,
+      "loss": 1.2411,
+      "step": 4150
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.651874797506873e-05,
+      "loss": 1.2297,
+      "step": 4155
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.650770041144563e-05,
+      "loss": 1.2246,
+      "step": 4160
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6496639053358126e-05,
+      "loss": 1.1892,
+      "step": 4165
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6485563924253142e-05,
+      "loss": 1.2321,
+      "step": 4170
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6474475047606783e-05,
+      "loss": 1.2214,
+      "step": 4175
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6463372446924296e-05,
+      "loss": 1.2432,
+      "step": 4180
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6452256145740023e-05,
+      "loss": 1.2332,
+      "step": 4185
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.644112616761734e-05,
+      "loss": 1.2089,
+      "step": 4190
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6429982536148628e-05,
+      "loss": 1.2415,
+      "step": 4195
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.641882527495519e-05,
+      "loss": 1.2295,
+      "step": 4200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6407654407687233e-05,
+      "loss": 1.2191,
+      "step": 4205
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6396469958023808e-05,
+      "loss": 1.2261,
+      "step": 4210
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6385271949672742e-05,
+      "loss": 1.2144,
+      "step": 4215
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.6374060406370613e-05,
+      "loss": 1.2292,
+      "step": 4220
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.636283535188269e-05,
+      "loss": 1.1946,
+      "step": 4225
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.6351596810002883e-05,
+      "loss": 1.2461,
+      "step": 4230
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.6340344804553683e-05,
+      "loss": 1.2374,
+      "step": 4235
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.6329079359386124e-05,
+      "loss": 1.2091,
+      "step": 4240
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.631780049837973e-05,
+      "loss": 1.2048,
+      "step": 4245
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.6306508245442463e-05,
+      "loss": 1.2057,
+      "step": 4250
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.629520262451067e-05,
+      "loss": 1.2597,
+      "step": 4255
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.6283883659549037e-05,
+      "loss": 1.2054,
+      "step": 4260
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.6272551374550532e-05,
+      "loss": 1.2482,
+      "step": 4265
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.626120579353636e-05,
+      "loss": 1.2499,
+      "step": 4270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.6249846940555905e-05,
+      "loss": 1.2561,
+      "step": 4275
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.6238474839686698e-05,
+      "loss": 1.2347,
+      "step": 4280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.622708951503433e-05,
+      "loss": 1.2299,
+      "step": 4285
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.6215690990732443e-05,
+      "loss": 1.2321,
+      "step": 4290
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.6204279290942647e-05,
+      "loss": 1.2416,
+      "step": 4295
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6192854439854482e-05,
+      "loss": 1.2178,
+      "step": 4300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6181416461685365e-05,
+      "loss": 1.2185,
+      "step": 4305
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6169965380680547e-05,
+      "loss": 1.2231,
+      "step": 4310
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6158501221113035e-05,
+      "loss": 1.2295,
+      "step": 4315
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.614702400728358e-05,
+      "loss": 1.2358,
+      "step": 4320
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6135533763520586e-05,
+      "loss": 1.2613,
+      "step": 4325
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.612403051418009e-05,
+      "loss": 1.2162,
+      "step": 4330
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6112514283645693e-05,
+      "loss": 1.2285,
+      "step": 4335
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6100985096328506e-05,
+      "loss": 1.2302,
+      "step": 4340
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6089442976667112e-05,
+      "loss": 1.2525,
+      "step": 4345
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6077887949127507e-05,
+      "loss": 1.2596,
+      "step": 4350
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6066320038203046e-05,
+      "loss": 1.2018,
+      "step": 4355
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.605473926841439e-05,
+      "loss": 1.1908,
+      "step": 4360
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6043145664309464e-05,
+      "loss": 1.2488,
+      "step": 4365
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.603153925046339e-05,
+      "loss": 1.2556,
+      "step": 4370
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.601992005147845e-05,
+      "loss": 1.2023,
+      "step": 4375
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.6008288091984025e-05,
+      "loss": 1.2343,
+      "step": 4380
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.599664339663654e-05,
+      "loss": 1.2438,
+      "step": 4385
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.598498599011942e-05,
+      "loss": 1.2226,
+      "step": 4390
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5973315897143043e-05,
+      "loss": 1.2143,
+      "step": 4395
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.596163314244466e-05,
+      "loss": 1.2081,
+      "step": 4400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.594993775078837e-05,
+      "loss": 1.254,
+      "step": 4405
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.593822974696507e-05,
+      "loss": 1.1906,
+      "step": 4410
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.592650915579237e-05,
+      "loss": 1.2095,
+      "step": 4415
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.591477600211458e-05,
+      "loss": 1.2251,
+      "step": 4420
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5903030310802628e-05,
+      "loss": 1.2008,
+      "step": 4425
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.589127210675402e-05,
+      "loss": 1.2362,
+      "step": 4430
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5879501414892793e-05,
+      "loss": 1.2216,
+      "step": 4435
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5867718260169446e-05,
+      "loss": 1.2083,
+      "step": 4440
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.585592266756089e-05,
+      "loss": 1.2232,
+      "step": 4445
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5844114662070423e-05,
+      "loss": 1.2777,
+      "step": 4450
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5832294268727634e-05,
+      "loss": 1.2242,
+      "step": 4455
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5820461512588377e-05,
+      "loss": 1.226,
+      "step": 4460
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5808616418734712e-05,
+      "loss": 1.2378,
+      "step": 4465
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.579675901227485e-05,
+      "loss": 1.2518,
+      "step": 4470
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5784889318343112e-05,
+      "loss": 1.2288,
+      "step": 4475
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5773007362099848e-05,
+      "loss": 1.2097,
+      "step": 4480
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.576111316873141e-05,
+      "loss": 1.2321,
+      "step": 4485
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5749206763450082e-05,
+      "loss": 1.1831,
+      "step": 4490
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5737288171494048e-05,
+      "loss": 1.2144,
+      "step": 4495
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.572535741812731e-05,
+      "loss": 1.2062,
+      "step": 4500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.571341452863966e-05,
+      "loss": 1.2381,
+      "step": 4505
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.57014595283466e-05,
+      "loss": 1.21,
+      "step": 4510
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5689492442589322e-05,
+      "loss": 1.2118,
+      "step": 4515
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5677513296734624e-05,
+      "loss": 1.2046,
+      "step": 4520
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5665522116174866e-05,
+      "loss": 1.2071,
+      "step": 4525
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5653518926327928e-05,
+      "loss": 1.2114,
+      "step": 4530
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.564150375263714e-05,
+      "loss": 1.1928,
+      "step": 4535
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5629476620571233e-05,
+      "loss": 1.2321,
+      "step": 4540
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.561743755562429e-05,
+      "loss": 1.2001,
+      "step": 4545
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.560538658331569e-05,
+      "loss": 1.2211,
+      "step": 4550
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5593323729190042e-05,
+      "loss": 1.209,
+      "step": 4555
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5581249018817155e-05,
+      "loss": 1.2271,
+      "step": 4560
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5569162477791956e-05,
+      "loss": 1.2167,
+      "step": 4565
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5557064131734462e-05,
+      "loss": 1.2143,
+      "step": 4570
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5544954006289706e-05,
+      "loss": 1.2107,
+      "step": 4575
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5532832127127694e-05,
+      "loss": 1.1899,
+      "step": 4580
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.552069851994334e-05,
+      "loss": 1.2185,
+      "step": 4585
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.550855321045643e-05,
+      "loss": 1.2003,
+      "step": 4590
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.549639622441154e-05,
+      "loss": 1.2139,
+      "step": 4595
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5484227587578008e-05,
+      "loss": 1.2446,
+      "step": 4600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5472047325749863e-05,
+      "loss": 1.2315,
+      "step": 4605
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.545985546474578e-05,
+      "loss": 1.2411,
+      "step": 4610
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5447652030409018e-05,
+      "loss": 1.2195,
+      "step": 4615
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.543543704860737e-05,
+      "loss": 1.2218,
+      "step": 4620
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5423210545233108e-05,
+      "loss": 1.1849,
+      "step": 4625
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5410972546202917e-05,
+      "loss": 1.2229,
+      "step": 4630
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.539872307745786e-05,
+      "loss": 1.2249,
+      "step": 4635
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.538646216496331e-05,
+      "loss": 1.2274,
+      "step": 4640
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5374189834708898e-05,
+      "loss": 1.2252,
+      "step": 4645
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5361906112708446e-05,
+      "loss": 1.2746,
+      "step": 4650
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5349611024999943e-05,
+      "loss": 1.2398,
+      "step": 4655
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.533730459764546e-05,
+      "loss": 1.2227,
+      "step": 4660
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5324986856731093e-05,
+      "loss": 1.203,
+      "step": 4665
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5312657828366946e-05,
+      "loss": 1.2301,
+      "step": 4670
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5300317538687025e-05,
+      "loss": 1.2158,
+      "step": 4675
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.528796601384922e-05,
+      "loss": 1.2231,
+      "step": 4680
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.527560328003523e-05,
+      "loss": 1.2181,
+      "step": 4685
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5263229363450517e-05,
+      "loss": 1.2443,
+      "step": 4690
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5250844290324248e-05,
+      "loss": 1.1862,
+      "step": 4695
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5238448086909237e-05,
+      "loss": 1.2155,
+      "step": 4700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5226040779481889e-05,
+      "loss": 1.279,
+      "step": 4705
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5213622394342156e-05,
+      "loss": 1.228,
+      "step": 4710
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5201192957813453e-05,
+      "loss": 1.2074,
+      "step": 4715
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5188752496242641e-05,
+      "loss": 1.2708,
+      "step": 4720
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5176301035999937e-05,
+      "loss": 1.2233,
+      "step": 4725
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.516383860347888e-05,
+      "loss": 1.2081,
+      "step": 4730
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5151365225096261e-05,
+      "loss": 1.2198,
+      "step": 4735
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.513888092729208e-05,
+      "loss": 1.2561,
+      "step": 4740
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5126385736529477e-05,
+      "loss": 1.1811,
+      "step": 4745
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5113879679294683e-05,
+      "loss": 1.2259,
+      "step": 4750
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5101362782096967e-05,
+      "loss": 1.2056,
+      "step": 4755
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.508883507146857e-05,
+      "loss": 1.2654,
+      "step": 4760
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5076296573964659e-05,
+      "loss": 1.2323,
+      "step": 4765
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5063747316163263e-05,
+      "loss": 1.2267,
+      "step": 4770
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5051187324665222e-05,
+      "loss": 1.2574,
+      "step": 4775
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5038616626094124e-05,
+      "loss": 1.2207,
+      "step": 4780
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.502603524709626e-05,
+      "loss": 1.2017,
+      "step": 4785
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5013443214340556e-05,
+      "loss": 1.2366,
+      "step": 4790
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5000840554518518e-05,
+      "loss": 1.1977,
+      "step": 4795
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4988227294344183e-05,
+      "loss": 1.2511,
+      "step": 4800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4975603460554063e-05,
+      "loss": 1.2417,
+      "step": 4805
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4962969079907068e-05,
+      "loss": 1.2186,
+      "step": 4810
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4950324179184479e-05,
+      "loss": 1.2132,
+      "step": 4815
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4937668785189867e-05,
+      "loss": 1.2351,
+      "step": 4820
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4925002924749053e-05,
+      "loss": 1.2336,
+      "step": 4825
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4912326624710032e-05,
+      "loss": 1.219,
+      "step": 4830
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4899639911942948e-05,
+      "loss": 1.1971,
+      "step": 4835
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4886942813339992e-05,
+      "loss": 1.2162,
+      "step": 4840
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4874235355815395e-05,
+      "loss": 1.2169,
+      "step": 4845
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4861517566305329e-05,
+      "loss": 1.2399,
+      "step": 4850
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4848789471767869e-05,
+      "loss": 1.2001,
+      "step": 4855
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4836051099182938e-05,
+      "loss": 1.194,
+      "step": 4860
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4823302475552247e-05,
+      "loss": 1.2079,
+      "step": 4865
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4810543627899235e-05,
+      "loss": 1.1689,
+      "step": 4870
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4797774583269005e-05,
+      "loss": 1.2265,
+      "step": 4875
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4784995368728283e-05,
+      "loss": 1.2116,
+      "step": 4880
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4772206011365355e-05,
+      "loss": 1.2244,
+      "step": 4885
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4759406538289995e-05,
+      "loss": 1.2176,
+      "step": 4890
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4746596976633436e-05,
+      "loss": 1.2163,
+      "step": 4895
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.4733777353548279e-05,
+      "loss": 1.2405,
+      "step": 4900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.4720947696208463e-05,
+      "loss": 1.2245,
+      "step": 4905
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.4708108031809192e-05,
+      "loss": 1.2066,
+      "step": 4910
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.4695258387566886e-05,
+      "loss": 1.2337,
+      "step": 4915
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.4682398790719115e-05,
+      "loss": 1.1965,
+      "step": 4920
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.4669529268524549e-05,
+      "loss": 1.2117,
+      "step": 4925
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.4656649848262895e-05,
+      "loss": 1.1858,
+      "step": 4930
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.4643760557234845e-05,
+      "loss": 1.2335,
+      "step": 4935
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.4630861422762009e-05,
+      "loss": 1.2177,
+      "step": 4940
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.4617952472186863e-05,
+      "loss": 1.192,
+      "step": 4945
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.4605033732872693e-05,
+      "loss": 1.2387,
+      "step": 4950
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.4592105232203533e-05,
+      "loss": 1.2091,
+      "step": 4955
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.4579166997584109e-05,
+      "loss": 1.2664,
+      "step": 4960
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.4566219056439777e-05,
+      "loss": 1.1871,
+      "step": 4965
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.4553261436216472e-05,
+      "loss": 1.1965,
+      "step": 4970
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.4540294164380649e-05,
+      "loss": 1.2692,
+      "step": 4975
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4527317268419207e-05,
+      "loss": 1.1682,
+      "step": 4980
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4514330775839465e-05,
+      "loss": 1.2188,
+      "step": 4985
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4501334714169073e-05,
+      "loss": 1.1854,
+      "step": 4990
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.448832911095596e-05,
+      "loss": 1.2459,
+      "step": 4995
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4475313993768292e-05,
+      "loss": 1.2555,
+      "step": 5000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4462289390194398e-05,
+      "loss": 1.2139,
+      "step": 5005
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4449255327842709e-05,
+      "loss": 1.2471,
+      "step": 5010
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4436211834341716e-05,
+      "loss": 1.2355,
+      "step": 5015
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.442315893733989e-05,
+      "loss": 1.1865,
+      "step": 5020
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4410096664505647e-05,
+      "loss": 1.2345,
+      "step": 5025
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4397025043527266e-05,
+      "loss": 1.2023,
+      "step": 5030
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4383944102112854e-05,
+      "loss": 1.2217,
+      "step": 5035
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.437085386799026e-05,
+      "loss": 1.1931,
+      "step": 5040
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4357754368907039e-05,
+      "loss": 1.216,
+      "step": 5045
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4344645632630387e-05,
+      "loss": 1.194,
+      "step": 5050
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4331527686947073e-05,
+      "loss": 1.2565,
+      "step": 5055
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.43184005596634e-05,
+      "loss": 1.2483,
+      "step": 5060
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4305264278605113e-05,
+      "loss": 1.2379,
+      "step": 5065
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4292118871617381e-05,
+      "loss": 1.2214,
+      "step": 5070
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4278964366564707e-05,
+      "loss": 1.2013,
+      "step": 5075
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.426580079133088e-05,
+      "loss": 1.2587,
+      "step": 5080
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4252628173818914e-05,
+      "loss": 1.2261,
+      "step": 5085
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4239446541950996e-05,
+      "loss": 1.2326,
+      "step": 5090
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4226255923668417e-05,
+      "loss": 1.1832,
+      "step": 5095
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4213056346931514e-05,
+      "loss": 1.2199,
+      "step": 5100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4199847839719618e-05,
+      "loss": 1.2185,
+      "step": 5105
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.418663043003099e-05,
+      "loss": 1.2202,
+      "step": 5110
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4173404145882755e-05,
+      "loss": 1.1711,
+      "step": 5115
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4160169015310856e-05,
+      "loss": 1.2137,
+      "step": 5120
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4146925066369988e-05,
+      "loss": 1.2195,
+      "step": 5125
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4133672327133536e-05,
+      "loss": 1.1873,
+      "step": 5130
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4120410825693518e-05,
+      "loss": 1.2104,
+      "step": 5135
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.410714059016052e-05,
+      "loss": 1.2083,
+      "step": 5140
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4093861648663656e-05,
+      "loss": 1.2012,
+      "step": 5145
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4080574029350484e-05,
+      "loss": 1.252,
+      "step": 5150
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4067277760386957e-05,
+      "loss": 1.2274,
+      "step": 5155
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4053972869957363e-05,
+      "loss": 1.2403,
+      "step": 5160
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4040659386264263e-05,
+      "loss": 1.2052,
+      "step": 5165
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4027337337528443e-05,
+      "loss": 1.2095,
+      "step": 5170
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4014006751988839e-05,
+      "loss": 1.1496,
+      "step": 5175
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4000667657902472e-05,
+      "loss": 1.189,
+      "step": 5180
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3987320083544413e-05,
+      "loss": 1.1891,
+      "step": 5185
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3973964057207706e-05,
+      "loss": 1.1995,
+      "step": 5190
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.39605996072033e-05,
+      "loss": 1.2122,
+      "step": 5195
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3947226761860022e-05,
+      "loss": 1.2182,
+      "step": 5200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.393384554952447e-05,
+      "loss": 1.1902,
+      "step": 5205
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.392045599856099e-05,
+      "loss": 1.2646,
+      "step": 5210
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3907058137351608e-05,
+      "loss": 1.1984,
+      "step": 5215
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3893651994295954e-05,
+      "loss": 1.221,
+      "step": 5220
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3880237597811222e-05,
+      "loss": 1.1956,
+      "step": 5225
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3866814976332092e-05,
+      "loss": 1.2005,
+      "step": 5230
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.385338415831069e-05,
+      "loss": 1.1886,
+      "step": 5235
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3839945172216509e-05,
+      "loss": 1.1807,
+      "step": 5240
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3826498046536356e-05,
+      "loss": 1.2072,
+      "step": 5245
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3813042809774295e-05,
+      "loss": 1.2331,
+      "step": 5250
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.379957949045158e-05,
+      "loss": 1.2243,
+      "step": 5255
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3786108117106598e-05,
+      "loss": 1.2175,
+      "step": 5260
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3772628718294811e-05,
+      "loss": 1.2079,
+      "step": 5265
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3759141322588694e-05,
+      "loss": 1.2489,
+      "step": 5270
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3745645958577663e-05,
+      "loss": 1.2316,
+      "step": 5275
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3732142654868033e-05,
+      "loss": 1.2503,
+      "step": 5280
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.371863144008295e-05,
+      "loss": 1.2269,
+      "step": 5285
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3705112342862328e-05,
+      "loss": 1.2174,
+      "step": 5290
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3691585391862785e-05,
+      "loss": 1.2045,
+      "step": 5295
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3678050615757593e-05,
+      "loss": 1.1688,
+      "step": 5300
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3664508043236602e-05,
+      "loss": 1.2451,
+      "step": 5305
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.36509577030062e-05,
+      "loss": 1.1965,
+      "step": 5310
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3637399623789233e-05,
+      "loss": 1.2355,
+      "step": 5315
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3623833834324951e-05,
+      "loss": 1.2132,
+      "step": 5320
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3610260363368952e-05,
+      "loss": 1.2369,
+      "step": 5325
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3596679239693113e-05,
+      "loss": 1.2358,
+      "step": 5330
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3583090492085529e-05,
+      "loss": 1.226,
+      "step": 5335
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.356949414935047e-05,
+      "loss": 1.2355,
+      "step": 5340
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3555890240308286e-05,
+      "loss": 1.1967,
+      "step": 5345
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.354227879379538e-05,
+      "loss": 1.2033,
+      "step": 5350
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3528659838664124e-05,
+      "loss": 1.2402,
+      "step": 5355
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3515033403782814e-05,
+      "loss": 1.2005,
+      "step": 5360
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3501399518035594e-05,
+      "loss": 1.2047,
+      "step": 5365
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.34877582103224e-05,
+      "loss": 1.1888,
+      "step": 5370
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3474109509558912e-05,
+      "loss": 1.1839,
+      "step": 5375
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3460453444676467e-05,
+      "loss": 1.1995,
+      "step": 5380
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3446790044622025e-05,
+      "loss": 1.1961,
+      "step": 5385
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.343311933835808e-05,
+      "loss": 1.2158,
+      "step": 5390
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3419441354862626e-05,
+      "loss": 1.2192,
+      "step": 5395
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.340575612312908e-05,
+      "loss": 1.1818,
+      "step": 5400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3392063672166213e-05,
+      "loss": 1.2125,
+      "step": 5405
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3378364030998113e-05,
+      "loss": 1.1822,
+      "step": 5410
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.33646572286641e-05,
+      "loss": 1.212,
+      "step": 5415
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3350943294218679e-05,
+      "loss": 1.2349,
+      "step": 5420
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3337222256731468e-05,
+      "loss": 1.234,
+      "step": 5425
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3323494145287144e-05,
+      "loss": 1.1979,
+      "step": 5430
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3309758988985379e-05,
+      "loss": 1.2097,
+      "step": 5435
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.329601681694078e-05,
+      "loss": 1.2107,
+      "step": 5440
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3282267658282815e-05,
+      "loss": 1.2127,
+      "step": 5445
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3268511542155778e-05,
+      "loss": 1.21,
+      "step": 5450
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3254748497718701e-05,
+      "loss": 1.2201,
+      "step": 5455
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3240978554145302e-05,
+      "loss": 1.1974,
+      "step": 5460
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3227201740623925e-05,
+      "loss": 1.1968,
+      "step": 5465
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3213418086357474e-05,
+      "loss": 1.2415,
+      "step": 5470
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3199627620563358e-05,
+      "loss": 1.2269,
+      "step": 5475
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3185830372473423e-05,
+      "loss": 1.225,
+      "step": 5480
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3172026371333889e-05,
+      "loss": 1.1938,
+      "step": 5485
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3158215646405287e-05,
+      "loss": 1.2142,
+      "step": 5490
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3144398226962416e-05,
+      "loss": 1.205,
+      "step": 5495
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3130574142294245e-05,
+      "loss": 1.2222,
+      "step": 5500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.311674342170389e-05,
+      "loss": 1.2527,
+      "step": 5505
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3102906094508519e-05,
+      "loss": 1.1925,
+      "step": 5510
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.308906219003931e-05,
+      "loss": 1.1779,
+      "step": 5515
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3075211737641385e-05,
+      "loss": 1.2298,
+      "step": 5520
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.306135476667374e-05,
+      "loss": 1.1898,
+      "step": 5525
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3047491306509196e-05,
+      "loss": 1.1648,
+      "step": 5530
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3033621386534323e-05,
+      "loss": 1.2179,
+      "step": 5535
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3019745036149387e-05,
+      "loss": 1.2535,
+      "step": 5540
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.300586228476828e-05,
+      "loss": 1.1626,
+      "step": 5545
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2991973161818472e-05,
+      "loss": 1.2298,
+      "step": 5550
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.297807769674093e-05,
+      "loss": 1.2106,
+      "step": 5555
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2964175918990067e-05,
+      "loss": 1.2126,
+      "step": 5560
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2950267858033676e-05,
+      "loss": 1.2308,
+      "step": 5565
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2936353543352876e-05,
+      "loss": 1.2084,
+      "step": 5570
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.292243300444203e-05,
+      "loss": 1.2145,
+      "step": 5575
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2908506270808703e-05,
+      "loss": 1.1958,
+      "step": 5580
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2894573371973587e-05,
+      "loss": 1.215,
+      "step": 5585
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2880634337470448e-05,
+      "loss": 1.1909,
+      "step": 5590
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2866689196846052e-05,
+      "loss": 1.2191,
+      "step": 5595
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.285273797966011e-05,
+      "loss": 1.2116,
+      "step": 5600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2838780715485215e-05,
+      "loss": 1.2007,
+      "step": 5605
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2824817433906774e-05,
+      "loss": 1.1645,
+      "step": 5610
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2810848164522957e-05,
+      "loss": 1.2255,
+      "step": 5615
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.279687293694462e-05,
+      "loss": 1.1885,
+      "step": 5620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2782891780795248e-05,
+      "loss": 1.2288,
+      "step": 5625
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2768904725710895e-05,
+      "loss": 1.2592,
+      "step": 5630
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.275491180134012e-05,
+      "loss": 1.2278,
+      "step": 5635
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.274091303734392e-05,
+      "loss": 1.2243,
+      "step": 5640
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.272690846339568e-05,
+      "loss": 1.1877,
+      "step": 5645
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2712898109181084e-05,
+      "loss": 1.2259,
+      "step": 5650
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.2698882004398083e-05,
+      "loss": 1.217,
+      "step": 5655
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.268486017875681e-05,
+      "loss": 1.2214,
+      "step": 5660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.2670832661979528e-05,
+      "loss": 1.2053,
+      "step": 5665
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.2656799483800558e-05,
+      "loss": 1.2355,
+      "step": 5670
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.2642760673966234e-05,
+      "loss": 1.1999,
+      "step": 5675
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.2628716262234813e-05,
+      "loss": 1.234,
+      "step": 5680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.2614666278376433e-05,
+      "loss": 1.257,
+      "step": 5685
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.2600610752173046e-05,
+      "loss": 1.2236,
+      "step": 5690
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.2586549713418348e-05,
+      "loss": 1.2325,
+      "step": 5695
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2572483191917719e-05,
+      "loss": 1.1781,
+      "step": 5700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2558411217488167e-05,
+      "loss": 1.2318,
+      "step": 5705
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2544333819958252e-05,
+      "loss": 1.2297,
+      "step": 5710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2530251029168032e-05,
+      "loss": 1.2306,
+      "step": 5715
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2516162874969002e-05,
+      "loss": 1.2306,
+      "step": 5720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2502069387224012e-05,
+      "loss": 1.2122,
+      "step": 5725
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2487970595807237e-05,
+      "loss": 1.2262,
+      "step": 5730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2473866530604077e-05,
+      "loss": 1.1918,
+      "step": 5735
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2459757221511125e-05,
+      "loss": 1.2169,
+      "step": 5740
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2445642698436075e-05,
+      "loss": 1.2145,
+      "step": 5745
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2431522991297683e-05,
+      "loss": 1.2204,
+      "step": 5750
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2417398130025694e-05,
+      "loss": 1.2344,
+      "step": 5755
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2403268144560773e-05,
+      "loss": 1.2057,
+      "step": 5760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2389133064854455e-05,
+      "loss": 1.2011,
+      "step": 5765
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2374992920869062e-05,
+      "loss": 1.2018,
+      "step": 5770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2360847742577658e-05,
+      "loss": 1.2136,
+      "step": 5775
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2346697559963974e-05,
+      "loss": 1.2266,
+      "step": 5780
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2332542403022363e-05,
+      "loss": 1.2201,
+      "step": 5785
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.23183823017577e-05,
+      "loss": 1.2362,
+      "step": 5790
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2304217286185351e-05,
+      "loss": 1.2057,
+      "step": 5795
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2290047386331108e-05,
+      "loss": 1.267,
+      "step": 5800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2275872632231098e-05,
+      "loss": 1.244,
+      "step": 5805
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2261693053931757e-05,
+      "loss": 1.2592,
+      "step": 5810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2247508681489732e-05,
+      "loss": 1.2501,
+      "step": 5815
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2233319544971836e-05,
+      "loss": 1.2519,
+      "step": 5820
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2219125674454986e-05,
+      "loss": 1.2571,
+      "step": 5825
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2204927100026125e-05,
+      "loss": 1.2122,
+      "step": 5830
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2190723851782176e-05,
+      "loss": 1.2303,
+      "step": 5835
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.217651595982996e-05,
+      "loss": 1.2295,
+      "step": 5840
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.216230345428615e-05,
+      "loss": 1.2525,
+      "step": 5845
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2148086365277197e-05,
+      "loss": 1.2364,
+      "step": 5850
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2133864722939257e-05,
+      "loss": 1.2606,
+      "step": 5855
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2119638557418155e-05,
+      "loss": 1.2489,
+      "step": 5860
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2105407898869283e-05,
+      "loss": 1.2264,
+      "step": 5865
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2091172777457583e-05,
+      "loss": 1.2541,
+      "step": 5870
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2076933223357433e-05,
+      "loss": 1.2275,
+      "step": 5875
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2062689266752622e-05,
+      "loss": 1.2687,
+      "step": 5880
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2048440937836264e-05,
+      "loss": 1.2548,
+      "step": 5885
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2034188266810736e-05,
+      "loss": 1.1997,
+      "step": 5890
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2019931283887635e-05,
+      "loss": 1.2176,
+      "step": 5895
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.200567001928769e-05,
+      "loss": 1.2211,
+      "step": 5900
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.19914045032407e-05,
+      "loss": 1.2673,
+      "step": 5905
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1977134765985481e-05,
+      "loss": 1.2317,
+      "step": 5910
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1962860837769794e-05,
+      "loss": 1.2284,
+      "step": 5915
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.194858274885029e-05,
+      "loss": 1.2226,
+      "step": 5920
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.193430052949244e-05,
+      "loss": 1.2361,
+      "step": 5925
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1920014209970456e-05,
+      "loss": 1.1998,
+      "step": 5930
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1905723820567257e-05,
+      "loss": 1.2052,
+      "step": 5935
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.189142939157438e-05,
+      "loss": 1.2024,
+      "step": 5940
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.1877130953291932e-05,
+      "loss": 1.231,
+      "step": 5945
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.1862828536028511e-05,
+      "loss": 1.2317,
+      "step": 5950
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.1848522170101155e-05,
+      "loss": 1.215,
+      "step": 5955
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.1834211885835267e-05,
+      "loss": 1.2352,
+      "step": 5960
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.1819897713564562e-05,
+      "loss": 1.2031,
+      "step": 5965
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.180557968363099e-05,
+      "loss": 1.2463,
+      "step": 5970
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.1791257826384682e-05,
+      "loss": 1.2238,
+      "step": 5975
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.1776932172183882e-05,
+      "loss": 1.2095,
+      "step": 5980
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.1762602751394878e-05,
+      "loss": 1.199,
+      "step": 5985
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.174826959439195e-05,
+      "loss": 1.2301,
+      "step": 5990
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.1733932731557292e-05,
+      "loss": 1.2277,
+      "step": 5995
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.171959219328095e-05,
+      "loss": 1.2206,
+      "step": 6000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.1705248009960773e-05,
+      "loss": 1.2307,
+      "step": 6005
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.1690900212002323e-05,
+      "loss": 1.2399,
+      "step": 6010
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.1676548829818831e-05,
+      "loss": 1.2248,
+      "step": 6015
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1662193893831129e-05,
+      "loss": 1.2077,
+      "step": 6020
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1647835434467569e-05,
+      "loss": 1.2343,
+      "step": 6025
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1633473482163982e-05,
+      "loss": 1.2142,
+      "step": 6030
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1619108067363608e-05,
+      "loss": 1.1984,
+      "step": 6035
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1604739220517012e-05,
+      "loss": 1.2062,
+      "step": 6040
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1590366972082046e-05,
+      "loss": 1.2442,
+      "step": 6045
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1575991352523765e-05,
+      "loss": 1.1937,
+      "step": 6050
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1561612392314374e-05,
+      "loss": 1.2324,
+      "step": 6055
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1547230121933153e-05,
+      "loss": 1.1834,
+      "step": 6060
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1532844571866415e-05,
+      "loss": 1.2164,
+      "step": 6065
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1518455772607401e-05,
+      "loss": 1.2182,
+      "step": 6070
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.150406375465626e-05,
+      "loss": 1.2061,
+      "step": 6075
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1489668548519955e-05,
+      "loss": 1.2228,
+      "step": 6080
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.14752701847122e-05,
+      "loss": 1.2176,
+      "step": 6085
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.146086869375342e-05,
+      "loss": 1.2637,
+      "step": 6090
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.1446464106170651e-05,
+      "loss": 1.2652,
+      "step": 6095
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.1432056452497504e-05,
+      "loss": 1.2238,
+      "step": 6100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.1417645763274088e-05,
+      "loss": 1.1944,
+      "step": 6105
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.140323206904694e-05,
+      "loss": 1.2122,
+      "step": 6110
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.1388815400368972e-05,
+      "loss": 1.264,
+      "step": 6115
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.1374395787799402e-05,
+      "loss": 1.2269,
+      "step": 6120
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.1359973261903682e-05,
+      "loss": 1.2083,
+      "step": 6125
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.1345547853253449e-05,
+      "loss": 1.2437,
+      "step": 6130
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.1331119592426445e-05,
+      "loss": 1.268,
+      "step": 6135
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.1316688510006452e-05,
+      "loss": 1.1825,
+      "step": 6140
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.1302254636583245e-05,
+      "loss": 1.1877,
+      "step": 6145
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.1287818002752508e-05,
+      "loss": 1.2309,
+      "step": 6150
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.1273378639115777e-05,
+      "loss": 1.1911,
+      "step": 6155
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.1258936576280375e-05,
+      "loss": 1.1993,
+      "step": 6160
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.1244491844859346e-05,
+      "loss": 1.2022,
+      "step": 6165
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.1230044475471389e-05,
+      "loss": 1.1643,
+      "step": 6170
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.12155944987408e-05,
+      "loss": 1.1865,
+      "step": 6175
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.1201141945297398e-05,
+      "loss": 1.1736,
+      "step": 6180
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.1186686845776466e-05,
+      "loss": 1.2263,
+      "step": 6185
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.117222923081868e-05,
+      "loss": 1.2192,
+      "step": 6190
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.1157769131070046e-05,
+      "loss": 1.2007,
+      "step": 6195
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.1143306577181849e-05,
+      "loss": 1.2362,
+      "step": 6200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.1128841599810567e-05,
+      "loss": 1.224,
+      "step": 6205
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.1114374229617817e-05,
+      "loss": 1.1885,
+      "step": 6210
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.1099904497270285e-05,
+      "loss": 1.2043,
+      "step": 6215
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.1085432433439666e-05,
+      "loss": 1.1884,
+      "step": 6220
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.10709580688026e-05,
+      "loss": 1.2164,
+      "step": 6225
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.1056481434040605e-05,
+      "loss": 1.2379,
+      "step": 6230
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.1042002559840003e-05,
+      "loss": 1.2351,
+      "step": 6235
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.1027521476891872e-05,
+      "loss": 1.2182,
+      "step": 6240
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.1013038215891964e-05,
+      "loss": 1.2553,
+      "step": 6245
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0998552807540657e-05,
+      "loss": 1.2363,
+      "step": 6250
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0984065282542869e-05,
+      "loss": 1.246,
+      "step": 6255
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0969575671608015e-05,
+      "loss": 1.1822,
+      "step": 6260
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0955084005449928e-05,
+      "loss": 1.1829,
+      "step": 6265
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0940590314786797e-05,
+      "loss": 1.2114,
+      "step": 6270
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.09260946303411e-05,
+      "loss": 1.2439,
+      "step": 6275
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0911596982839547e-05,
+      "loss": 1.2342,
+      "step": 6280
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0897097403013006e-05,
+      "loss": 1.2245,
+      "step": 6285
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0882595921596439e-05,
+      "loss": 1.1962,
+      "step": 6290
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.086809256932884e-05,
+      "loss": 1.2116,
+      "step": 6295
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.0853587376953173e-05,
+      "loss": 1.2159,
+      "step": 6300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.08390803752163e-05,
+      "loss": 1.2285,
+      "step": 6305
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.0824571594868912e-05,
+      "loss": 1.2596,
+      "step": 6310
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.0810061066665476e-05,
+      "loss": 1.2369,
+      "step": 6315
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.0795548821364168e-05,
+      "loss": 1.2217,
+      "step": 6320
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.0781034889726796e-05,
+      "loss": 1.2177,
+      "step": 6325
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.0766519302518747e-05,
+      "loss": 1.2042,
+      "step": 6330
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.0752002090508911e-05,
+      "loss": 1.2437,
+      "step": 6335
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.0737483284469634e-05,
+      "loss": 1.2252,
+      "step": 6340
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.0722962915176634e-05,
+      "loss": 1.209,
+      "step": 6345
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.070844101340894e-05,
+      "loss": 1.2108,
+      "step": 6350
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.069391760994883e-05,
+      "loss": 1.1724,
+      "step": 6355
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.0679392735581771e-05,
+      "loss": 1.2203,
+      "step": 6360
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.0664866421096338e-05,
+      "loss": 1.1942,
+      "step": 6365
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.0650338697284177e-05,
+      "loss": 1.171,
+      "step": 6370
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.0635809594939898e-05,
+      "loss": 1.2633,
+      "step": 6375
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.0621279144861047e-05,
+      "loss": 1.162,
+      "step": 6380
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.0606747377848028e-05,
+      "loss": 1.1969,
+      "step": 6385
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.0592214324704027e-05,
+      "loss": 1.2075,
+      "step": 6390
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.0577680016234966e-05,
+      "loss": 1.1996,
+      "step": 6395
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.0563144483249421e-05,
+      "loss": 1.1869,
+      "step": 6400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.0548607756558572e-05,
+      "loss": 1.1898,
+      "step": 6405
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.0534069866976113e-05,
+      "loss": 1.1758,
+      "step": 6410
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.0519530845318224e-05,
+      "loss": 1.2532,
+      "step": 6415
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.050499072240347e-05,
+      "loss": 1.2316,
+      "step": 6420
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.0490449529052755e-05,
+      "loss": 1.1916,
+      "step": 6425
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.0475907296089252e-05,
+      "loss": 1.214,
+      "step": 6430
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.0461364054338339e-05,
+      "loss": 1.1996,
+      "step": 6435
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.0446819834627526e-05,
+      "loss": 1.1973,
+      "step": 6440
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.0432274667786409e-05,
+      "loss": 1.2462,
+      "step": 6445
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.0417728584646574e-05,
+      "loss": 1.1982,
+      "step": 6450
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.0403181616041564e-05,
+      "loss": 1.2201,
+      "step": 6455
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.0388633792806792e-05,
+      "loss": 1.193,
+      "step": 6460
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.0374085145779486e-05,
+      "loss": 1.2079,
+      "step": 6465
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.035953570579862e-05,
+      "loss": 1.2354,
+      "step": 6470
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.0344985503704841e-05,
+      "loss": 1.2036,
+      "step": 6475
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.0330434570340423e-05,
+      "loss": 1.2018,
+      "step": 6480
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.0315882936549181e-05,
+      "loss": 1.2133,
+      "step": 6485
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.0301330633176425e-05,
+      "loss": 1.229,
+      "step": 6490
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.028677769106887e-05,
+      "loss": 1.1936,
+      "step": 6495
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.0272224141074596e-05,
+      "loss": 1.2142,
+      "step": 6500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.0257670014042969e-05,
+      "loss": 1.249,
+      "step": 6505
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.0243115340824577e-05,
+      "loss": 1.1975,
+      "step": 6510
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.0228560152271167e-05,
+      "loss": 1.2481,
+      "step": 6515
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.0214004479235578e-05,
+      "loss": 1.2024,
+      "step": 6520
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.0199448352571673e-05,
+      "loss": 1.199,
+      "step": 6525
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.0184891803134277e-05,
+      "loss": 1.2229,
+      "step": 6530
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.0170334861779123e-05,
+      "loss": 1.1969,
+      "step": 6535
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.0155777559362754e-05,
+      "loss": 1.1906,
+      "step": 6540
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.0141219926742496e-05,
+      "loss": 1.227,
+      "step": 6545
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.0126661994776365e-05,
+      "loss": 1.2076,
+      "step": 6550
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.0112103794323018e-05,
+      "loss": 1.2424,
+      "step": 6555
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.0097545356241676e-05,
+      "loss": 1.1896,
+      "step": 6560
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.0082986711392064e-05,
+      "loss": 1.2207,
+      "step": 6565
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.0068427890634352e-05,
+      "loss": 1.2317,
+      "step": 6570
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.005386892482907e-05,
+      "loss": 1.181,
+      "step": 6575
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.003930984483707e-05,
+      "loss": 1.1798,
+      "step": 6580
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.0024750681519442e-05,
+      "loss": 1.1891,
+      "step": 6585
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.0010191465737437e-05,
+      "loss": 1.2098,
+      "step": 6590
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.995632228352445e-06,
+      "loss": 1.1901,
+      "step": 6595
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.981073000225873e-06,
+      "loss": 1.2055,
+      "step": 6600
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.966513812219135e-06,
+      "loss": 1.1969,
+      "step": 6605
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.95195469519354e-06,
+      "loss": 1.2149,
+      "step": 6610
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.937395680010256e-06,
+      "loss": 1.2011,
+      "step": 6615
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.922836797530234e-06,
+      "loss": 1.2269,
+      "step": 6620
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.908278078614139e-06,
+      "loss": 1.2265,
+      "step": 6625
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.893719554122298e-06,
+      "loss": 1.2074,
+      "step": 6630
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.879161254914615e-06,
+      "loss": 1.2202,
+      "step": 6635
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.864603211850526e-06,
+      "loss": 1.2004,
+      "step": 6640
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.85004545578892e-06,
+      "loss": 1.2033,
+      "step": 6645
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.835488017588078e-06,
+      "loss": 1.2044,
+      "step": 6650
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.820930928105603e-06,
+      "loss": 1.211,
+      "step": 6655
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.80637421819837e-06,
+      "loss": 1.2072,
+      "step": 6660
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.791817918722438e-06,
+      "loss": 1.1979,
+      "step": 6665
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.777262060533003e-06,
+      "loss": 1.1651,
+      "step": 6670
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.762706674484322e-06,
+      "loss": 1.2283,
+      "step": 6675
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.748151791429651e-06,
+      "loss": 1.2109,
+      "step": 6680
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.733597442221182e-06,
+      "loss": 1.2074,
+      "step": 6685
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.719043657709973e-06,
+      "loss": 1.2098,
+      "step": 6690
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.70449046874589e-06,
+      "loss": 1.2382,
+      "step": 6695
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.689937906177527e-06,
+      "loss": 1.2357,
+      "step": 6700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.675386000852165e-06,
+      "loss": 1.2269,
+      "step": 6705
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.660834783615674e-06,
+      "loss": 1.2063,
+      "step": 6710
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.646284285312475e-06,
+      "loss": 1.21,
+      "step": 6715
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.631734536785476e-06,
+      "loss": 1.192,
+      "step": 6720
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.617185568875971e-06,
+      "loss": 1.1985,
+      "step": 6725
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.60263741242362e-06,
+      "loss": 1.1954,
+      "step": 6730
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.588090098266354e-06,
+      "loss": 1.2143,
+      "step": 6735
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.573543657240318e-06,
+      "loss": 1.2209,
+      "step": 6740
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.558998120179812e-06,
+      "loss": 1.2339,
+      "step": 6745
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.544453517917214e-06,
+      "loss": 1.192,
+      "step": 6750
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.529909881282922e-06,
+      "loss": 1.186,
+      "step": 6755
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.515367241105288e-06,
+      "loss": 1.2202,
+      "step": 6760
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.500825628210551e-06,
+      "loss": 1.2257,
+      "step": 6765
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.486285073422774e-06,
+      "loss": 1.198,
+      "step": 6770
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.471745607563778e-06,
+      "loss": 1.2022,
+      "step": 6775
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.457207261453073e-06,
+      "loss": 1.162,
+      "step": 6780
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.442670065907794e-06,
+      "loss": 1.1896,
+      "step": 6785
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.428134051742644e-06,
+      "loss": 1.1667,
+      "step": 6790
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.413599249769814e-06,
+      "loss": 1.2223,
+      "step": 6795
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.39906569079893e-06,
+      "loss": 1.1757,
+      "step": 6800
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.38453340563698e-06,
+      "loss": 1.2003,
+      "step": 6805
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.370002425088257e-06,
+      "loss": 1.2075,
+      "step": 6810
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.355472779954283e-06,
+      "loss": 1.1785,
+      "step": 6815
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.340944501033754e-06,
+      "loss": 1.1796,
+      "step": 6820
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.326417619122464e-06,
+      "loss": 1.2092,
+      "step": 6825
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.311892165013253e-06,
+      "loss": 1.2006,
+      "step": 6830
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.297368169495932e-06,
+      "loss": 1.2492,
+      "step": 6835
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.282845663357219e-06,
+      "loss": 1.2004,
+      "step": 6840
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.268324677380674e-06,
+      "loss": 1.2288,
+      "step": 6845
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.253805242346633e-06,
+      "loss": 1.2611,
+      "step": 6850
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.23928738903215e-06,
+      "loss": 1.1787,
+      "step": 6855
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.224771148210927e-06,
+      "loss": 1.2036,
+      "step": 6860
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.210256550653238e-06,
+      "loss": 1.2009,
+      "step": 6865
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.195743627125888e-06,
+      "loss": 1.2028,
+      "step": 6870
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.181232408392118e-06,
+      "loss": 1.1931,
+      "step": 6875
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.166722925211562e-06,
+      "loss": 1.2075,
+      "step": 6880
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.152215208340187e-06,
+      "loss": 1.2256,
+      "step": 6885
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.137709288530196e-06,
+      "loss": 1.1985,
+      "step": 6890
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.123205196529997e-06,
+      "loss": 1.2383,
+      "step": 6895
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.108702963084113e-06,
+      "loss": 1.1923,
+      "step": 6900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.094202618933138e-06,
+      "loss": 1.2249,
+      "step": 6905
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.079704194813656e-06,
+      "loss": 1.2201,
+      "step": 6910
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.06520772145818e-06,
+      "loss": 1.1944,
+      "step": 6915
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.050713229595087e-06,
+      "loss": 1.1923,
+      "step": 6920
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.036220749948558e-06,
+      "loss": 1.1828,
+      "step": 6925
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.021730313238506e-06,
+      "loss": 1.2321,
+      "step": 6930
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.007241950180511e-06,
+      "loss": 1.2126,
+      "step": 6935
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.992755691485767e-06,
+      "loss": 1.2093,
+      "step": 6940
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.978271567860997e-06,
+      "loss": 1.1877,
+      "step": 6945
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.963789610008406e-06,
+      "loss": 1.2082,
+      "step": 6950
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.949309848625598e-06,
+      "loss": 1.2163,
+      "step": 6955
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.934832314405537e-06,
+      "loss": 1.1988,
+      "step": 6960
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.920357038036447e-06,
+      "loss": 1.2413,
+      "step": 6965
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.90588405020178e-06,
+      "loss": 1.1834,
+      "step": 6970
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.891413381580131e-06,
+      "loss": 1.2152,
+      "step": 6975
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.876945062845182e-06,
+      "loss": 1.1847,
+      "step": 6980
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.862479124665634e-06,
+      "loss": 1.2041,
+      "step": 6985
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.848015597705133e-06,
+      "loss": 1.1963,
+      "step": 6990
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.83355451262223e-06,
+      "loss": 1.1942,
+      "step": 6995
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.819095900070286e-06,
+      "loss": 1.1791,
+      "step": 7000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.804639790697432e-06,
+      "loss": 1.184,
+      "step": 7005
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.79018621514648e-06,
+      "loss": 1.1875,
+      "step": 7010
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.77573520405488e-06,
+      "loss": 1.2182,
+      "step": 7015
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.761286788054643e-06,
+      "loss": 1.2175,
+      "step": 7020
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.746840997772286e-06,
+      "loss": 1.1648,
+      "step": 7025
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.73239786382875e-06,
+      "loss": 1.2407,
+      "step": 7030
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.71795741683935e-06,
+      "loss": 1.219,
+      "step": 7035
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.703519687413704e-06,
+      "loss": 1.2323,
+      "step": 7040
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.68908470615567e-06,
+      "loss": 1.2143,
+      "step": 7045
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.67465250366329e-06,
+      "loss": 1.1867,
+      "step": 7050
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.6602231105287e-06,
+      "loss": 1.2237,
+      "step": 7055
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.645796557338088e-06,
+      "loss": 1.1962,
+      "step": 7060
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.631372874671624e-06,
+      "loss": 1.2004,
+      "step": 7065
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.616952093103393e-06,
+      "loss": 1.1985,
+      "step": 7070
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.60253424320133e-06,
+      "loss": 1.2212,
+      "step": 7075
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.588119355527148e-06,
+      "loss": 1.1748,
+      "step": 7080
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.573707460636296e-06,
+      "loss": 1.2085,
+      "step": 7085
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.559298589077866e-06,
+      "loss": 1.1849,
+      "step": 7090
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.54489277139455e-06,
+      "loss": 1.1957,
+      "step": 7095
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.53049003812256e-06,
+      "loss": 1.2436,
+      "step": 7100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.516090419791569e-06,
+      "loss": 1.2056,
+      "step": 7105
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.501693946924662e-06,
+      "loss": 1.2326,
+      "step": 7110
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.487300650038238e-06,
+      "loss": 1.1899,
+      "step": 7115
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.472910559641975e-06,
+      "loss": 1.1745,
+      "step": 7120
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.45852370623875e-06,
+      "loss": 1.1973,
+      "step": 7125
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.444140120324575e-06,
+      "loss": 1.2242,
+      "step": 7130
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.429759832388545e-06,
+      "loss": 1.1936,
+      "step": 7135
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.415382872912758e-06,
+      "loss": 1.1897,
+      "step": 7140
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.401009272372256e-06,
+      "loss": 1.1977,
+      "step": 7145
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.386639061234967e-06,
+      "loss": 1.2119,
+      "step": 7150
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.372272269961626e-06,
+      "loss": 1.1745,
+      "step": 7155
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.35790892900572e-06,
+      "loss": 1.226,
+      "step": 7160
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.343549068813437e-06,
+      "loss": 1.2171,
+      "step": 7165
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.329192719823569e-06,
+      "loss": 1.2331,
+      "step": 7170
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.314839912467468e-06,
+      "loss": 1.1704,
+      "step": 7175
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.300490677168986e-06,
+      "loss": 1.2284,
+      "step": 7180
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.2861450443444e-06,
+      "loss": 1.177,
+      "step": 7185
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.27180304440235e-06,
+      "loss": 1.1817,
+      "step": 7190
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.257464707743778e-06,
+      "loss": 1.1838,
+      "step": 7195
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.243130064761852e-06,
+      "loss": 1.2397,
+      "step": 7200
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.228799145841922e-06,
+      "loss": 1.1763,
+      "step": 7205
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.214471981361436e-06,
+      "loss": 1.2055,
+      "step": 7210
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.200148601689887e-06,
+      "loss": 1.1789,
+      "step": 7215
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.185829037188751e-06,
+      "loss": 1.2063,
+      "step": 7220
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.171513318211403e-06,
+      "loss": 1.2263,
+      "step": 7225
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.15720147510308e-06,
+      "loss": 1.2081,
+      "step": 7230
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.142893538200796e-06,
+      "loss": 1.238,
+      "step": 7235
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.128589537833289e-06,
+      "loss": 1.1645,
+      "step": 7240
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.114289504320946e-06,
+      "loss": 1.1719,
+      "step": 7245
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.099993467975752e-06,
+      "loss": 1.2348,
+      "step": 7250
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.085701459101216e-06,
+      "loss": 1.1812,
+      "step": 7255
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.071413507992312e-06,
+      "loss": 1.2247,
+      "step": 7260
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.057129644935411e-06,
+      "loss": 1.2152,
+      "step": 7265
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.04284990020822e-06,
+      "loss": 1.1798,
+      "step": 7270
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.028574304079716e-06,
+      "loss": 1.2312,
+      "step": 7275
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.014302886810078e-06,
+      "loss": 1.1956,
+      "step": 7280
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.000035678650638e-06,
+      "loss": 1.1897,
+      "step": 7285
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.985772709843789e-06,
+      "loss": 1.1886,
+      "step": 7290
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.971514010622953e-06,
+      "loss": 1.1841,
+      "step": 7295
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.957259611212495e-06,
+      "loss": 1.1941,
+      "step": 7300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.943009541827667e-06,
+      "loss": 1.215,
+      "step": 7305
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.928763832674541e-06,
+      "loss": 1.2149,
+      "step": 7310
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.91452251394995e-06,
+      "loss": 1.223,
+      "step": 7315
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.900285615841415e-06,
+      "loss": 1.1619,
+      "step": 7320
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.886053168527085e-06,
+      "loss": 1.1972,
+      "step": 7325
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.871825202175695e-06,
+      "loss": 1.2099,
+      "step": 7330
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.857601746946455e-06,
+      "loss": 1.2023,
+      "step": 7335
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.843382832989023e-06,
+      "loss": 1.1891,
+      "step": 7340
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.829168490443435e-06,
+      "loss": 1.2149,
+      "step": 7345
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.814958749440034e-06,
+      "loss": 1.2554,
+      "step": 7350
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.800753640099408e-06,
+      "loss": 1.1641,
+      "step": 7355
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.786553192532326e-06,
+      "loss": 1.1868,
+      "step": 7360
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.772357436839678e-06,
+      "loss": 1.2185,
+      "step": 7365
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.758166403112409e-06,
+      "loss": 1.1744,
+      "step": 7370
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.743980121431449e-06,
+      "loss": 1.2052,
+      "step": 7375
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.729798621867662e-06,
+      "loss": 1.1814,
+      "step": 7380
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.715621934481776e-06,
+      "loss": 1.1882,
+      "step": 7385
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.701450089324312e-06,
+      "loss": 1.1744,
+      "step": 7390
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.687283116435531e-06,
+      "loss": 1.1682,
+      "step": 7395
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.673121045845367e-06,
+      "loss": 1.1881,
+      "step": 7400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.65896390757336e-06,
+      "loss": 1.1725,
+      "step": 7405
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.644811731628591e-06,
+      "loss": 1.2273,
+      "step": 7410
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.630664548009634e-06,
+      "loss": 1.2153,
+      "step": 7415
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.616522386704469e-06,
+      "loss": 1.2349,
+      "step": 7420
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.602385277690437e-06,
+      "loss": 1.2027,
+      "step": 7425
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.5882532509341675e-06,
+      "loss": 1.1793,
+      "step": 7430
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.574126336391514e-06,
+      "loss": 1.2009,
+      "step": 7435
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.560004564007502e-06,
+      "loss": 1.1525,
+      "step": 7440
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.545887963716248e-06,
+      "loss": 1.215,
+      "step": 7445
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.531776565440914e-06,
+      "loss": 1.2096,
+      "step": 7450
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.517670399093622e-06,
+      "loss": 1.2314,
+      "step": 7455
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.503569494575417e-06,
+      "loss": 1.1918,
+      "step": 7460
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.489473881776183e-06,
+      "loss": 1.2383,
+      "step": 7465
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.475383590574592e-06,
+      "loss": 1.1914,
+      "step": 7470
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.461298650838029e-06,
+      "loss": 1.1904,
+      "step": 7475
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.4472190924225465e-06,
+      "loss": 1.1907,
+      "step": 7480
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.433144945172777e-06,
+      "loss": 1.1948,
+      "step": 7485
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.41907623892189e-06,
+      "loss": 1.1836,
+      "step": 7490
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.405013003491518e-06,
+      "loss": 1.2358,
+      "step": 7495
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.3909552686917066e-06,
+      "loss": 1.2118,
+      "step": 7500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.37690306432083e-06,
+      "loss": 1.226,
+      "step": 7505
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.362856420165548e-06,
+      "loss": 1.254,
+      "step": 7510
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.348815366000726e-06,
+      "loss": 1.1963,
+      "step": 7515
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.334779931589384e-06,
+      "loss": 1.2039,
+      "step": 7520
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.320750146682638e-06,
+      "loss": 1.2323,
+      "step": 7525
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.306726041019613e-06,
+      "loss": 1.1836,
+      "step": 7530
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.292707644327406e-06,
+      "loss": 1.1918,
+      "step": 7535
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.278694986321011e-06,
+      "loss": 1.2147,
+      "step": 7540
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.264688096703256e-06,
+      "loss": 1.1997,
+      "step": 7545
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.250687005164743e-06,
+      "loss": 1.1924,
+      "step": 7550
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.236691741383783e-06,
+      "loss": 1.1877,
+      "step": 7555
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.222702335026337e-06,
+      "loss": 1.1737,
+      "step": 7560
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.208718815745945e-06,
+      "loss": 1.1912,
+      "step": 7565
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.194741213183672e-06,
+      "loss": 1.1553,
+      "step": 7570
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.1807695569680325e-06,
+      "loss": 1.2434,
+      "step": 7575
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.1668038767149515e-06,
+      "loss": 1.2464,
+      "step": 7580
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.152844202027673e-06,
+      "loss": 1.2227,
+      "step": 7585
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.138890562496721e-06,
+      "loss": 1.2013,
+      "step": 7590
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.12494298769982e-06,
+      "loss": 1.2091,
+      "step": 7595
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.111001507201839e-06,
+      "loss": 1.1877,
+      "step": 7600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.097066150554729e-06,
+      "loss": 1.1683,
+      "step": 7605
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.083136947297471e-06,
+      "loss": 1.2056,
+      "step": 7610
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.069213926955989e-06,
+      "loss": 1.2276,
+      "step": 7615
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.055297119043105e-06,
+      "loss": 1.2392,
+      "step": 7620
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.041386553058474e-06,
+      "loss": 1.1601,
+      "step": 7625
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.027482258488516e-06,
+      "loss": 1.1857,
+      "step": 7630
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.013584264806366e-06,
+      "loss": 1.2088,
+      "step": 7635
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.999692601471795e-06,
+      "loss": 1.2398,
+      "step": 7640
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.985807297931155e-06,
+      "loss": 1.176,
+      "step": 7645
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.971928383617321e-06,
+      "loss": 1.2103,
+      "step": 7650
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.9580558879496265e-06,
+      "loss": 1.1732,
+      "step": 7655
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.944189840333792e-06,
+      "loss": 1.1833,
+      "step": 7660
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.930330270161878e-06,
+      "loss": 1.2134,
+      "step": 7665
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.91647720681221e-06,
+      "loss": 1.1887,
+      "step": 7670
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.902630679649322e-06,
+      "loss": 1.1708,
+      "step": 7675
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.888790718023892e-06,
+      "loss": 1.1898,
+      "step": 7680
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.874957351272684e-06,
+      "loss": 1.1996,
+      "step": 7685
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.861130608718478e-06,
+      "loss": 1.1972,
+      "step": 7690
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.847310519670018e-06,
+      "loss": 1.1998,
+      "step": 7695
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.833497113421938e-06,
+      "loss": 1.2253,
+      "step": 7700
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.819690419254714e-06,
+      "loss": 1.2025,
+      "step": 7705
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.805890466434588e-06,
+      "loss": 1.1773,
+      "step": 7710
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.792097284213515e-06,
+      "loss": 1.166,
+      "step": 7715
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.778310901829099e-06,
+      "loss": 1.1618,
+      "step": 7720
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.764531348504531e-06,
+      "loss": 1.2067,
+      "step": 7725
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.750758653448524e-06,
+      "loss": 1.1873,
+      "step": 7730
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.73699284585525e-06,
+      "loss": 1.1972,
+      "step": 7735
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.723233954904289e-06,
+      "loss": 1.1999,
+      "step": 7740
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.709482009760555e-06,
+      "loss": 1.1828,
+      "step": 7745
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.695737039574241e-06,
+      "loss": 1.1715,
+      "step": 7750
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.681999073480756e-06,
+      "loss": 1.1678,
+      "step": 7755
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.668268140600659e-06,
+      "loss": 1.2098,
+      "step": 7760
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.6545442700396e-06,
+      "loss": 1.1956,
+      "step": 7765
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.64082749088826e-06,
+      "loss": 1.1618,
+      "step": 7770
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.627117832222297e-06,
+      "loss": 1.219,
+      "step": 7775
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.613415323102262e-06,
+      "loss": 1.1813,
+      "step": 7780
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.599719992573559e-06,
+      "loss": 1.2043,
+      "step": 7785
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.586031869666371e-06,
+      "loss": 1.2182,
+      "step": 7790
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.572350983395608e-06,
+      "loss": 1.2232,
+      "step": 7795
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.5586773627608366e-06,
+      "loss": 1.1888,
+      "step": 7800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.545011036746226e-06,
+      "loss": 1.184,
+      "step": 7805
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.531352034320475e-06,
+      "loss": 1.1932,
+      "step": 7810
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.517700384436767e-06,
+      "loss": 1.2017,
+      "step": 7815
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.504056116032698e-06,
+      "loss": 1.2025,
+      "step": 7820
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.490419258030212e-06,
+      "loss": 1.1919,
+      "step": 7825
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.476789839335551e-06,
+      "loss": 1.2148,
+      "step": 7830
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.46316788883919e-06,
+      "loss": 1.1688,
+      "step": 7835
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.449553435415768e-06,
+      "loss": 1.1924,
+      "step": 7840
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.4359465079240315e-06,
+      "loss": 1.1657,
+      "step": 7845
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.422347135206779e-06,
+      "loss": 1.206,
+      "step": 7850
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.4087553460907926e-06,
+      "loss": 1.1948,
+      "step": 7855
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.395171169386778e-06,
+      "loss": 1.2175,
+      "step": 7860
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.381594633889306e-06,
+      "loss": 1.1656,
+      "step": 7865
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.368025768376754e-06,
+      "loss": 1.1889,
+      "step": 7870
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.354464601611233e-06,
+      "loss": 1.2039,
+      "step": 7875
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.340911162338546e-06,
+      "loss": 1.205,
+      "step": 7880
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.3273654792880975e-06,
+      "loss": 1.1897,
+      "step": 7885
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.3138275811728765e-06,
+      "loss": 1.1711,
+      "step": 7890
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.3002974966893525e-06,
+      "loss": 1.2157,
+      "step": 7895
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.286775254517433e-06,
+      "loss": 1.2382,
+      "step": 7900
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.273260883320409e-06,
+      "loss": 1.2144,
+      "step": 7905
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.2597544117448804e-06,
+      "loss": 1.1768,
+      "step": 7910
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.24625586842071e-06,
+      "loss": 1.19,
+      "step": 7915
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.232765281960947e-06,
+      "loss": 1.1892,
+      "step": 7920
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.219282680961782e-06,
+      "loss": 1.2026,
+      "step": 7925
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.205808094002469e-06,
+      "loss": 1.1891,
+      "step": 7930
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.192341549645283e-06,
+      "loss": 1.2035,
+      "step": 7935
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.178883076435447e-06,
+      "loss": 1.1731,
+      "step": 7940
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.165432702901079e-06,
+      "loss": 1.2171,
+      "step": 7945
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.151990457553125e-06,
+      "loss": 1.1853,
+      "step": 7950
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.1385563688853e-06,
+      "loss": 1.1638,
+      "step": 7955
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.125130465374034e-06,
+      "loss": 1.1707,
+      "step": 7960
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.111712775478402e-06,
+      "loss": 1.1835,
+      "step": 7965
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.098303327640075e-06,
+      "loss": 1.1866,
+      "step": 7970
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.084902150283243e-06,
+      "loss": 1.2072,
+      "step": 7975
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.071509271814573e-06,
+      "loss": 1.2018,
+      "step": 7980
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.058124720623137e-06,
+      "loss": 1.1941,
+      "step": 7985
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.044748525080359e-06,
+      "loss": 1.1463,
+      "step": 7990
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 1.199765920639038,
+      "eval_runtime": 1581.4767,
+      "eval_samples_per_second": 17.899,
+      "eval_steps_per_second": 1.119,
+      "step": 7993
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.031380713539949e-06,
+      "loss": 1.1697,
+      "step": 7995
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.018021314337847e-06,
+      "loss": 1.1959,
+      "step": 8000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.004670355792159e-06,
+      "loss": 1.197,
+      "step": 8005
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.9913278662031005e-06,
+      "loss": 1.1953,
+      "step": 8010
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.977993873852935e-06,
+      "loss": 1.164,
+      "step": 8015
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.964668407005913e-06,
+      "loss": 1.1513,
+      "step": 8020
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.951351493908215e-06,
+      "loss": 1.1789,
+      "step": 8025
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.938043162787891e-06,
+      "loss": 1.1676,
+      "step": 8030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.9247434418547966e-06,
+      "loss": 1.2024,
+      "step": 8035
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.911452359300541e-06,
+      "loss": 1.159,
+      "step": 8040
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.898169943298415e-06,
+      "loss": 1.1962,
+      "step": 8045
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.884896222003343e-06,
+      "loss": 1.1911,
+      "step": 8050
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.8716312235518234e-06,
+      "loss": 1.1425,
+      "step": 8055
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.858374976061863e-06,
+      "loss": 1.1781,
+      "step": 8060
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.845127507632908e-06,
+      "loss": 1.1479,
+      "step": 8065
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.831888846345809e-06,
+      "loss": 1.1595,
+      "step": 8070
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.8186590202627495e-06,
+      "loss": 1.2249,
+      "step": 8075
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.805438057427166e-06,
+      "loss": 1.1899,
+      "step": 8080
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.792225985863728e-06,
+      "loss": 1.1675,
+      "step": 8085
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.7790228335782476e-06,
+      "loss": 1.192,
+      "step": 8090
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.765828628557632e-06,
+      "loss": 1.1825,
+      "step": 8095
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.7526433987698275e-06,
+      "loss": 1.1564,
+      "step": 8100
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.739467172163744e-06,
+      "loss": 1.1637,
+      "step": 8105
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.726299976669225e-06,
+      "loss": 1.1615,
+      "step": 8110
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.713141840196956e-06,
+      "loss": 1.1871,
+      "step": 8115
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.699992790638429e-06,
+      "loss": 1.1879,
+      "step": 8120
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.686852855865862e-06,
+      "loss": 1.1675,
+      "step": 8125
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.673722063732163e-06,
+      "loss": 1.1688,
+      "step": 8130
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.660600442070858e-06,
+      "loss": 1.1638,
+      "step": 8135
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.647488018696034e-06,
+      "loss": 1.163,
+      "step": 8140
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.634384821402281e-06,
+      "loss": 1.2309,
+      "step": 8145
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.621290877964629e-06,
+      "loss": 1.1362,
+      "step": 8150
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.608206216138495e-06,
+      "loss": 1.1658,
+      "step": 8155
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.595130863659618e-06,
+      "loss": 1.1622,
+      "step": 8160
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.58206484824402e-06,
+      "loss": 1.174,
+      "step": 8165
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.569008197587904e-06,
+      "loss": 1.174,
+      "step": 8170
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.5559609393676425e-06,
+      "loss": 1.2279,
+      "step": 8175
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.542923101239692e-06,
+      "loss": 1.2327,
+      "step": 8180
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.529894710840543e-06,
+      "loss": 1.1632,
+      "step": 8185
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.516875795786658e-06,
+      "loss": 1.1902,
+      "step": 8190
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.503866383674414e-06,
+      "loss": 1.1656,
+      "step": 8195
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.490866502080046e-06,
+      "loss": 1.1589,
+      "step": 8200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.477876178559588e-06,
+      "loss": 1.1773,
+      "step": 8205
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.46489544064881e-06,
+      "loss": 1.1916,
+      "step": 8210
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.451924315863166e-06,
+      "loss": 1.1983,
+      "step": 8215
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.438962831697732e-06,
+      "loss": 1.2077,
+      "step": 8220
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.426011015627151e-06,
+      "loss": 1.227,
+      "step": 8225
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.413068895105567e-06,
+      "loss": 1.1851,
+      "step": 8230
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.400136497566577e-06,
+      "loss": 1.1802,
+      "step": 8235
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.3872138504231666e-06,
+      "loss": 1.2172,
+      "step": 8240
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.374300981067653e-06,
+      "loss": 1.1786,
+      "step": 8245
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.361397916871629e-06,
+      "loss": 1.1479,
+      "step": 8250
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.3485046851859005e-06,
+      "loss": 1.1466,
+      "step": 8255
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.3356213133404335e-06,
+      "loss": 1.2016,
+      "step": 8260
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.322747828644295e-06,
+      "loss": 1.1705,
+      "step": 8265
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.309884258385587e-06,
+      "loss": 1.2049,
+      "step": 8270
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.297030629831399e-06,
+      "loss": 1.1736,
+      "step": 8275
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.284186970227758e-06,
+      "loss": 1.1645,
+      "step": 8280
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.271353306799546e-06,
+      "loss": 1.1796,
+      "step": 8285
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.2585296667504606e-06,
+      "loss": 1.1631,
+      "step": 8290
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.245716077262952e-06,
+      "loss": 1.1809,
+      "step": 8295
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.232912565498167e-06,
+      "loss": 1.1861,
+      "step": 8300
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.220119158595891e-06,
+      "loss": 1.2004,
+      "step": 8305
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.207335883674491e-06,
+      "loss": 1.1687,
+      "step": 8310
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.194562767830851e-06,
+      "loss": 1.168,
+      "step": 8315
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.181799838140326e-06,
+      "loss": 1.1301,
+      "step": 8320
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.1690471216566785e-06,
+      "loss": 1.1684,
+      "step": 8325
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.156304645412017e-06,
+      "loss": 1.1532,
+      "step": 8330
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.143572436416757e-06,
+      "loss": 1.2316,
+      "step": 8335
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.1308505216595395e-06,
+      "loss": 1.1776,
+      "step": 8340
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.1181389281071835e-06,
+      "loss": 1.1898,
+      "step": 8345
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.105437682704634e-06,
+      "loss": 1.1783,
+      "step": 8350
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.0927468123749065e-06,
+      "loss": 1.1873,
+      "step": 8355
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.080066344019008e-06,
+      "loss": 1.1751,
+      "step": 8360
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.06739630451591e-06,
+      "loss": 1.1886,
+      "step": 8365
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.054736720722475e-06,
+      "loss": 1.1584,
+      "step": 8370
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.0420876194734e-06,
+      "loss": 1.1647,
+      "step": 8375
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.029449027581166e-06,
+      "loss": 1.1969,
+      "step": 8380
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.016820971835967e-06,
+      "loss": 1.1391,
+      "step": 8385
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.004203479005682e-06,
+      "loss": 1.1647,
+      "step": 8390
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.991596575835783e-06,
+      "loss": 1.1682,
+      "step": 8395
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.979000289049305e-06,
+      "loss": 1.1603,
+      "step": 8400
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.966414645346767e-06,
+      "loss": 1.1454,
+      "step": 8405
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.95383967140614e-06,
+      "loss": 1.1395,
+      "step": 8410
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.941275393882771e-06,
+      "loss": 1.1685,
+      "step": 8415
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.9287218394093414e-06,
+      "loss": 1.1629,
+      "step": 8420
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.916179034595794e-06,
+      "loss": 1.1836,
+      "step": 8425
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.90364700602929e-06,
+      "loss": 1.1993,
+      "step": 8430
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.891125780274148e-06,
+      "loss": 1.1676,
+      "step": 8435
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.878615383871781e-06,
+      "loss": 1.1483,
+      "step": 8440
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.866115843340666e-06,
+      "loss": 1.2043,
+      "step": 8445
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.853627185176245e-06,
+      "loss": 1.2105,
+      "step": 8450
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.841149435850905e-06,
+      "loss": 1.1823,
+      "step": 8455
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.828682621813907e-06,
+      "loss": 1.2282,
+      "step": 8460
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.816226769491335e-06,
+      "loss": 1.1808,
+      "step": 8465
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.8037819052860316e-06,
+      "loss": 1.1932,
+      "step": 8470
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.791348055577554e-06,
+      "loss": 1.1824,
+      "step": 8475
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.778925246722107e-06,
+      "loss": 1.1406,
+      "step": 8480
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.766513505052495e-06,
+      "loss": 1.2093,
+      "step": 8485
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.7541128568780614e-06,
+      "loss": 1.153,
+      "step": 8490
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.741723328484636e-06,
+      "loss": 1.1722,
+      "step": 8495
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.729344946134476e-06,
+      "loss": 1.1899,
+      "step": 8500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.716977736066213e-06,
+      "loss": 1.211,
+      "step": 8505
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.704621724494797e-06,
+      "loss": 1.1518,
+      "step": 8510
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6922769376114405e-06,
+      "loss": 1.1493,
+      "step": 8515
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.679943401583562e-06,
+      "loss": 1.1777,
+      "step": 8520
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6676211425547336e-06,
+      "loss": 1.1897,
+      "step": 8525
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.655310186644618e-06,
+      "loss": 1.1579,
+      "step": 8530
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.643010559948926e-06,
+      "loss": 1.1936,
+      "step": 8535
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.630722288539347e-06,
+      "loss": 1.1686,
+      "step": 8540
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.618445398463509e-06,
+      "loss": 1.173,
+      "step": 8545
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.606179915744897e-06,
+      "loss": 1.1778,
+      "step": 8550
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.593925866382839e-06,
+      "loss": 1.1726,
+      "step": 8555
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.5816832763524136e-06,
+      "loss": 1.1755,
+      "step": 8560
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.569452171604411e-06,
+      "loss": 1.1774,
+      "step": 8565
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.5572325780652745e-06,
+      "loss": 1.1861,
+      "step": 8570
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.545024521637053e-06,
+      "loss": 1.1517,
+      "step": 8575
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.532828028197332e-06,
+      "loss": 1.1635,
+      "step": 8580
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.520643123599195e-06,
+      "loss": 1.1598,
+      "step": 8585
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.508469833671155e-06,
+      "loss": 1.1551,
+      "step": 8590
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.496308184217103e-06,
+      "loss": 1.1823,
+      "step": 8595
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.484158201016262e-06,
+      "loss": 1.1431,
+      "step": 8600
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.472019909823121e-06,
+      "loss": 1.1759,
+      "step": 8605
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.459893336367384e-06,
+      "loss": 1.188,
+      "step": 8610
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.44777850635393e-06,
+      "loss": 1.179,
+      "step": 8615
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.4356754454627285e-06,
+      "loss": 1.1395,
+      "step": 8620
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.423584179348809e-06,
+      "loss": 1.1666,
+      "step": 8625
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.411504733642199e-06,
+      "loss": 1.1652,
+      "step": 8630
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.399437133947874e-06,
+      "loss": 1.1654,
+      "step": 8635
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.387381405845688e-06,
+      "loss": 1.1818,
+      "step": 8640
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.375337574890341e-06,
+      "loss": 1.1823,
+      "step": 8645
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.363305666611314e-06,
+      "loss": 1.1409,
+      "step": 8650
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.351285706512809e-06,
+      "loss": 1.1484,
+      "step": 8655
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.339277720073708e-06,
+      "loss": 1.1561,
+      "step": 8660
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.3272817327475035e-06,
+      "loss": 1.2047,
+      "step": 8665
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.315297769962267e-06,
+      "loss": 1.1645,
+      "step": 8670
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.303325857120572e-06,
+      "loss": 1.158,
+      "step": 8675
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.291366019599453e-06,
+      "loss": 1.1813,
+      "step": 8680
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.279418282750338e-06,
+      "loss": 1.1655,
+      "step": 8685
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.2674826718990185e-06,
+      "loss": 1.1897,
+      "step": 8690
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.255559212345577e-06,
+      "loss": 1.1897,
+      "step": 8695
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.243647929364339e-06,
+      "loss": 1.1659,
+      "step": 8700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.23174884820382e-06,
+      "loss": 1.1479,
+      "step": 8705
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.2198619940866684e-06,
+      "loss": 1.1797,
+      "step": 8710
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.207987392209617e-06,
+      "loss": 1.1888,
+      "step": 8715
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.1961250677434255e-06,
+      "loss": 1.1719,
+      "step": 8720
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.184275045832838e-06,
+      "loss": 1.1756,
+      "step": 8725
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.172437351596506e-06,
+      "loss": 1.1592,
+      "step": 8730
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.1606120101269564e-06,
+      "loss": 1.1912,
+      "step": 8735
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.1487990464905355e-06,
+      "loss": 1.1786,
+      "step": 8740
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.13699848572735e-06,
+      "loss": 1.1869,
+      "step": 8745
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.125210352851211e-06,
+      "loss": 1.1445,
+      "step": 8750
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.113434672849593e-06,
+      "loss": 1.14,
+      "step": 8755
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.101671470683572e-06,
+      "loss": 1.1824,
+      "step": 8760
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.08992077128777e-06,
+      "loss": 1.1622,
+      "step": 8765
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.078182599570314e-06,
+      "loss": 1.1942,
+      "step": 8770
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.0664569804127695e-06,
+      "loss": 1.1671,
+      "step": 8775
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.054743938670099e-06,
+      "loss": 1.1566,
+      "step": 8780
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.043043499170601e-06,
+      "loss": 1.1989,
+      "step": 8785
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.031355686715864e-06,
+      "loss": 1.2168,
+      "step": 8790
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.019680526080706e-06,
+      "loss": 1.1633,
+      "step": 8795
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.008018042013131e-06,
+      "loss": 1.1677,
+      "step": 8800
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.996368259234274e-06,
+      "loss": 1.186,
+      "step": 8805
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.984731202438339e-06,
+      "loss": 1.1441,
+      "step": 8810
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.973106896292563e-06,
+      "loss": 1.1849,
+      "step": 8815
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9614953654371504e-06,
+      "loss": 1.1486,
+      "step": 8820
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.949896634485227e-06,
+      "loss": 1.1722,
+      "step": 8825
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.938310728022789e-06,
+      "loss": 1.1613,
+      "step": 8830
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.926737670608641e-06,
+      "loss": 1.1718,
+      "step": 8835
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.915177486774361e-06,
+      "loss": 1.1727,
+      "step": 8840
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.903630201024231e-06,
+      "loss": 1.145,
+      "step": 8845
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.892095837835196e-06,
+      "loss": 1.159,
+      "step": 8850
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.880574421656809e-06,
+      "loss": 1.2088,
+      "step": 8855
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.869065976911177e-06,
+      "loss": 1.1756,
+      "step": 8860
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8575705279929145e-06,
+      "loss": 1.2112,
+      "step": 8865
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.846088099269085e-06,
+      "loss": 1.1465,
+      "step": 8870
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.834618715079161e-06,
+      "loss": 1.1797,
+      "step": 8875
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.823162399734949e-06,
+      "loss": 1.1635,
+      "step": 8880
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.811719177520562e-06,
+      "loss": 1.1898,
+      "step": 8885
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.800289072692368e-06,
+      "loss": 1.1456,
+      "step": 8890
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.7888721094789173e-06,
+      "loss": 1.1754,
+      "step": 8895
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.7774683120809065e-06,
+      "loss": 1.1631,
+      "step": 8900
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.766077704671128e-06,
+      "loss": 1.141,
+      "step": 8905
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.7547003113944135e-06,
+      "loss": 1.218,
+      "step": 8910
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.743336156367582e-06,
+      "loss": 1.196,
+      "step": 8915
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.7319852636793975e-06,
+      "loss": 1.1632,
+      "step": 8920
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.7206476573905016e-06,
+      "loss": 1.2215,
+      "step": 8925
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.70932336153338e-06,
+      "loss": 1.1395,
+      "step": 8930
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.698012400112303e-06,
+      "loss": 1.162,
+      "step": 8935
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.6867147971032724e-06,
+      "loss": 1.1958,
+      "step": 8940
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.6754305764539834e-06,
+      "loss": 1.2104,
+      "step": 8945
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.664159762083754e-06,
+      "loss": 1.1548,
+      "step": 8950
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.6529023778834895e-06,
+      "loss": 1.154,
+      "step": 8955
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.6416584477156246e-06,
+      "loss": 1.1636,
+      "step": 8960
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.630427995414081e-06,
+      "loss": 1.1223,
+      "step": 8965
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.619211044784199e-06,
+      "loss": 1.1882,
+      "step": 8970
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.6080076196027116e-06,
+      "loss": 1.107,
+      "step": 8975
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.5968177436176777e-06,
+      "loss": 1.2013,
+      "step": 8980
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.585641440548434e-06,
+      "loss": 1.1529,
+      "step": 8985
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.574478734085549e-06,
+      "loss": 1.2166,
+      "step": 8990
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.5633296478907686e-06,
+      "loss": 1.163,
+      "step": 8995
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.5521942055969648e-06,
+      "loss": 1.1727,
+      "step": 9000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.541072430808099e-06,
+      "loss": 1.1943,
+      "step": 9005
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.529964347099153e-06,
+      "loss": 1.1785,
+      "step": 9010
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.518869978016083e-06,
+      "loss": 1.1891,
+      "step": 9015
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.507789347075783e-06,
+      "loss": 1.1429,
+      "step": 9020
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.4967224777660215e-06,
+      "loss": 1.2049,
+      "step": 9025
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.4856693935453988e-06,
+      "loss": 1.1758,
+      "step": 9030
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.4746301178432938e-06,
+      "loss": 1.1723,
+      "step": 9035
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.463604674059814e-06,
+      "loss": 1.1551,
+      "step": 9040
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.4525930855657473e-06,
+      "loss": 1.1858,
+      "step": 9045
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.4415953757025165e-06,
+      "loss": 1.1908,
+      "step": 9050
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.4306115677821193e-06,
+      "loss": 1.1901,
+      "step": 9055
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.4196416850870896e-06,
+      "loss": 1.1711,
+      "step": 9060
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.408685750870443e-06,
+      "loss": 1.1615,
+      "step": 9065
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.3977437883556265e-06,
+      "loss": 1.213,
+      "step": 9070
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.3868158207364733e-06,
+      "loss": 1.1598,
+      "step": 9075
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.375901871177151e-06,
+      "loss": 1.1988,
+      "step": 9080
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.365001962812111e-06,
+      "loss": 1.1659,
+      "step": 9085
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.3541161187460446e-06,
+      "loss": 1.154,
+      "step": 9090
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.343244362053828e-06,
+      "loss": 1.1817,
+      "step": 9095
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.332386715780478e-06,
+      "loss": 1.1801,
+      "step": 9100
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.321543202941101e-06,
+      "loss": 1.1337,
+      "step": 9105
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.310713846520842e-06,
+      "loss": 1.1684,
+      "step": 9110
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.2998986694748425e-06,
+      "loss": 1.1321,
+      "step": 9115
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.289097694728186e-06,
+      "loss": 1.1259,
+      "step": 9120
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.278310945175851e-06,
+      "loss": 1.2283,
+      "step": 9125
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.267538443682662e-06,
+      "loss": 1.1637,
+      "step": 9130
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.2567802130832417e-06,
+      "loss": 1.1836,
+      "step": 9135
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.2460362761819653e-06,
+      "loss": 1.1788,
+      "step": 9140
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.2353066557529067e-06,
+      "loss": 1.1665,
+      "step": 9145
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.2245913745397928e-06,
+      "loss": 1.2191,
+      "step": 9150
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.213890455255961e-06,
+      "loss": 1.1818,
+      "step": 9155
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.2032039205842947e-06,
+      "loss": 1.1682,
+      "step": 9160
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.1925317931771904e-06,
+      "loss": 1.1711,
+      "step": 9165
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.1818740956565155e-06,
+      "loss": 1.1761,
+      "step": 9170
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.171230850613537e-06,
+      "loss": 1.1476,
+      "step": 9175
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.1606020806088932e-06,
+      "loss": 1.1709,
+      "step": 9180
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.1499878081725365e-06,
+      "loss": 1.1981,
+      "step": 9185
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.1393880558036906e-06,
+      "loss": 1.1849,
+      "step": 9190
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.1288028459707998e-06,
+      "loss": 1.1955,
+      "step": 9195
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.118232201111487e-06,
+      "loss": 1.1681,
+      "step": 9200
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.10767614363249e-06,
+      "loss": 1.1559,
+      "step": 9205
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.097134695909636e-06,
+      "loss": 1.1599,
+      "step": 9210
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0866078802877807e-06,
+      "loss": 1.164,
+      "step": 9215
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0760957190807607e-06,
+      "loss": 1.1875,
+      "step": 9220
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0655982345713585e-06,
+      "loss": 1.1636,
+      "step": 9225
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0551154490112365e-06,
+      "loss": 1.2126,
+      "step": 9230
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0446473846209056e-06,
+      "loss": 1.1627,
+      "step": 9235
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.034194063589666e-06,
+      "loss": 1.2011,
+      "step": 9240
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0237555080755754e-06,
+      "loss": 1.1718,
+      "step": 9245
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.013331740205381e-06,
+      "loss": 1.1891,
+      "step": 9250
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.0029227820744922e-06,
+      "loss": 1.1399,
+      "step": 9255
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.992528655746926e-06,
+      "loss": 1.1963,
+      "step": 9260
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9821493832552583e-06,
+      "loss": 1.1582,
+      "step": 9265
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9717849866005777e-06,
+      "loss": 1.1445,
+      "step": 9270
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.961435487752442e-06,
+      "loss": 1.1635,
+      "step": 9275
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.951100908648834e-06,
+      "loss": 1.1826,
+      "step": 9280
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.940781271196105e-06,
+      "loss": 1.1903,
+      "step": 9285
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9304765972689375e-06,
+      "loss": 1.1467,
+      "step": 9290
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9201869087102896e-06,
+      "loss": 1.2058,
+      "step": 9295
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9099122273313617e-06,
+      "loss": 1.1651,
+      "step": 9300
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.899652574911542e-06,
+      "loss": 1.1611,
+      "step": 9305
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8894079731983593e-06,
+      "loss": 1.2096,
+      "step": 9310
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.879178443907441e-06,
+      "loss": 1.1904,
+      "step": 9315
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8689640087224648e-06,
+      "loss": 1.1889,
+      "step": 9320
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.858764689295113e-06,
+      "loss": 1.1462,
+      "step": 9325
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8485805072450223e-06,
+      "loss": 1.1646,
+      "step": 9330
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8384114841597597e-06,
+      "loss": 1.1795,
+      "step": 9335
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.828257641594735e-06,
+      "loss": 1.1599,
+      "step": 9340
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.818119001073195e-06,
+      "loss": 1.1647,
+      "step": 9345
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8079955840861595e-06,
+      "loss": 1.1744,
+      "step": 9350
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.7978874120923783e-06,
+      "loss": 1.1848,
+      "step": 9355
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.7877945065182876e-06,
+      "loss": 1.2079,
+      "step": 9360
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.7777168887579585e-06,
+      "loss": 1.1826,
+      "step": 9365
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.767654580173064e-06,
+      "loss": 1.1547,
+      "step": 9370
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.7576076020928165e-06,
+      "loss": 1.1624,
+      "step": 9375
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.747575975813942e-06,
+      "loss": 1.1586,
+      "step": 9380
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.7375597226006167e-06,
+      "loss": 1.1859,
+      "step": 9385
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.7275588636844364e-06,
+      "loss": 1.1429,
+      "step": 9390
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.7175734202643613e-06,
+      "loss": 1.1979,
+      "step": 9395
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.7076034135066766e-06,
+      "loss": 1.1656,
+      "step": 9400
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6976488645449484e-06,
+      "loss": 1.1626,
+      "step": 9405
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6877097944799723e-06,
+      "loss": 1.175,
+      "step": 9410
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.677786224379736e-06,
+      "loss": 1.1628,
+      "step": 9415
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6678781752793727e-06,
+      "loss": 1.1507,
+      "step": 9420
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.657985668181111e-06,
+      "loss": 1.1735,
+      "step": 9425
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.64810872405424e-06,
+      "loss": 1.1624,
+      "step": 9430
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.638247363835057e-06,
+      "loss": 1.154,
+      "step": 9435
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6284016084268292e-06,
+      "loss": 1.1474,
+      "step": 9440
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.618571478699735e-06,
+      "loss": 1.1559,
+      "step": 9445
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6087569954908466e-06,
+      "loss": 1.1702,
+      "step": 9450
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.5989581796040607e-06,
+      "loss": 1.1853,
+      "step": 9455
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.5891750518100654e-06,
+      "loss": 1.167,
+      "step": 9460
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.579407632846296e-06,
+      "loss": 1.1673,
+      "step": 9465
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.5696559434168856e-06,
+      "loss": 1.1511,
+      "step": 9470
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.5599200041926297e-06,
+      "loss": 1.2293,
+      "step": 9475
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.550199835810936e-06,
+      "loss": 1.2052,
+      "step": 9480
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.5404954588757845e-06,
+      "loss": 1.1939,
+      "step": 9485
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.5308068939576757e-06,
+      "loss": 1.1613,
+      "step": 9490
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.521134161593599e-06,
+      "loss": 1.172,
+      "step": 9495
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.5114772822869794e-06,
+      "loss": 1.1662,
+      "step": 9500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.5018362765076453e-06,
+      "loss": 1.1645,
+      "step": 9505
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.4922111646917713e-06,
+      "loss": 1.2127,
+      "step": 9510
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.4826019672418412e-06,
+      "loss": 1.17,
+      "step": 9515
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.4730087045266073e-06,
+      "loss": 1.1377,
+      "step": 9520
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.4634313968810442e-06,
+      "loss": 1.1795,
+      "step": 9525
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.453870064606307e-06,
+      "loss": 1.1749,
+      "step": 9530
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.4443247279696834e-06,
+      "loss": 1.1911,
+      "step": 9535
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.43479540720456e-06,
+      "loss": 1.2201,
+      "step": 9540
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.425282122510373e-06,
+      "loss": 1.1828,
+      "step": 9545
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.415784894052565e-06,
+      "loss": 1.1711,
+      "step": 9550
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.4063037419625433e-06,
+      "loss": 1.1984,
+      "step": 9555
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.3968386863376447e-06,
+      "loss": 1.1813,
+      "step": 9560
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.38738974724108e-06,
+      "loss": 1.175,
+      "step": 9565
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.377956944701897e-06,
+      "loss": 1.1703,
+      "step": 9570
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3685402987149442e-06,
+      "loss": 1.1822,
+      "step": 9575
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.359139829240812e-06,
+      "loss": 1.1786,
+      "step": 9580
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3497555562058126e-06,
+      "loss": 1.2086,
+      "step": 9585
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3403874995019214e-06,
+      "loss": 1.1834,
+      "step": 9590
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3310356789867407e-06,
+      "loss": 1.2014,
+      "step": 9595
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3217001144834563e-06,
+      "loss": 1.176,
+      "step": 9600
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3123808257807944e-06,
+      "loss": 1.1343,
+      "step": 9605
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3030778326329827e-06,
+      "loss": 1.1911,
+      "step": 9610
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.29379115475971e-06,
+      "loss": 1.1506,
+      "step": 9615
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2845208118460814e-06,
+      "loss": 1.187,
+      "step": 9620
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2752668235425658e-06,
+      "loss": 1.2089,
+      "step": 9625
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.266029209464976e-06,
+      "loss": 1.164,
+      "step": 9630
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.256807989194413e-06,
+      "loss": 1.1565,
+      "step": 9635
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.247603182277228e-06,
+      "loss": 1.1692,
+      "step": 9640
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.23841480822498e-06,
+      "loss": 1.1403,
+      "step": 9645
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2292428865143943e-06,
+      "loss": 1.1913,
+      "step": 9650
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.2200874365873236e-06,
+      "loss": 1.1724,
+      "step": 9655
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.210948477850704e-06,
+      "loss": 1.1649,
+      "step": 9660
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.201826029676516e-06,
+      "loss": 1.1733,
+      "step": 9665
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1927201114017405e-06,
+      "loss": 1.1931,
+      "step": 9670
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.183630742328322e-06,
+      "loss": 1.181,
+      "step": 9675
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.174557941723123e-06,
+      "loss": 1.1893,
+      "step": 9680
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1655017288178893e-06,
+      "loss": 1.1415,
+      "step": 9685
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1564621228092007e-06,
+      "loss": 1.1901,
+      "step": 9690
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1474391428584394e-06,
+      "loss": 1.1582,
+      "step": 9695
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.138432808091744e-06,
+      "loss": 1.1787,
+      "step": 9700
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.129443137599967e-06,
+      "loss": 1.1789,
+      "step": 9705
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1204701504386405e-06,
+      "loss": 1.1459,
+      "step": 9710
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1115138656279333e-06,
+      "loss": 1.1737,
+      "step": 9715
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1025743021526067e-06,
+      "loss": 1.1846,
+      "step": 9720
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.093651478961982e-06,
+      "loss": 1.1408,
+      "step": 9725
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.084745414969892e-06,
+      "loss": 1.2022,
+      "step": 9730
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0758561290546454e-06,
+      "loss": 1.1632,
+      "step": 9735
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0669836400589893e-06,
+      "loss": 1.1438,
+      "step": 9740
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.058127966790061e-06,
+      "loss": 1.2249,
+      "step": 9745
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0492891280193593e-06,
+      "loss": 1.1636,
+      "step": 9750
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0404671424826925e-06,
+      "loss": 1.1621,
+      "step": 9755
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.03166202888015e-06,
+      "loss": 1.19,
+      "step": 9760
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.022873805876057e-06,
+      "loss": 1.2048,
+      "step": 9765
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0141024920989284e-06,
+      "loss": 1.1891,
+      "step": 9770
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.0053481061414447e-06,
+      "loss": 1.1601,
+      "step": 9775
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.996610666560399e-06,
+      "loss": 1.2149,
+      "step": 9780
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9878901918766704e-06,
+      "loss": 1.144,
+      "step": 9785
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9791867005751685e-06,
+      "loss": 1.1475,
+      "step": 9790
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9705002111048077e-06,
+      "loss": 1.1591,
+      "step": 9795
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9618307418784634e-06,
+      "loss": 1.1655,
+      "step": 9800
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9531783112729297e-06,
+      "loss": 1.1642,
+      "step": 9805
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9445429376288893e-06,
+      "loss": 1.1591,
+      "step": 9810
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9359246392508603e-06,
+      "loss": 1.1608,
+      "step": 9815
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9273234344071745e-06,
+      "loss": 1.1634,
+      "step": 9820
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.918739341329927e-06,
+      "loss": 1.1532,
+      "step": 9825
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9101723782149404e-06,
+      "loss": 1.1976,
+      "step": 9830
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9016225632217256e-06,
+      "loss": 1.1569,
+      "step": 9835
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.8930899144734505e-06,
+      "loss": 1.1807,
+      "step": 9840
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.8845744500568897e-06,
+      "loss": 1.1487,
+      "step": 9845
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.876076188022392e-06,
+      "loss": 1.1534,
+      "step": 9850
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.867595146383846e-06,
+      "loss": 1.1435,
+      "step": 9855
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.8591313431186308e-06,
+      "loss": 1.1449,
+      "step": 9860
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.8506847961675923e-06,
+      "loss": 1.1302,
+      "step": 9865
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.8422555234349936e-06,
+      "loss": 1.137,
+      "step": 9870
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.833843542788486e-06,
+      "loss": 1.15,
+      "step": 9875
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.8254488720590612e-06,
+      "loss": 1.2089,
+      "step": 9880
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.8170715290410223e-06,
+      "loss": 1.1646,
+      "step": 9885
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.8087115314919378e-06,
+      "loss": 1.1925,
+      "step": 9890
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.8003688971326194e-06,
+      "loss": 1.2079,
+      "step": 9895
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7920436436470667e-06,
+      "loss": 1.1814,
+      "step": 9900
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7837357886824292e-06,
+      "loss": 1.1754,
+      "step": 9905
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7754453498489899e-06,
+      "loss": 1.1678,
+      "step": 9910
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7671723447201083e-06,
+      "loss": 1.1599,
+      "step": 9915
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.758916790832189e-06,
+      "loss": 1.1596,
+      "step": 9920
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7506787056846463e-06,
+      "loss": 1.1768,
+      "step": 9925
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.742458106739867e-06,
+      "loss": 1.1702,
+      "step": 9930
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7342550114231692e-06,
+      "loss": 1.1346,
+      "step": 9935
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7260694371227705e-06,
+      "loss": 1.1624,
+      "step": 9940
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7179014011897487e-06,
+      "loss": 1.1864,
+      "step": 9945
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7097509209380036e-06,
+      "loss": 1.1504,
+      "step": 9950
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.701618013644224e-06,
+      "loss": 1.1959,
+      "step": 9955
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6935026965478474e-06,
+      "loss": 1.1755,
+      "step": 9960
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.685404986851027e-06,
+      "loss": 1.1496,
+      "step": 9965
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6773249017185923e-06,
+      "loss": 1.1901,
+      "step": 9970
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6692624582780136e-06,
+      "loss": 1.1193,
+      "step": 9975
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6612176736193652e-06,
+      "loss": 1.2188,
+      "step": 9980
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6531905647952928e-06,
+      "loss": 1.127,
+      "step": 9985
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.645181148820969e-06,
+      "loss": 1.1813,
+      "step": 9990
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6371894426740686e-06,
+      "loss": 1.1807,
+      "step": 9995
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6292154632947233e-06,
+      "loss": 1.1314,
+      "step": 10000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6212592275854887e-06,
+      "loss": 1.1893,
+      "step": 10005
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6133207524113115e-06,
+      "loss": 1.1806,
+      "step": 10010
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6054000545994885e-06,
+      "loss": 1.1782,
+      "step": 10015
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5974971509396342e-06,
+      "loss": 1.1501,
+      "step": 10020
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.589612058183646e-06,
+      "loss": 1.1734,
+      "step": 10025
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5817447930456652e-06,
+      "loss": 1.1662,
+      "step": 10030
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5738953722020466e-06,
+      "loss": 1.1779,
+      "step": 10035
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5660638122913164e-06,
+      "loss": 1.136,
+      "step": 10040
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5582501299141461e-06,
+      "loss": 1.1589,
+      "step": 10045
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.550454341633304e-06,
+      "loss": 1.1938,
+      "step": 10050
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5426764639736313e-06,
+      "loss": 1.1742,
+      "step": 10055
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5349165134220113e-06,
+      "loss": 1.1765,
+      "step": 10060
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.527174506427317e-06,
+      "loss": 1.1339,
+      "step": 10065
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5194504594003901e-06,
+      "loss": 1.1435,
+      "step": 10070
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5117443887140026e-06,
+      "loss": 1.132,
+      "step": 10075
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5040563107028206e-06,
+      "loss": 1.1837,
+      "step": 10080
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.4963862416633722e-06,
+      "loss": 1.1965,
+      "step": 10085
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.4887341978540137e-06,
+      "loss": 1.1546,
+      "step": 10090
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4811001954948846e-06,
+      "loss": 1.2142,
+      "step": 10095
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4734842507678904e-06,
+      "loss": 1.1881,
+      "step": 10100
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4658863798166567e-06,
+      "loss": 1.1412,
+      "step": 10105
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.458306598746495e-06,
+      "loss": 1.121,
+      "step": 10110
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4507449236243797e-06,
+      "loss": 1.1218,
+      "step": 10115
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4432013704788973e-06,
+      "loss": 1.1891,
+      "step": 10120
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.435675955300223e-06,
+      "loss": 1.1624,
+      "step": 10125
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4281686940400874e-06,
+      "loss": 1.1775,
+      "step": 10130
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4206796026117385e-06,
+      "loss": 1.1552,
+      "step": 10135
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4132086968899062e-06,
+      "loss": 1.1934,
+      "step": 10140
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4057559927107767e-06,
+      "loss": 1.1665,
+      "step": 10145
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.3983215058719512e-06,
+      "loss": 1.1541,
+      "step": 10150
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.3909052521324174e-06,
+      "loss": 1.1505,
+      "step": 10155
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.3835072472125122e-06,
+      "loss": 1.186,
+      "step": 10160
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.3761275067938917e-06,
+      "loss": 1.1351,
+      "step": 10165
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.368766046519493e-06,
+      "loss": 1.1838,
+      "step": 10170
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.3614228819935128e-06,
+      "loss": 1.194,
+      "step": 10175
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.3540980287813599e-06,
+      "loss": 1.1774,
+      "step": 10180
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.3467915024096246e-06,
+      "loss": 1.1288,
+      "step": 10185
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.3395033183660555e-06,
+      "loss": 1.172,
+      "step": 10190
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.3322334920995205e-06,
+      "loss": 1.1381,
+      "step": 10195
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.324982039019972e-06,
+      "loss": 1.1779,
+      "step": 10200
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.3177489744984184e-06,
+      "loss": 1.1817,
+      "step": 10205
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.3105343138668868e-06,
+      "loss": 1.1436,
+      "step": 10210
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.3033380724183976e-06,
+      "loss": 1.1776,
+      "step": 10215
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2961602654069228e-06,
+      "loss": 1.1716,
+      "step": 10220
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.289000908047363e-06,
+      "loss": 1.1948,
+      "step": 10225
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2818600155155092e-06,
+      "loss": 1.1613,
+      "step": 10230
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2747376029480108e-06,
+      "loss": 1.199,
+      "step": 10235
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2676336854423465e-06,
+      "loss": 1.1528,
+      "step": 10240
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2605482780567913e-06,
+      "loss": 1.1716,
+      "step": 10245
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.253481395810382e-06,
+      "loss": 1.1594,
+      "step": 10250
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2464330536828906e-06,
+      "loss": 1.1797,
+      "step": 10255
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.239403266614785e-06,
+      "loss": 1.1644,
+      "step": 10260
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2323920495072051e-06,
+      "loss": 1.1757,
+      "step": 10265
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2253994172219276e-06,
+      "loss": 1.1698,
+      "step": 10270
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.218425384581332e-06,
+      "loss": 1.1585,
+      "step": 10275
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.211469966368376e-06,
+      "loss": 1.1688,
+      "step": 10280
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2045331773265556e-06,
+      "loss": 1.1422,
+      "step": 10285
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1976150321598822e-06,
+      "loss": 1.1771,
+      "step": 10290
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.1907155455328435e-06,
+      "loss": 1.1598,
+      "step": 10295
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.1838347320703814e-06,
+      "loss": 1.1842,
+      "step": 10300
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.1769726063578514e-06,
+      "loss": 1.1867,
+      "step": 10305
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.1701291829409988e-06,
+      "loss": 1.1725,
+      "step": 10310
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.1633044763259238e-06,
+      "loss": 1.1797,
+      "step": 10315
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.1564985009790552e-06,
+      "loss": 1.1703,
+      "step": 10320
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.149711271327114e-06,
+      "loss": 1.1667,
+      "step": 10325
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.1429428017570887e-06,
+      "loss": 1.1551,
+      "step": 10330
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.1361931066161936e-06,
+      "loss": 1.1705,
+      "step": 10335
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.1294622002118593e-06,
+      "loss": 1.1738,
+      "step": 10340
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.122750096811681e-06,
+      "loss": 1.1964,
+      "step": 10345
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.1160568106433978e-06,
+      "loss": 1.1661,
+      "step": 10350
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.1093823558948647e-06,
+      "loss": 1.1858,
+      "step": 10355
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.1027267467140156e-06,
+      "loss": 1.1658,
+      "step": 10360
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0960899972088413e-06,
+      "loss": 1.1378,
+      "step": 10365
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0894721214473514e-06,
+      "loss": 1.1789,
+      "step": 10370
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.0828731334575537e-06,
+      "loss": 1.1808,
+      "step": 10375
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.0762930472274125e-06,
+      "loss": 1.156,
+      "step": 10380
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.0697318767048315e-06,
+      "loss": 1.1687,
+      "step": 10385
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.0631896357976124e-06,
+      "loss": 1.1478,
+      "step": 10390
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.056666338373442e-06,
+      "loss": 1.1457,
+      "step": 10395
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.0501619982598432e-06,
+      "loss": 1.1407,
+      "step": 10400
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.043676629244157e-06,
+      "loss": 1.1324,
+      "step": 10405
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.037210245073511e-06,
+      "loss": 1.1507,
+      "step": 10410
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.0307628594547925e-06,
+      "loss": 1.1436,
+      "step": 10415
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.0243344860546178e-06,
+      "loss": 1.1894,
+      "step": 10420
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.017925138499295e-06,
+      "loss": 1.1733,
+      "step": 10425
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.0115348303748128e-06,
+      "loss": 1.1448,
+      "step": 10430
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.0051635752267952e-06,
+      "loss": 1.1906,
+      "step": 10435
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.988113865604843e-07,
+      "loss": 1.1686,
+      "step": 10440
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.924782778406994e-07,
+      "loss": 1.1932,
+      "step": 10445
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.861642624918256e-07,
+      "loss": 1.1864,
+      "step": 10450
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.798693538977688e-07,
+      "loss": 1.1869,
+      "step": 10455
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.735935654019346e-07,
+      "loss": 1.2176,
+      "step": 10460
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.673369103072029e-07,
+      "loss": 1.1739,
+      "step": 10465
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.610994018758901e-07,
+      "loss": 1.1697,
+      "step": 10470
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.548810533297325e-07,
+      "loss": 1.197,
+      "step": 10475
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.486818778498519e-07,
+      "loss": 1.1945,
+      "step": 10480
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.425018885767278e-07,
+      "loss": 1.2414,
+      "step": 10485
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.363410986101695e-07,
+      "loss": 1.1554,
+      "step": 10490
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.301995210092918e-07,
+      "loss": 1.1748,
+      "step": 10495
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.240771687924821e-07,
+      "loss": 1.1781,
+      "step": 10500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.179740549373762e-07,
+      "loss": 1.1481,
+      "step": 10505
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.118901923808365e-07,
+      "loss": 1.1643,
+      "step": 10510
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.058255940189053e-07,
+      "loss": 1.1522,
+      "step": 10515
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.997802727068006e-07,
+      "loss": 1.1656,
+      "step": 10520
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.93754241258874e-07,
+      "loss": 1.1682,
+      "step": 10525
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.877475124485901e-07,
+      "loss": 1.157,
+      "step": 10530
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.817600990084974e-07,
+      "loss": 1.1502,
+      "step": 10535
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.75792013630199e-07,
+      "loss": 1.1264,
+      "step": 10540
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.69843268964331e-07,
+      "loss": 1.195,
+      "step": 10545
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.639138776205302e-07,
+      "loss": 1.1672,
+      "step": 10550
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.580038521674128e-07,
+      "loss": 1.183,
+      "step": 10555
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.521132051325432e-07,
+      "loss": 1.1569,
+      "step": 10560
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.462419490024065e-07,
+      "loss": 1.1799,
+      "step": 10565
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.403900962223899e-07,
+      "loss": 1.1377,
+      "step": 10570
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.345576591967463e-07,
+      "loss": 1.1696,
+      "step": 10575
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.287446502885766e-07,
+      "loss": 1.1694,
+      "step": 10580
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.229510818197961e-07,
+      "loss": 1.2007,
+      "step": 10585
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.17176966071116e-07,
+      "loss": 1.1403,
+      "step": 10590
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.114223152820078e-07,
+      "loss": 1.1403,
+      "step": 10595
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.056871416506884e-07,
+      "loss": 1.1806,
+      "step": 10600
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.999714573340855e-07,
+      "loss": 1.1422,
+      "step": 10605
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.942752744478166e-07,
+      "loss": 1.1549,
+      "step": 10610
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.885986050661576e-07,
+      "loss": 1.1747,
+      "step": 10615
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.829414612220287e-07,
+      "loss": 1.1669,
+      "step": 10620
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.773038549069567e-07,
+      "loss": 1.1943,
+      "step": 10625
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.716857980710535e-07,
+      "loss": 1.1826,
+      "step": 10630
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.660873026229953e-07,
+      "loss": 1.1865,
+      "step": 10635
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.605083804299906e-07,
+      "loss": 1.1773,
+      "step": 10640
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.549490433177609e-07,
+      "loss": 1.1726,
+      "step": 10645
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.494093030705108e-07,
+      "loss": 1.1596,
+      "step": 10650
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.438891714309071e-07,
+      "loss": 1.1506,
+      "step": 10655
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.383886601000479e-07,
+      "loss": 1.1446,
+      "step": 10660
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.329077807374463e-07,
+      "loss": 1.1694,
+      "step": 10665
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.274465449609969e-07,
+      "loss": 1.1619,
+      "step": 10670
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.22004964346964e-07,
+      "loss": 1.1949,
+      "step": 10675
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.165830504299387e-07,
+      "loss": 1.1794,
+      "step": 10680
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.111808147028299e-07,
+      "loss": 1.2018,
+      "step": 10685
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.05798268616833e-07,
+      "loss": 1.192,
+      "step": 10690
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.004354235814059e-07,
+      "loss": 1.1708,
+      "step": 10695
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.950922909642488e-07,
+      "loss": 1.1778,
+      "step": 10700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.897688820912729e-07,
+      "loss": 1.1172,
+      "step": 10705
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.84465208246583e-07,
+      "loss": 1.1617,
+      "step": 10710
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.79181280672454e-07,
+      "loss": 1.1738,
+      "step": 10715
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.739171105692987e-07,
+      "loss": 1.1818,
+      "step": 10720
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.686727090956535e-07,
+      "loss": 1.2152,
+      "step": 10725
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.634480873681526e-07,
+      "loss": 1.1738,
+      "step": 10730
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.582432564614993e-07,
+      "loss": 1.1458,
+      "step": 10735
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.530582274084463e-07,
+      "loss": 1.1648,
+      "step": 10740
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.47893011199775e-07,
+      "loss": 1.1281,
+      "step": 10745
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.42747618784263e-07,
+      "loss": 1.1768,
+      "step": 10750
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.376220610686734e-07,
+      "loss": 1.1573,
+      "step": 10755
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.325163489177244e-07,
+      "loss": 1.2035,
+      "step": 10760
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.274304931540643e-07,
+      "loss": 1.1829,
+      "step": 10765
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.223645045582549e-07,
+      "loss": 1.1542,
+      "step": 10770
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.173183938687421e-07,
+      "loss": 1.2316,
+      "step": 10775
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.122921717818375e-07,
+      "loss": 1.1714,
+      "step": 10780
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.072858489516975e-07,
+      "loss": 1.1665,
+      "step": 10785
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.022994359902957e-07,
+      "loss": 1.1824,
+      "step": 10790
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.97332943467398e-07,
+      "loss": 1.2014,
+      "step": 10795
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.923863819105513e-07,
+      "loss": 1.1609,
+      "step": 10800
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.874597618050526e-07,
+      "loss": 1.1358,
+      "step": 10805
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.825530935939261e-07,
+      "loss": 1.1847,
+      "step": 10810
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.776663876779076e-07,
+      "loss": 1.1616,
+      "step": 10815
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.727996544154147e-07,
+      "loss": 1.162,
+      "step": 10820
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.679529041225318e-07,
+      "loss": 1.1603,
+      "step": 10825
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.631261470729832e-07,
+      "loss": 1.1467,
+      "step": 10830
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.583193934981146e-07,
+      "loss": 1.1861,
+      "step": 10835
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.53532653586869e-07,
+      "loss": 1.1476,
+      "step": 10840
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.487659374857668e-07,
+      "loss": 1.1849,
+      "step": 10845
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.440192552988833e-07,
+      "loss": 1.144,
+      "step": 10850
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.392926170878277e-07,
+      "loss": 1.1875,
+      "step": 10855
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.345860328717222e-07,
+      "loss": 1.1862,
+      "step": 10860
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.298995126271789e-07,
+      "loss": 1.1436,
+      "step": 10865
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.252330662882809e-07,
+      "loss": 1.181,
+      "step": 10870
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.205867037465606e-07,
+      "loss": 1.1517,
+      "step": 10875
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.159604348509784e-07,
+      "loss": 1.1746,
+      "step": 10880
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.113542694079021e-07,
+      "loss": 1.1842,
+      "step": 10885
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.06768217181085e-07,
+      "loss": 1.1808,
+      "step": 10890
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.022022878916466e-07,
+      "loss": 1.1588,
+      "step": 10895
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.976564912180526e-07,
+      "loss": 1.1522,
+      "step": 10900
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.931308367960919e-07,
+      "loss": 1.1722,
+      "step": 10905
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.886253342188574e-07,
+      "loss": 1.1574,
+      "step": 10910
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.841399930367264e-07,
+      "loss": 1.1837,
+      "step": 10915
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.796748227573411e-07,
+      "loss": 1.1755,
+      "step": 10920
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.7522983284558513e-07,
+      "loss": 1.167,
+      "step": 10925
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.7080503272356693e-07,
+      "loss": 1.1314,
+      "step": 10930
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.6640043177059725e-07,
+      "loss": 1.1654,
+      "step": 10935
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.620160393231732e-07,
+      "loss": 1.1853,
+      "step": 10940
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5765186467494837e-07,
+      "loss": 1.137,
+      "step": 10945
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.533079170767274e-07,
+      "loss": 1.1647,
+      "step": 10950
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.489842057364391e-07,
+      "loss": 1.164,
+      "step": 10955
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.4468073981911555e-07,
+      "loss": 1.1759,
+      "step": 10960
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.40397528446872e-07,
+      "loss": 1.2138,
+      "step": 10965
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3613458069889257e-07,
+      "loss": 1.2001,
+      "step": 10970
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.3189190561140904e-07,
+      "loss": 1.1872,
+      "step": 10975
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.276695121776786e-07,
+      "loss": 1.1352,
+      "step": 10980
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.234674093479696e-07,
+      "loss": 1.1686,
+      "step": 10985
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.192856060295358e-07,
+      "loss": 1.1468,
+      "step": 10990
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.1512411108660775e-07,
+      "loss": 1.1489,
+      "step": 10995
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.1098293334036255e-07,
+      "loss": 1.1682,
+      "step": 11000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.068620815689139e-07,
+      "loss": 1.1469,
+      "step": 11005
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.027615645072902e-07,
+      "loss": 1.1656,
+      "step": 11010
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.986813908474152e-07,
+      "loss": 1.1273,
+      "step": 11015
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.946215692380906e-07,
+      "loss": 1.152,
+      "step": 11020
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.90582108284977e-07,
+      "loss": 1.1835,
+      "step": 11025
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.865630165505785e-07,
+      "loss": 1.1859,
+      "step": 11030
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.8256430255421805e-07,
+      "loss": 1.1777,
+      "step": 11035
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.7858597477202777e-07,
+      "loss": 1.1778,
+      "step": 11040
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.746280416369241e-07,
+      "loss": 1.1447,
+      "step": 11045
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.7069051153859394e-07,
+      "loss": 1.1695,
+      "step": 11050
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6677339282347624e-07,
+      "loss": 1.1507,
+      "step": 11055
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.628766937947414e-07,
+      "loss": 1.1656,
+      "step": 11060
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.5900042271227897e-07,
+      "loss": 1.1587,
+      "step": 11065
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.5514458779267514e-07,
+      "loss": 1.182,
+      "step": 11070
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.5130919720919865e-07,
+      "loss": 1.1883,
+      "step": 11075
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.474942590917774e-07,
+      "loss": 1.1568,
+      "step": 11080
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.436997815269927e-07,
+      "loss": 1.1241,
+      "step": 11085
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.399257725580518e-07,
+      "loss": 1.1565,
+      "step": 11090
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.3617224018477335e-07,
+      "loss": 1.1627,
+      "step": 11095
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.3243919236357503e-07,
+      "loss": 1.1669,
+      "step": 11100
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.287266370074493e-07,
+      "loss": 1.186,
+      "step": 11105
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.2503458198595237e-07,
+      "loss": 1.1563,
+      "step": 11110
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.2136303512518396e-07,
+      "loss": 1.1903,
+      "step": 11115
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.177120042077786e-07,
+      "loss": 1.15,
+      "step": 11120
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.140814969728734e-07,
+      "loss": 1.1537,
+      "step": 11125
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.104715211161069e-07,
+      "loss": 1.123,
+      "step": 11130
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.0688208428959696e-07,
+      "loss": 1.1995,
+      "step": 11135
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.0331319410192297e-07,
+      "loss": 1.1394,
+      "step": 11140
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.997648581181112e-07,
+      "loss": 1.1821,
+      "step": 11145
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.9623708385961956e-07,
+      "loss": 1.2199,
+      "step": 11150
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.927298788043209e-07,
+      "loss": 1.1818,
+      "step": 11155
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.892432503864884e-07,
+      "loss": 1.1638,
+      "step": 11160
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.857772059967767e-07,
+      "loss": 1.189,
+      "step": 11165
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.8233175298221005e-07,
+      "loss": 1.1623,
+      "step": 11170
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.789068986461618e-07,
+      "loss": 1.1262,
+      "step": 11175
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.755026502483449e-07,
+      "loss": 1.1949,
+      "step": 11180
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.721190150047925e-07,
+      "loss": 1.1691,
+      "step": 11185
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.687560000878453e-07,
+      "loss": 1.1966,
+      "step": 11190
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.6541361262613307e-07,
+      "loss": 1.1539,
+      "step": 11195
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.6209185970455963e-07,
+      "loss": 1.2227,
+      "step": 11200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.5879074836429375e-07,
+      "loss": 1.1795,
+      "step": 11205
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.5551028560274803e-07,
+      "loss": 1.114,
+      "step": 11210
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.5225047837356354e-07,
+      "loss": 1.1542,
+      "step": 11215
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.490113335866018e-07,
+      "loss": 1.1837,
+      "step": 11220
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.4579285810792054e-07,
+      "loss": 1.1338,
+      "step": 11225
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.425950587597714e-07,
+      "loss": 1.1599,
+      "step": 11230
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.3941794232057334e-07,
+      "loss": 1.1572,
+      "step": 11235
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.3626151552490485e-07,
+      "loss": 1.1441,
+      "step": 11240
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.3312578506348828e-07,
+      "loss": 1.2149,
+      "step": 11245
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.3001075758317448e-07,
+      "loss": 1.1484,
+      "step": 11250
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.269164396869339e-07,
+      "loss": 1.1565,
+      "step": 11255
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.238428379338342e-07,
+      "loss": 1.1702,
+      "step": 11260
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.2078995883903276e-07,
+      "loss": 1.1957,
+      "step": 11265
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1775780887376086e-07,
+      "loss": 1.1675,
+      "step": 11270
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.147463944653072e-07,
+      "loss": 1.1828,
+      "step": 11275
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1175572199701233e-07,
+      "loss": 1.1801,
+      "step": 11280
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0878579780824525e-07,
+      "loss": 1.1313,
+      "step": 11285
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0583662819439686e-07,
+      "loss": 1.1136,
+      "step": 11290
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0290821940686323e-07,
+      "loss": 1.1462,
+      "step": 11295
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.000005776530345e-07,
+      "loss": 1.1481,
+      "step": 11300
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9711370909627935e-07,
+      "loss": 1.1498,
+      "step": 11305
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.94247619855934e-07,
+      "loss": 1.1507,
+      "step": 11310
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9140231600728866e-07,
+      "loss": 1.128,
+      "step": 11315
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.885778035815722e-07,
+      "loss": 1.1342,
+      "step": 11320
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8577408856594536e-07,
+      "loss": 1.2074,
+      "step": 11325
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.82991176903482e-07,
+      "loss": 1.1833,
+      "step": 11330
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8022907449316007e-07,
+      "loss": 1.1808,
+      "step": 11335
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7748778718984394e-07,
+      "loss": 1.1571,
+      "step": 11340
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7476732080428215e-07,
+      "loss": 1.1712,
+      "step": 11345
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7206768110308524e-07,
+      "loss": 1.177,
+      "step": 11350
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6938887380871683e-07,
+      "loss": 1.1719,
+      "step": 11355
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.667309045994825e-07,
+      "loss": 1.194,
+      "step": 11360
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6409377910951763e-07,
+      "loss": 1.1867,
+      "step": 11365
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6147750292877296e-07,
+      "loss": 1.2122,
+      "step": 11370
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.588820816030079e-07,
+      "loss": 1.1308,
+      "step": 11375
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5630752063377274e-07,
+      "loss": 1.1633,
+      "step": 11380
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5375382547840102e-07,
+      "loss": 1.211,
+      "step": 11385
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5122100154999597e-07,
+      "loss": 1.1444,
+      "step": 11390
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4870905421741967e-07,
+      "loss": 1.1885,
+      "step": 11395
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4621798880528505e-07,
+      "loss": 1.1674,
+      "step": 11400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4374781059393495e-07,
+      "loss": 1.2194,
+      "step": 11405
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4129852481944317e-07,
+      "loss": 1.1413,
+      "step": 11410
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.388701366735956e-07,
+      "loss": 1.1609,
+      "step": 11415
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.3646265130387916e-07,
+      "loss": 1.1603,
+      "step": 11420
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.3407607381347609e-07,
+      "loss": 1.1425,
+      "step": 11425
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.3171040926124757e-07,
+      "loss": 1.1537,
+      "step": 11430
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2936566266172568e-07,
+      "loss": 1.1574,
+      "step": 11435
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.270418389851047e-07,
+      "loss": 1.1733,
+      "step": 11440
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2473894315722434e-07,
+      "loss": 1.1765,
+      "step": 11445
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2245698005956762e-07,
+      "loss": 1.166,
+      "step": 11450
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.2019595452924193e-07,
+      "loss": 1.1591,
+      "step": 11455
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1795587135897568e-07,
+      "loss": 1.1618,
+      "step": 11460
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1573673529710506e-07,
+      "loss": 1.1745,
+      "step": 11465
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1353855104756395e-07,
+      "loss": 1.1854,
+      "step": 11470
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1136132326987403e-07,
+      "loss": 1.179,
+      "step": 11475
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0920505657913694e-07,
+      "loss": 1.1991,
+      "step": 11480
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0706975554601983e-07,
+      "loss": 1.1686,
+      "step": 11485
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0495542469675213e-07,
+      "loss": 1.1939,
+      "step": 11490
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.0286206851310987e-07,
+      "loss": 1.1729,
+      "step": 11495
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.0078969143241024e-07,
+      "loss": 1.1781,
+      "step": 11500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.873829784749933e-08,
+      "loss": 1.1544,
+      "step": 11505
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.670789210674547e-08,
+      "loss": 1.1651,
+      "step": 11510
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.469847851402924e-08,
+      "loss": 1.1744,
+      "step": 11515
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.27100613287324e-08,
+      "loss": 1.1868,
+      "step": 11520
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.074264476573003e-08,
+      "loss": 1.185,
+      "step": 11525
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.879623299538398e-08,
+      "loss": 1.1915,
+      "step": 11530
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.68708301435306e-08,
+      "loss": 1.1508,
+      "step": 11535
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.496644029147184e-08,
+      "loss": 1.1461,
+      "step": 11540
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.308306747597306e-08,
+      "loss": 1.1553,
+      "step": 11545
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.122071568924305e-08,
+      "loss": 1.156,
+      "step": 11550
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.937938887893626e-08,
+      "loss": 1.1598,
+      "step": 11555
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.75590909481394e-08,
+      "loss": 1.2028,
+      "step": 11560
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.575982575536267e-08,
+      "loss": 1.1864,
+      "step": 11565
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.398159711453634e-08,
+      "loss": 1.1595,
+      "step": 11570
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.222440879499415e-08,
+      "loss": 1.1541,
+      "step": 11575
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.048826452147329e-08,
+      "loss": 1.1588,
+      "step": 11580
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.877316797410549e-08,
+      "loss": 1.1519,
+      "step": 11585
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.707912278840267e-08,
+      "loss": 1.1602,
+      "step": 11590
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.540613255525796e-08,
+      "loss": 1.1866,
+      "step": 11595
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.37542008209302e-08,
+      "loss": 1.15,
+      "step": 11600
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.212333108704505e-08,
+      "loss": 1.1623,
+      "step": 11605
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.051352681057831e-08,
+      "loss": 1.1788,
+      "step": 11610
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.8924791403854876e-08,
+      "loss": 1.1751,
+      "step": 11615
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.735712823453976e-08,
+      "loss": 1.14,
+      "step": 11620
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.5810540625632625e-08,
+      "loss": 1.167,
+      "step": 11625
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.428503185545442e-08,
+      "loss": 1.2238,
+      "step": 11630
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.27806051576496e-08,
+      "loss": 1.1706,
+      "step": 11635
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.129726372117061e-08,
+      "loss": 1.1763,
+      "step": 11640
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9835010690278965e-08,
+      "loss": 1.1779,
+      "step": 11645
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.839384916453194e-08,
+      "loss": 1.1674,
+      "step": 11650
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.6973782198780346e-08,
+      "loss": 1.1617,
+      "step": 11655
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.5574812803160786e-08,
+      "loss": 1.1492,
+      "step": 11660
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.419694394308782e-08,
+      "loss": 1.1992,
+      "step": 11665
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.28401785392496e-08,
+      "loss": 1.2058,
+      "step": 11670
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1504519467601146e-08,
+      "loss": 1.1759,
+      "step": 11675
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.018996955935772e-08,
+      "loss": 1.1841,
+      "step": 11680
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.889653160098816e-08,
+      "loss": 1.1799,
+      "step": 11685
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.762420833421265e-08,
+      "loss": 1.164,
+      "step": 11690
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.6373002455992734e-08,
+      "loss": 1.1624,
+      "step": 11695
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.514291661852687e-08,
+      "loss": 1.1601,
+      "step": 11700
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.3933953429244884e-08,
+      "loss": 1.1445,
+      "step": 11705
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.274611545080353e-08,
+      "loss": 1.1838,
+      "step": 11710
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1579405201079826e-08,
+      "loss": 1.1666,
+      "step": 11715
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.043382515316551e-08,
+      "loss": 1.142,
+      "step": 11720
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9309377735362578e-08,
+      "loss": 1.1925,
+      "step": 11725
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8206065331179978e-08,
+      "loss": 1.1776,
+      "step": 11730
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.7123890279322495e-08,
+      "loss": 1.1628,
+      "step": 11735
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.6062854873691866e-08,
+      "loss": 1.1829,
+      "step": 11740
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.5022961363381227e-08,
+      "loss": 1.1725,
+      "step": 11745
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4004211952666223e-08,
+      "loss": 1.1593,
+      "step": 11750
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3006608801006136e-08,
+      "loss": 1.1767,
+      "step": 11755
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.203015402303166e-08,
+      "loss": 1.1831,
+      "step": 11760
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1074849688550446e-08,
+      "loss": 1.1479,
+      "step": 11765
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0140697822533807e-08,
+      "loss": 1.1891,
+      "step": 11770
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.922770040511557e-08,
+      "loss": 1.1339,
+      "step": 11775
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.833585937158988e-08,
+      "loss": 1.1524,
+      "step": 11780
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7465176612405656e-08,
+      "loss": 1.1724,
+      "step": 11785
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6615653973159895e-08,
+      "loss": 1.1897,
+      "step": 11790
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5787293254598822e-08,
+      "loss": 1.1795,
+      "step": 11795
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4980096212608985e-08,
+      "loss": 1.1562,
+      "step": 11800
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4194064558219478e-08,
+      "loss": 1.162,
+      "step": 11805
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3429199957590844e-08,
+      "loss": 1.1705,
+      "step": 11810
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2685504032019514e-08,
+      "loss": 1.1413,
+      "step": 11815
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1962978357925593e-08,
+      "loss": 1.145,
+      "step": 11820
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1261624466858411e-08,
+      "loss": 1.1927,
+      "step": 11825
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.058144384548765e-08,
+      "loss": 1.1835,
+      "step": 11830
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.922437935601104e-09,
+      "loss": 1.1355,
+      "step": 11835
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.28460813410359e-09,
+      "loss": 1.1966,
+      "step": 11840
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.667955793011384e-09,
+      "loss": 1.1331,
+      "step": 11845
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.072482219452227e-09,
+      "loss": 1.1918,
+      "step": 11850
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.498188675658658e-09,
+      "loss": 1.2045,
+      "step": 11855
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.945076378969129e-09,
+      "loss": 1.1174,
+      "step": 11860
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.413146501824674e-09,
+      "loss": 1.1947,
+      "step": 11865
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.902400171762246e-09,
+      "loss": 1.1244,
+      "step": 11870
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.412838471420267e-09,
+      "loss": 1.1453,
+      "step": 11875
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.944462438528641e-09,
+      "loss": 1.1657,
+      "step": 11880
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.497273065910968e-09,
+      "loss": 1.1785,
+      "step": 11885
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.071271301480107e-09,
+      "loss": 1.1535,
+      "step": 11890
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.6664580482392852e-09,
+      "loss": 1.1868,
+      "step": 11895
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.282834164275439e-09,
+      "loss": 1.1743,
+      "step": 11900
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9204004627625404e-09,
+      "loss": 1.182,
+      "step": 11905
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5791577119560484e-09,
+      "loss": 1.1752,
+      "step": 11910
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2591066351929093e-09,
+      "loss": 1.175,
+      "step": 11915
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9602479108904448e-09,
+      "loss": 1.1506,
+      "step": 11920
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6825821725430236e-09,
+      "loss": 1.171,
+      "step": 11925
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4261100087231694e-09,
+      "loss": 1.132,
+      "step": 11930
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1908319630782316e-09,
+      "loss": 1.153,
+      "step": 11935
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.76748534330385e-10,
+      "loss": 1.1679,
+      "step": 11940
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.838601762755194e-10,
+      "loss": 1.1599,
+      "step": 11945
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.121672977821292e-10,
+      "loss": 1.1823,
+      "step": 11950
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.616702627890934e-10,
+      "loss": 1.1515,
+      "step": 11955
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.3236939030789615e-10,
+      "loss": 1.1925,
+      "step": 11960
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.242649544192954e-10,
+      "loss": 1.1738,
+      "step": 11965
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3735718427332346e-10,
+      "loss": 1.1623,
+      "step": 11970
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.164626409039699e-11,
+      "loss": 1.1534,
+      "step": 11975
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7132333157986466e-11,
+      "loss": 1.1755,
+      "step": 11980
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.815485832836529e-12,
+      "loss": 1.1623,
+      "step": 11985
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 1.1927711963653564,
+      "eval_runtime": 1568.8559,
+      "eval_samples_per_second": 18.043,
+      "eval_steps_per_second": 1.128,
+      "step": 11988
+    },
+    {
+      "epoch": 3.0,
+      "step": 11988,
+      "total_flos": 4513156835573760.0,
+      "train_loss": 1.2212401726045408,
+      "train_runtime": 158201.6468,
+      "train_samples_per_second": 4.851,
+      "train_steps_per_second": 0.076
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 11988,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 100,
+  "total_flos": 4513156835573760.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}