diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,56563 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8047349566481491,
+  "eval_steps": 500,
+  "global_step": 46500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 11.0832,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 10.9995,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 10.8287,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 10.5347,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 10.1317,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 9.7568,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2e-05,
+      "loss": 9.4066,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 9.0828,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 8.7974,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 8.4785,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 8.2844,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 8.1159,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.8e-05,
+      "loss": 8.04,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.4e-05,
+      "loss": 8.0479,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 7.785,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 7.8489,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000102,
+      "loss": 7.7499,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 7.596,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 7.6188,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 7.5564,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 7.5576,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 7.5225,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000138,
+      "loss": 7.4858,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 7.2874,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015,
+      "loss": 7.3418,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000156,
+      "loss": 7.2647,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000162,
+      "loss": 7.1983,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000168,
+      "loss": 7.0973,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 7.1693,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 7.1801,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000186,
+      "loss": 7.0995,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 6.9925,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000198,
+      "loss": 6.9763,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000204,
+      "loss": 6.9261,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 6.9174,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 6.9063,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 6.8622,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 6.897,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000234,
+      "loss": 6.8112,
+      "step": 195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 6.7515,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 6.7586,
+      "step": 205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 6.5961,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000258,
+      "loss": 6.6598,
+      "step": 215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 6.6071,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00027,
+      "loss": 6.5918,
+      "step": 225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000276,
+      "loss": 6.539,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 6.4539,
+      "step": 235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 6.4418,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000294,
+      "loss": 6.5303,
+      "step": 245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0003,
+      "loss": 6.4021,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00030599999999999996,
+      "loss": 6.4768,
+      "step": 255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000312,
+      "loss": 6.3929,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000318,
+      "loss": 6.3891,
+      "step": 265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000324,
+      "loss": 6.4381,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00033,
+      "loss": 6.3544,
+      "step": 275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000336,
+      "loss": 6.355,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00034199999999999996,
+      "loss": 6.3141,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00034799999999999995,
+      "loss": 6.2884,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035399999999999993,
+      "loss": 6.3167,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 6.2576,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036599999999999995,
+      "loss": 6.2405,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000372,
+      "loss": 6.2546,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00037799999999999997,
+      "loss": 6.1438,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00038399999999999996,
+      "loss": 6.1532,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00039,
+      "loss": 6.1473,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000396,
+      "loss": 6.2154,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000402,
+      "loss": 6.0882,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000408,
+      "loss": 6.2324,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004139999999999999,
+      "loss": 6.2041,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00041999999999999996,
+      "loss": 6.0757,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042599999999999995,
+      "loss": 6.0776,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00043199999999999993,
+      "loss": 6.032,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00043799999999999997,
+      "loss": 6.1352,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00044399999999999995,
+      "loss": 5.9888,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00045,
+      "loss": 6.0221,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00045599999999999997,
+      "loss": 6.0419,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00046199999999999995,
+      "loss": 5.9985,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000468,
+      "loss": 5.8769,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000474,
+      "loss": 5.9902,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00047999999999999996,
+      "loss": 5.9365,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000486,
+      "loss": 5.895,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004919999999999999,
+      "loss": 5.9953,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000498,
+      "loss": 5.8948,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005039999999999999,
+      "loss": 5.9516,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005099999999999999,
+      "loss": 5.8119,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000516,
+      "loss": 5.8849,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000522,
+      "loss": 5.8657,
+      "step": 435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005279999999999999,
+      "loss": 5.7593,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000534,
+      "loss": 5.8779,
+      "step": 445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00054,
+      "loss": 5.8816,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005459999999999999,
+      "loss": 5.813,
+      "step": 455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000552,
+      "loss": 5.8118,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000558,
+      "loss": 5.7594,
+      "step": 465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005639999999999999,
+      "loss": 5.8195,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00057,
+      "loss": 5.7352,
+      "step": 475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005759999999999999,
+      "loss": 5.6983,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005819999999999999,
+      "loss": 5.7638,
+      "step": 485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000588,
+      "loss": 5.8061,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005939999999999999,
+      "loss": 5.8155,
+      "step": 495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006,
+      "loss": 5.667,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 5.787242889404297,
+      "eval_runtime": 151.0777,
+      "eval_samples_per_second": 12.186,
+      "eval_steps_per_second": 0.768,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999999887207687,
+      "loss": 5.7718,
+      "step": 505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999999548830758,
+      "loss": 5.673,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999998984869237,
+      "loss": 5.6713,
+      "step": 515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999998195323167,
+      "loss": 5.6457,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999997180192608,
+      "loss": 5.7369,
+      "step": 525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999995939477636,
+      "loss": 5.7537,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999994473178343,
+      "loss": 5.721,
+      "step": 535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999992781294842,
+      "loss": 5.6126,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999990863827257,
+      "loss": 5.5628,
+      "step": 545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999988720775734,
+      "loss": 5.5213,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999986352140434,
+      "loss": 5.5386,
+      "step": 555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999983757921536,
+      "loss": 5.6003,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999980938119233,
+      "loss": 5.5606,
+      "step": 565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999977892733739,
+      "loss": 5.6028,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999974621765282,
+      "loss": 5.6032,
+      "step": 575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999971125214107,
+      "loss": 5.5227,
+      "step": 580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999967403080478,
+      "loss": 5.5631,
+      "step": 585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999963455364676,
+      "loss": 5.5767,
+      "step": 590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999959282066996,
+      "loss": 5.5166,
+      "step": 595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999954883187754,
+      "loss": 5.5309,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999950258727278,
+      "loss": 5.4728,
+      "step": 605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999945408685917,
+      "loss": 5.4537,
+      "step": 610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999940333064035,
+      "loss": 5.4461,
+      "step": 615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999935031862015,
+      "loss": 5.4692,
+      "step": 620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999929505080256,
+      "loss": 5.4129,
+      "step": 625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999923752719172,
+      "loss": 5.4537,
+      "step": 630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999917774779196,
+      "loss": 5.3996,
+      "step": 635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999911571260778,
+      "loss": 5.4168,
+      "step": 640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999905142164384,
+      "loss": 5.4616,
+      "step": 645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999898487490498,
+      "loss": 5.4191,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999891607239619,
+      "loss": 5.3498,
+      "step": 655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999884501412266,
+      "loss": 5.5168,
+      "step": 660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999877170008972,
+      "loss": 5.4529,
+      "step": 665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999869613030289,
+      "loss": 5.4247,
+      "step": 670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999861830476786,
+      "loss": 5.3704,
+      "step": 675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999853822349047,
+      "loss": 5.314,
+      "step": 680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999845588647674,
+      "loss": 5.263,
+      "step": 685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999837129373288,
+      "loss": 5.3963,
+      "step": 690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999828444526523,
+      "loss": 5.3916,
+      "step": 695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999819534108033,
+      "loss": 5.3184,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999810398118487,
+      "loss": 5.4009,
+      "step": 705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999801036558574,
+      "loss": 5.3957,
+      "step": 710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999791449428996,
+      "loss": 5.2775,
+      "step": 715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999781636730476,
+      "loss": 5.2826,
+      "step": 720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999771598463749,
+      "loss": 5.2916,
+      "step": 725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999761334629572,
+      "loss": 5.3436,
+      "step": 730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999750845228717,
+      "loss": 5.3674,
+      "step": 735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999740130261971,
+      "loss": 5.2603,
+      "step": 740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999729189730139,
+      "loss": 5.2142,
+      "step": 745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999718023634048,
+      "loss": 5.3698,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999706631974534,
+      "loss": 5.2052,
+      "step": 755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999695014752454,
+      "loss": 5.2727,
+      "step": 760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999683171968683,
+      "loss": 5.2406,
+      "step": 765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000599967110362411,
+      "loss": 5.2282,
+      "step": 770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999658809719644,
+      "loss": 5.1632,
+      "step": 775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999646290256208,
+      "loss": 5.1701,
+      "step": 780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999633545234743,
+      "loss": 5.1785,
+      "step": 785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999620574656209,
+      "loss": 5.2017,
+      "step": 790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000599960737852158,
+      "loss": 5.2983,
+      "step": 795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000599959395683185,
+      "loss": 5.1982,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999580309588026,
+      "loss": 5.1416,
+      "step": 805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999566436791137,
+      "loss": 5.1912,
+      "step": 810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999552338442223,
+      "loss": 5.2655,
+      "step": 815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999538014542347,
+      "loss": 5.188,
+      "step": 820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999523465092583,
+      "loss": 5.1928,
+      "step": 825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999508690094028,
+      "loss": 5.1108,
+      "step": 830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000599949368954779,
+      "loss": 5.1586,
+      "step": 835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999478463455001,
+      "loss": 5.1363,
+      "step": 840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999463011816801,
+      "loss": 5.142,
+      "step": 845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999447334634358,
+      "loss": 5.1496,
+      "step": 850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999431431908844,
+      "loss": 5.052,
+      "step": 855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999415303641459,
+      "loss": 5.0762,
+      "step": 860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005999398949833416,
+      "loss": 5.1514,
+      "step": 865
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999382370485942,
+      "loss": 5.1176,
+      "step": 870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999365565600288,
+      "loss": 5.1016,
+      "step": 875
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999348535177713,
+      "loss": 5.1446,
+      "step": 880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00059993312792195,
+      "loss": 4.9957,
+      "step": 885
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999313797726947,
+      "loss": 5.1482,
+      "step": 890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999296090701366,
+      "loss": 5.0768,
+      "step": 895
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999278158144091,
+      "loss": 5.2004,
+      "step": 900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999260000056469,
+      "loss": 5.0756,
+      "step": 905
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999241616439867,
+      "loss": 5.1679,
+      "step": 910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999223007295666,
+      "loss": 5.0844,
+      "step": 915
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999204172625265,
+      "loss": 5.1273,
+      "step": 920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999185112430082,
+      "loss": 5.0067,
+      "step": 925
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999165826711547,
+      "loss": 5.0033,
+      "step": 930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999146315471115,
+      "loss": 5.0368,
+      "step": 935
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999126578710249,
+      "loss": 5.0261,
+      "step": 940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999106616430435,
+      "loss": 5.0727,
+      "step": 945
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999086428633173,
+      "loss": 5.0198,
+      "step": 950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999066015319981,
+      "loss": 5.0701,
+      "step": 955
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999045376492396,
+      "loss": 5.0427,
+      "step": 960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999024512151967,
+      "loss": 5.1675,
+      "step": 965
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005999003422300267,
+      "loss": 5.0997,
+      "step": 970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998982106938878,
+      "loss": 5.1069,
+      "step": 975
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998960566069404,
+      "loss": 5.0461,
+      "step": 980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998938799693464,
+      "loss": 5.0177,
+      "step": 985
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998916807812696,
+      "loss": 4.9937,
+      "step": 990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998894590428755,
+      "loss": 5.0079,
+      "step": 995
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998872147543309,
+      "loss": 4.9904,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 5.063974857330322,
+      "eval_runtime": 150.7726,
+      "eval_samples_per_second": 12.21,
+      "eval_steps_per_second": 0.769,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998849479158045,
+      "loss": 5.0673,
+      "step": 1005
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998826585274671,
+      "loss": 4.9611,
+      "step": 1010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998803465894905,
+      "loss": 5.0122,
+      "step": 1015
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998780121020487,
+      "loss": 5.0227,
+      "step": 1020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998756550653174,
+      "loss": 4.9941,
+      "step": 1025
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998732754794736,
+      "loss": 5.0426,
+      "step": 1030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998708733446962,
+      "loss": 4.9343,
+      "step": 1035
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000599868448661166,
+      "loss": 5.0083,
+      "step": 1040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998660014290653,
+      "loss": 5.0275,
+      "step": 1045
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998635316485781,
+      "loss": 5.0044,
+      "step": 1050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998610393198902,
+      "loss": 5.0977,
+      "step": 1055
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998585244431887,
+      "loss": 4.9482,
+      "step": 1060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998559870186629,
+      "loss": 5.007,
+      "step": 1065
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998534270465038,
+      "loss": 4.9826,
+      "step": 1070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998508445269035,
+      "loss": 4.9924,
+      "step": 1075
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998482394600566,
+      "loss": 4.9346,
+      "step": 1080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998456118461587,
+      "loss": 4.9776,
+      "step": 1085
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998429616854074,
+      "loss": 5.0336,
+      "step": 1090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998402889780021,
+      "loss": 4.9965,
+      "step": 1095
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998375937241438,
+      "loss": 4.9999,
+      "step": 1100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998348759240351,
+      "loss": 4.9897,
+      "step": 1105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998321355778803,
+      "loss": 4.904,
+      "step": 1110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998293726858855,
+      "loss": 5.0172,
+      "step": 1115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998265872482585,
+      "loss": 4.9544,
+      "step": 1120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998237792652088,
+      "loss": 4.9249,
+      "step": 1125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998209487369473,
+      "loss": 5.0203,
+      "step": 1130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998180956636872,
+      "loss": 4.9552,
+      "step": 1135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998152200456427,
+      "loss": 4.9835,
+      "step": 1140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998123218830302,
+      "loss": 4.9234,
+      "step": 1145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998094011760677,
+      "loss": 4.9115,
+      "step": 1150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998064579249748,
+      "loss": 4.914,
+      "step": 1155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998034921299726,
+      "loss": 4.9211,
+      "step": 1160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005998005037912842,
+      "loss": 4.9112,
+      "step": 1165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997974929091345,
+      "loss": 4.9139,
+      "step": 1170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997944594837497,
+      "loss": 4.9096,
+      "step": 1175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997914035153581,
+      "loss": 4.8065,
+      "step": 1180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997883250041892,
+      "loss": 4.9562,
+      "step": 1185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997852239504747,
+      "loss": 4.939,
+      "step": 1190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997821003544478,
+      "loss": 4.8459,
+      "step": 1195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997789542163432,
+      "loss": 4.8906,
+      "step": 1200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997757855363976,
+      "loss": 4.8547,
+      "step": 1205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997725943148494,
+      "loss": 4.9617,
+      "step": 1210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997693805519383,
+      "loss": 4.8849,
+      "step": 1215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997661442479061,
+      "loss": 4.8716,
+      "step": 1220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997628854029961,
+      "loss": 4.83,
+      "step": 1225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997596040174534,
+      "loss": 4.8464,
+      "step": 1230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997563000915248,
+      "loss": 4.8273,
+      "step": 1235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997529736254585,
+      "loss": 4.8679,
+      "step": 1240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997496246195049,
+      "loss": 4.8066,
+      "step": 1245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997462530739157,
+      "loss": 4.7826,
+      "step": 1250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997428589889444,
+      "loss": 4.8587,
+      "step": 1255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997394423648463,
+      "loss": 4.8974,
+      "step": 1260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997360032018783,
+      "loss": 4.8758,
+      "step": 1265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000599732541500299,
+      "loss": 4.8788,
+      "step": 1270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997290572603687,
+      "loss": 4.8967,
+      "step": 1275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997255504823492,
+      "loss": 4.8774,
+      "step": 1280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997220211665045,
+      "loss": 4.8542,
+      "step": 1285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997184693130999,
+      "loss": 4.7275,
+      "step": 1290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997148949224023,
+      "loss": 4.807,
+      "step": 1295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997112979946809,
+      "loss": 4.8433,
+      "step": 1300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997076785302057,
+      "loss": 4.9526,
+      "step": 1305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000599704036529249,
+      "loss": 4.8017,
+      "step": 1310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005997003719920847,
+      "loss": 4.8443,
+      "step": 1315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996966849189884,
+      "loss": 4.823,
+      "step": 1320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996929753102374,
+      "loss": 4.8656,
+      "step": 1325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996892431661104,
+      "loss": 4.7528,
+      "step": 1330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996854884868884,
+      "loss": 4.804,
+      "step": 1335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996817112728534,
+      "loss": 4.8494,
+      "step": 1340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996779115242896,
+      "loss": 4.8679,
+      "step": 1345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996740892414827,
+      "loss": 4.7677,
+      "step": 1350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996702444247201,
+      "loss": 4.7851,
+      "step": 1355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996663770742908,
+      "loss": 4.8101,
+      "step": 1360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996624871904858,
+      "loss": 4.8664,
+      "step": 1365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996585747735975,
+      "loss": 4.8919,
+      "step": 1370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996546398239201,
+      "loss": 4.7878,
+      "step": 1375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996506823417495,
+      "loss": 4.8748,
+      "step": 1380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996467023273832,
+      "loss": 4.8132,
+      "step": 1385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996426997811207,
+      "loss": 4.8323,
+      "step": 1390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996386747032627,
+      "loss": 4.7567,
+      "step": 1395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996346270941119,
+      "loss": 4.7993,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996305569539728,
+      "loss": 4.7315,
+      "step": 1405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996264642831516,
+      "loss": 4.8493,
+      "step": 1410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996223490819556,
+      "loss": 4.7694,
+      "step": 1415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996182113506947,
+      "loss": 4.7332,
+      "step": 1420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996140510896796,
+      "loss": 4.8223,
+      "step": 1425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996098682992235,
+      "loss": 4.7319,
+      "step": 1430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996056629796408,
+      "loss": 4.8006,
+      "step": 1435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005996014351312476,
+      "loss": 4.844,
+      "step": 1440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995971847543621,
+      "loss": 4.7525,
+      "step": 1445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995929118493034,
+      "loss": 4.7649,
+      "step": 1450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995886164163934,
+      "loss": 4.7703,
+      "step": 1455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995842984559547,
+      "loss": 4.8856,
+      "step": 1460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995799579683121,
+      "loss": 4.7844,
+      "step": 1465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995755949537919,
+      "loss": 4.7843,
+      "step": 1470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995712094127224,
+      "loss": 4.7452,
+      "step": 1475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995668013454332,
+      "loss": 4.7533,
+      "step": 1480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995623707522558,
+      "loss": 4.7856,
+      "step": 1485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995579176335233,
+      "loss": 4.8798,
+      "step": 1490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995534419895707,
+      "loss": 4.7446,
+      "step": 1495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995489438207344,
+      "loss": 4.7522,
+      "step": 1500
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.799483776092529,
+      "eval_runtime": 150.7804,
+      "eval_samples_per_second": 12.21,
+      "eval_steps_per_second": 0.769,
+      "step": 1500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995444231273526,
+      "loss": 4.6847,
+      "step": 1505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995398799097654,
+      "loss": 4.846,
+      "step": 1510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995353141683144,
+      "loss": 4.7079,
+      "step": 1515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995307259033428,
+      "loss": 4.7785,
+      "step": 1520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995261151151957,
+      "loss": 4.8516,
+      "step": 1525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995214818042198,
+      "loss": 4.7627,
+      "step": 1530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995168259707635,
+      "loss": 4.7492,
+      "step": 1535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995121476151769,
+      "loss": 4.767,
+      "step": 1540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995074467378117,
+      "loss": 4.7328,
+      "step": 1545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005995027233390215,
+      "loss": 4.622,
+      "step": 1550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994979774191614,
+      "loss": 4.7061,
+      "step": 1555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994932089785883,
+      "loss": 4.7718,
+      "step": 1560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994884180176608,
+      "loss": 4.8252,
+      "step": 1565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994836045367391,
+      "loss": 4.7833,
+      "step": 1570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994787685361852,
+      "loss": 4.7973,
+      "step": 1575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994739100163627,
+      "loss": 4.8151,
+      "step": 1580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994690289776368,
+      "loss": 4.6815,
+      "step": 1585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994641254203749,
+      "loss": 4.6592,
+      "step": 1590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994591993449455,
+      "loss": 4.6454,
+      "step": 1595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994542507517188,
+      "loss": 4.7652,
+      "step": 1600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994492796410673,
+      "loss": 4.6894,
+      "step": 1605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994442860133647,
+      "loss": 4.7193,
+      "step": 1610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994392698689862,
+      "loss": 4.6321,
+      "step": 1615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994342312083093,
+      "loss": 4.7376,
+      "step": 1620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994291700317127,
+      "loss": 4.7212,
+      "step": 1625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994240863395772,
+      "loss": 4.6763,
+      "step": 1630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994189801322849,
+      "loss": 4.6659,
+      "step": 1635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994138514102198,
+      "loss": 4.6895,
+      "step": 1640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994087001737674,
+      "loss": 4.7019,
+      "step": 1645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005994035264233154,
+      "loss": 4.6882,
+      "step": 1650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993983301592525,
+      "loss": 4.724,
+      "step": 1655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993931113819697,
+      "loss": 4.6239,
+      "step": 1660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993878700918592,
+      "loss": 4.5808,
+      "step": 1665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993826062893152,
+      "loss": 4.6235,
+      "step": 1670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993773199747335,
+      "loss": 4.6813,
+      "step": 1675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993720111485115,
+      "loss": 4.6841,
+      "step": 1680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993666798110488,
+      "loss": 4.6773,
+      "step": 1685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993613259627458,
+      "loss": 4.6088,
+      "step": 1690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993559496040054,
+      "loss": 4.592,
+      "step": 1695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993505507352317,
+      "loss": 4.7731,
+      "step": 1700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993451293568307,
+      "loss": 4.7196,
+      "step": 1705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993396854692102,
+      "loss": 4.6916,
+      "step": 1710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993342190727793,
+      "loss": 4.6838,
+      "step": 1715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993287301679494,
+      "loss": 4.6393,
+      "step": 1720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993232187551328,
+      "loss": 4.705,
+      "step": 1725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993176848347444,
+      "loss": 4.6213,
+      "step": 1730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993121284071999,
+      "loss": 4.7283,
+      "step": 1735
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993065494729173,
+      "loss": 4.6904,
+      "step": 1740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005993009480323162,
+      "loss": 4.7068,
+      "step": 1745
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005992953240858177,
+      "loss": 4.6864,
+      "step": 1750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005992896776338446,
+      "loss": 4.6971,
+      "step": 1755
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005992840086768216,
+      "loss": 4.7385,
+      "step": 1760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000599278317215175,
+      "loss": 4.6804,
+      "step": 1765
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005992726032493326,
+      "loss": 4.7191,
+      "step": 1770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005992668667797244,
+      "loss": 4.5827,
+      "step": 1775
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005992611078067813,
+      "loss": 4.7553,
+      "step": 1780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005992553263309367,
+      "loss": 4.6555,
+      "step": 1785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005992495223526253,
+      "loss": 4.5248,
+      "step": 1790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005992436958722834,
+      "loss": 4.4933,
+      "step": 1795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005992378468903489,
+      "loss": 4.622,
+      "step": 1800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005992319754072622,
+      "loss": 4.5995,
+      "step": 1805
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005992260814234644,
+      "loss": 4.6473,
+      "step": 1810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005992201649393989,
+      "loss": 4.6518,
+      "step": 1815
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005992142259555103,
+      "loss": 4.6183,
+      "step": 1820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005992082644722454,
+      "loss": 4.5708,
+      "step": 1825
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005992022804900525,
+      "loss": 4.5672,
+      "step": 1830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005991962740093813,
+      "loss": 4.5723,
+      "step": 1835
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005991902450306839,
+      "loss": 4.574,
+      "step": 1840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005991841935544133,
+      "loss": 4.6351,
+      "step": 1845
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005991781195810246,
+      "loss": 4.6235,
+      "step": 1850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005991720231109745,
+      "loss": 4.6356,
+      "step": 1855
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005991659041447217,
+      "loss": 4.6934,
+      "step": 1860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000599159762682726,
+      "loss": 4.6972,
+      "step": 1865
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005991535987254494,
+      "loss": 4.5976,
+      "step": 1870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005991474122733552,
+      "loss": 4.5597,
+      "step": 1875
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005991412033269087,
+      "loss": 4.552,
+      "step": 1880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005991349718865768,
+      "loss": 4.5887,
+      "step": 1885
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000599128717952828,
+      "loss": 4.6184,
+      "step": 1890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005991224415261328,
+      "loss": 4.6031,
+      "step": 1895
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005991161426069629,
+      "loss": 4.6395,
+      "step": 1900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000599109821195792,
+      "loss": 4.6755,
+      "step": 1905
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005991034772930955,
+      "loss": 4.5333,
+      "step": 1910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005990971108993503,
+      "loss": 4.5863,
+      "step": 1915
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005990907220150353,
+      "loss": 4.5263,
+      "step": 1920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005990843106406308,
+      "loss": 4.5766,
+      "step": 1925
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005990778767766188,
+      "loss": 4.6123,
+      "step": 1930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005990714204234833,
+      "loss": 4.5906,
+      "step": 1935
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005990649415817097,
+      "loss": 4.592,
+      "step": 1940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005990584402517853,
+      "loss": 4.5306,
+      "step": 1945
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005990519164341987,
+      "loss": 4.538,
+      "step": 1950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005990453701294406,
+      "loss": 4.7278,
+      "step": 1955
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005990388013380033,
+      "loss": 4.5994,
+      "step": 1960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005990322100603806,
+      "loss": 4.6766,
+      "step": 1965
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005990255962970684,
+      "loss": 4.5969,
+      "step": 1970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005990189600485637,
+      "loss": 4.5924,
+      "step": 1975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005990123013153657,
+      "loss": 4.6688,
+      "step": 1980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000599005620097975,
+      "loss": 4.6841,
+      "step": 1985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005989989163968941,
+      "loss": 4.6268,
+      "step": 1990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000598992190212627,
+      "loss": 4.6271,
+      "step": 1995
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005989854415456796,
+      "loss": 4.5934,
+      "step": 2000
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.622138977050781,
+      "eval_runtime": 150.27,
+      "eval_samples_per_second": 12.251,
+      "eval_steps_per_second": 0.772,
+      "step": 2000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005989786703965592,
+      "loss": 4.6169,
+      "step": 2005
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000598971876765775,
+      "loss": 4.5989,
+      "step": 2010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005989650606538379,
+      "loss": 4.5947,
+      "step": 2015
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005989582220612605,
+      "loss": 4.6156,
+      "step": 2020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005989513609885568,
+      "loss": 4.6182,
+      "step": 2025
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005989444774362429,
+      "loss": 4.6214,
+      "step": 2030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005989375714048363,
+      "loss": 4.6239,
+      "step": 2035
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005989306428948565,
+      "loss": 4.5639,
+      "step": 2040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005989236919068243,
+      "loss": 4.5897,
+      "step": 2045
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005989167184412624,
+      "loss": 4.6397,
+      "step": 2050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005989097224986952,
+      "loss": 4.6781,
+      "step": 2055
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005989027040796488,
+      "loss": 4.6692,
+      "step": 2060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005988956631846509,
+      "loss": 4.5066,
+      "step": 2065
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005988885998142308,
+      "loss": 4.5364,
+      "step": 2070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005988815139689199,
+      "loss": 4.5657,
+      "step": 2075
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005988744056492508,
+      "loss": 4.4681,
+      "step": 2080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005988672748557583,
+      "loss": 4.5448,
+      "step": 2085
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005988601215889782,
+      "loss": 4.5786,
+      "step": 2090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005988529458494488,
+      "loss": 4.5426,
+      "step": 2095
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005988457476377092,
+      "loss": 4.5494,
+      "step": 2100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005988385269543011,
+      "loss": 4.5845,
+      "step": 2105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005988312837997673,
+      "loss": 4.5108,
+      "step": 2110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005988240181746524,
+      "loss": 4.5646,
+      "step": 2115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005988167300795029,
+      "loss": 4.5582,
+      "step": 2120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005988094195148666,
+      "loss": 4.5773,
+      "step": 2125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005988020864812933,
+      "loss": 4.5119,
+      "step": 2130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005987947309793344,
+      "loss": 4.4035,
+      "step": 2135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005987873530095432,
+      "loss": 4.4899,
+      "step": 2140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005987799525724742,
+      "loss": 4.56,
+      "step": 2145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000598772529668684,
+      "loss": 4.4948,
+      "step": 2150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005987650842987307,
+      "loss": 4.4912,
+      "step": 2155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005987576164631743,
+      "loss": 4.6292,
+      "step": 2160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005987501261625762,
+      "loss": 4.563,
+      "step": 2165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005987426133974997,
+      "loss": 4.5965,
+      "step": 2170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005987350781685097,
+      "loss": 4.546,
+      "step": 2175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005987275204761729,
+      "loss": 4.6035,
+      "step": 2180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005987199403210574,
+      "loss": 4.6014,
+      "step": 2185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005987123377037333,
+      "loss": 4.5493,
+      "step": 2190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005987047126247723,
+      "loss": 4.5345,
+      "step": 2195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005986970650847478,
+      "loss": 4.5793,
+      "step": 2200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005986893950842347,
+      "loss": 4.5242,
+      "step": 2205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005986817026238099,
+      "loss": 4.5865,
+      "step": 2210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005986739877040518,
+      "loss": 4.5695,
+      "step": 2215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005986662503255405,
+      "loss": 4.5274,
+      "step": 2220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005986584904888579,
+      "loss": 4.5146,
+      "step": 2225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005986507081945872,
+      "loss": 4.4585,
+      "step": 2230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000598642903443314,
+      "loss": 4.5071,
+      "step": 2235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005986350762356249,
+      "loss": 4.6897,
+      "step": 2240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005986272265721086,
+      "loss": 4.5492,
+      "step": 2245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005986193544533553,
+      "loss": 4.4015,
+      "step": 2250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000598611459879957,
+      "loss": 4.5966,
+      "step": 2255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000598603542852507,
+      "loss": 4.5583,
+      "step": 2260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005985956033716011,
+      "loss": 4.4743,
+      "step": 2265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005985876414378361,
+      "loss": 4.557,
+      "step": 2270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005985796570518108,
+      "loss": 4.5134,
+      "step": 2275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005985716502141254,
+      "loss": 4.5188,
+      "step": 2280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005985636209253821,
+      "loss": 4.5209,
+      "step": 2285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005985555691861845,
+      "loss": 4.4911,
+      "step": 2290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005985474949971383,
+      "loss": 4.5194,
+      "step": 2295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005985393983588503,
+      "loss": 4.5059,
+      "step": 2300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005985312792719297,
+      "loss": 4.4679,
+      "step": 2305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005985231377369869,
+      "loss": 4.5583,
+      "step": 2310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000598514973754634,
+      "loss": 4.5782,
+      "step": 2315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005985067873254848,
+      "loss": 4.5104,
+      "step": 2320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005984985784501551,
+      "loss": 4.5072,
+      "step": 2325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005984903471292621,
+      "loss": 4.435,
+      "step": 2330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005984820933634247,
+      "loss": 4.4558,
+      "step": 2335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005984738171532636,
+      "loss": 4.5055,
+      "step": 2340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005984655184994011,
+      "loss": 4.521,
+      "step": 2345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005984571974024614,
+      "loss": 4.4612,
+      "step": 2350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005984488538630698,
+      "loss": 4.4435,
+      "step": 2355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005984404878818539,
+      "loss": 4.5881,
+      "step": 2360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005984320994594429,
+      "loss": 4.5542,
+      "step": 2365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005984236885964674,
+      "loss": 4.3867,
+      "step": 2370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00059841525529356,
+      "loss": 4.5786,
+      "step": 2375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005984067995513547,
+      "loss": 4.5679,
+      "step": 2380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005983983213704874,
+      "loss": 4.4752,
+      "step": 2385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005983898207515957,
+      "loss": 4.4446,
+      "step": 2390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005983812976953185,
+      "loss": 4.4343,
+      "step": 2395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000598372752202297,
+      "loss": 4.3717,
+      "step": 2400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005983641842731736,
+      "loss": 4.4689,
+      "step": 2405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005983555939085927,
+      "loss": 4.4949,
+      "step": 2410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005983469811092002,
+      "loss": 4.5136,
+      "step": 2415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005983383458756437,
+      "loss": 4.4673,
+      "step": 2420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005983296882085726,
+      "loss": 4.521,
+      "step": 2425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005983210081086378,
+      "loss": 4.4922,
+      "step": 2430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005983123055764921,
+      "loss": 4.5239,
+      "step": 2435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005983035806127897,
+      "loss": 4.4424,
+      "step": 2440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005982948332181871,
+      "loss": 4.4227,
+      "step": 2445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005982860633933415,
+      "loss": 4.5284,
+      "step": 2450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005982772711389128,
+      "loss": 4.5241,
+      "step": 2455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005982684564555619,
+      "loss": 4.4501,
+      "step": 2460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005982596193439517,
+      "loss": 4.4509,
+      "step": 2465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005982507598047466,
+      "loss": 4.4052,
+      "step": 2470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000598241877838613,
+      "loss": 4.4956,
+      "step": 2475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005982329734462186,
+      "loss": 4.3242,
+      "step": 2480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005982240466282329,
+      "loss": 4.392,
+      "step": 2485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005982150973853274,
+      "loss": 4.5077,
+      "step": 2490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005982061257181749,
+      "loss": 4.4919,
+      "step": 2495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00059819713162745,
+      "loss": 4.5178,
+      "step": 2500
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 4.496661186218262,
+      "eval_runtime": 150.774,
+      "eval_samples_per_second": 12.21,
+      "eval_steps_per_second": 0.769,
+      "step": 2500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005981881151138291,
+      "loss": 4.5157,
+      "step": 2505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00059817907617799,
+      "loss": 4.4359,
+      "step": 2510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005981700148206126,
+      "loss": 4.4987,
+      "step": 2515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005981609310423781,
+      "loss": 4.3829,
+      "step": 2520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005981518248439697,
+      "loss": 4.4533,
+      "step": 2525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005981426962260722,
+      "loss": 4.4787,
+      "step": 2530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005981335451893718,
+      "loss": 4.4843,
+      "step": 2535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005981243717345567,
+      "loss": 4.4322,
+      "step": 2540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005981151758623167,
+      "loss": 4.504,
+      "step": 2545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005981059575733432,
+      "loss": 4.4743,
+      "step": 2550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005980967168683297,
+      "loss": 4.5483,
+      "step": 2555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005980874537479706,
+      "loss": 4.4412,
+      "step": 2560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005980781682129627,
+      "loss": 4.4774,
+      "step": 2565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005980688602640042,
+      "loss": 4.3842,
+      "step": 2570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000598059529901795,
+      "loss": 4.453,
+      "step": 2575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005980501771270366,
+      "loss": 4.3691,
+      "step": 2580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005980408019404324,
+      "loss": 4.4649,
+      "step": 2585
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005980314043426874,
+      "loss": 4.4571,
+      "step": 2590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005980219843345082,
+      "loss": 4.2869,
+      "step": 2595
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000598012541916603,
+      "loss": 4.442,
+      "step": 2600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000598003077089682,
+      "loss": 4.4842,
+      "step": 2605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005979935898544568,
+      "loss": 4.4487,
+      "step": 2610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005979840802116409,
+      "loss": 4.4768,
+      "step": 2615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005979745481619493,
+      "loss": 4.4794,
+      "step": 2620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005979649937060988,
+      "loss": 4.4843,
+      "step": 2625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005979554168448078,
+      "loss": 4.4868,
+      "step": 2630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005979458175787963,
+      "loss": 4.52,
+      "step": 2635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005979361959087865,
+      "loss": 4.5309,
+      "step": 2640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005979265518355015,
+      "loss": 4.4523,
+      "step": 2645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005979168853596668,
+      "loss": 4.4153,
+      "step": 2650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000597907196482009,
+      "loss": 4.3197,
+      "step": 2655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005978974852032568,
+      "loss": 4.4264,
+      "step": 2660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005978877515241404,
+      "loss": 4.4645,
+      "step": 2665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005978779954453917,
+      "loss": 4.4296,
+      "step": 2670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005978682169677444,
+      "loss": 4.4271,
+      "step": 2675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005978584160919339,
+      "loss": 4.4266,
+      "step": 2680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005978485928186968,
+      "loss": 4.4608,
+      "step": 2685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000597838747148772,
+      "loss": 4.455,
+      "step": 2690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005978288790828998,
+      "loss": 4.4432,
+      "step": 2695
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005978189886218223,
+      "loss": 4.4275,
+      "step": 2700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005978090757662831,
+      "loss": 4.4242,
+      "step": 2705
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005977991405170277,
+      "loss": 4.4197,
+      "step": 2710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005977891828748031,
+      "loss": 4.3557,
+      "step": 2715
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005977792028403581,
+      "loss": 4.4138,
+      "step": 2720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005977692004144431,
+      "loss": 4.4321,
+      "step": 2725
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005977591755978104,
+      "loss": 4.3348,
+      "step": 2730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005977491283912135,
+      "loss": 4.3667,
+      "step": 2735
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005977390587954082,
+      "loss": 4.5263,
+      "step": 2740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005977289668111515,
+      "loss": 4.4123,
+      "step": 2745
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005977188524392025,
+      "loss": 4.4297,
+      "step": 2750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005977087156803214,
+      "loss": 4.4391,
+      "step": 2755
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005976985565352707,
+      "loss": 4.4199,
+      "step": 2760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005976883750048142,
+      "loss": 4.4723,
+      "step": 2765
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005976781710897175,
+      "loss": 4.3106,
+      "step": 2770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000597667944790748,
+      "loss": 4.4756,
+      "step": 2775
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005976576961086745,
+      "loss": 4.4701,
+      "step": 2780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005976474250442678,
+      "loss": 4.316,
+      "step": 2785
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005976371315983,
+      "loss": 4.3714,
+      "step": 2790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005976268157715453,
+      "loss": 4.2413,
+      "step": 2795
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005976164775647795,
+      "loss": 4.4715,
+      "step": 2800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005976061169787798,
+      "loss": 4.4679,
+      "step": 2805
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005975957340143254,
+      "loss": 4.3745,
+      "step": 2810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005975853286721968,
+      "loss": 4.348,
+      "step": 2815
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005975749009531767,
+      "loss": 4.4823,
+      "step": 2820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005975644508580489,
+      "loss": 4.3,
+      "step": 2825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005975539783875995,
+      "loss": 4.3079,
+      "step": 2830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000597543483542616,
+      "loss": 4.4583,
+      "step": 2835
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005975329663238872,
+      "loss": 4.3141,
+      "step": 2840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005975224267322043,
+      "loss": 4.4546,
+      "step": 2845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005975118647683597,
+      "loss": 4.4115,
+      "step": 2850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005975012804331476,
+      "loss": 4.3319,
+      "step": 2855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005974906737273638,
+      "loss": 4.3583,
+      "step": 2860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000597480044651806,
+      "loss": 4.3746,
+      "step": 2865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005974693932072734,
+      "loss": 4.3578,
+      "step": 2870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000597458719394567,
+      "loss": 4.4066,
+      "step": 2875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005974480232144891,
+      "loss": 4.3589,
+      "step": 2880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005974373046678445,
+      "loss": 4.4172,
+      "step": 2885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005974265637554389,
+      "loss": 4.4344,
+      "step": 2890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005974158004780799,
+      "loss": 4.3983,
+      "step": 2895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000597405014836577,
+      "loss": 4.3429,
+      "step": 2900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005973942068317411,
+      "loss": 4.3561,
+      "step": 2905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005973833764643851,
+      "loss": 4.3462,
+      "step": 2910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005973725237353231,
+      "loss": 4.3672,
+      "step": 2915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005973616486453715,
+      "loss": 4.4243,
+      "step": 2920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005973507511953478,
+      "loss": 4.4428,
+      "step": 2925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005973398313860714,
+      "loss": 4.3471,
+      "step": 2930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005973288892183636,
+      "loss": 4.2282,
+      "step": 2935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005973179246930472,
+      "loss": 4.3884,
+      "step": 2940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005973069378109466,
+      "loss": 4.3978,
+      "step": 2945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005972959285728877,
+      "loss": 4.4103,
+      "step": 2950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005972848969796988,
+      "loss": 4.3988,
+      "step": 2955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005972738430322092,
+      "loss": 4.4573,
+      "step": 2960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005972627667312501,
+      "loss": 4.3979,
+      "step": 2965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005972516680776543,
+      "loss": 4.3337,
+      "step": 2970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005972405470722565,
+      "loss": 4.4286,
+      "step": 2975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000597229403715893,
+      "loss": 4.3805,
+      "step": 2980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005972182380094015,
+      "loss": 4.3886,
+      "step": 2985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005972070499536217,
+      "loss": 4.3983,
+      "step": 2990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000597195839549395,
+      "loss": 4.4782,
+      "step": 2995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005971846067975642,
+      "loss": 4.3503,
+      "step": 3000
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 4.413268089294434,
+      "eval_runtime": 150.2737,
+      "eval_samples_per_second": 12.251,
+      "eval_steps_per_second": 0.772,
+      "step": 3000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000597173351698974,
+      "loss": 4.4155,
+      "step": 3005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005971620742544709,
+      "loss": 4.4261,
+      "step": 3010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005971507744649025,
+      "loss": 4.348,
+      "step": 3015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000597139452331119,
+      "loss": 4.4353,
+      "step": 3020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005971281078539712,
+      "loss": 4.3666,
+      "step": 3025
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005971167410343127,
+      "loss": 4.4247,
+      "step": 3030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005971053518729979,
+      "loss": 4.4093,
+      "step": 3035
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005970939403708832,
+      "loss": 4.4617,
+      "step": 3040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005970825065288269,
+      "loss": 4.3769,
+      "step": 3045
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005970710503476885,
+      "loss": 4.3765,
+      "step": 3050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005970595718283296,
+      "loss": 4.4443,
+      "step": 3055
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005970480709716134,
+      "loss": 4.4562,
+      "step": 3060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005970365477784046,
+      "loss": 4.4314,
+      "step": 3065
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005970250022495696,
+      "loss": 4.3781,
+      "step": 3070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005970134343859767,
+      "loss": 4.2969,
+      "step": 3075
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005970018441884957,
+      "loss": 4.3114,
+      "step": 3080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005969902316579982,
+      "loss": 4.4437,
+      "step": 3085
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005969785967953572,
+      "loss": 4.4663,
+      "step": 3090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005969669396014479,
+      "loss": 4.397,
+      "step": 3095
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005969552600771465,
+      "loss": 4.3915,
+      "step": 3100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005969435582233314,
+      "loss": 4.3023,
+      "step": 3105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005969318340408827,
+      "loss": 4.2951,
+      "step": 3110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005969200875306817,
+      "loss": 4.4636,
+      "step": 3115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005969083186936118,
+      "loss": 4.3918,
+      "step": 3120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000596896527530558,
+      "loss": 4.3059,
+      "step": 3125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000596884714042407,
+      "loss": 4.2898,
+      "step": 3130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005968728782300469,
+      "loss": 4.3249,
+      "step": 3135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005968610200943678,
+      "loss": 4.3538,
+      "step": 3140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005968491396362614,
+      "loss": 4.2482,
+      "step": 3145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000596837236856621,
+      "loss": 4.3515,
+      "step": 3150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005968253117563417,
+      "loss": 4.4214,
+      "step": 3155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005968133643363202,
+      "loss": 4.3378,
+      "step": 3160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005968013945974549,
+      "loss": 4.3318,
+      "step": 3165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005967894025406456,
+      "loss": 4.2822,
+      "step": 3170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0005967773881667944,
+      "loss": 4.2495,
+      "step": 3175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005967653514768046,
+      "loss": 4.268,
+      "step": 3180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005967532924715811,
+      "loss": 4.3754,
+      "step": 3185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000596741211152031,
+      "loss": 4.4025,
+      "step": 3190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005967291075190625,
+      "loss": 4.4083,
+      "step": 3195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005967169815735858,
+      "loss": 4.3213,
+      "step": 3200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005967048333165128,
+      "loss": 4.3461,
+      "step": 3205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005966926627487569,
+      "loss": 4.3768,
+      "step": 3210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005966804698712333,
+      "loss": 4.4196,
+      "step": 3215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005966682546848589,
+      "loss": 4.5157,
+      "step": 3220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000596656017190552,
+      "loss": 4.5326,
+      "step": 3225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005966437573892331,
+      "loss": 4.4102,
+      "step": 3230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005966314752818239,
+      "loss": 4.405,
+      "step": 3235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005966191708692478,
+      "loss": 4.4479,
+      "step": 3240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005966068441524304,
+      "loss": 4.2742,
+      "step": 3245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005965944951322984,
+      "loss": 4.4105,
+      "step": 3250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005965821238097803,
+      "loss": 4.3222,
+      "step": 3255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005965697301858064,
+      "loss": 4.3689,
+      "step": 3260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005965573142613088,
+      "loss": 4.3729,
+      "step": 3265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005965448760372209,
+      "loss": 4.3759,
+      "step": 3270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005965324155144782,
+      "loss": 4.3199,
+      "step": 3275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005965199326940174,
+      "loss": 4.352,
+      "step": 3280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005965074275767774,
+      "loss": 4.3888,
+      "step": 3285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005964949001636985,
+      "loss": 4.2492,
+      "step": 3290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005964823504557226,
+      "loss": 4.2244,
+      "step": 3295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005964697784537933,
+      "loss": 4.2883,
+      "step": 3300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005964571841588561,
+      "loss": 4.371,
+      "step": 3305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005964445675718579,
+      "loss": 4.334,
+      "step": 3310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005964319286937475,
+      "loss": 4.3165,
+      "step": 3315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005964192675254753,
+      "loss": 4.2529,
+      "step": 3320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005964065840679933,
+      "loss": 4.3381,
+      "step": 3325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005963938783222553,
+      "loss": 4.283,
+      "step": 3330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005963811502892165,
+      "loss": 4.3161,
+      "step": 3335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005963683999698342,
+      "loss": 4.3748,
+      "step": 3340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000596355627365067,
+      "loss": 4.3108,
+      "step": 3345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005963428324758755,
+      "loss": 4.3598,
+      "step": 3350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005963300153032217,
+      "loss": 4.3048,
+      "step": 3355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005963171758480695,
+      "loss": 4.3148,
+      "step": 3360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005963043141113842,
+      "loss": 4.1902,
+      "step": 3365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000596291430094133,
+      "loss": 4.2621,
+      "step": 3370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005962785237972848,
+      "loss": 4.3392,
+      "step": 3375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00059626559522181,
+      "loss": 4.4537,
+      "step": 3380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005962526443686807,
+      "loss": 4.3562,
+      "step": 3385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005962396712388709,
+      "loss": 4.3759,
+      "step": 3390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000596226675833356,
+      "loss": 4.3024,
+      "step": 3395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005962136581531132,
+      "loss": 4.3345,
+      "step": 3400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005962006181991214,
+      "loss": 4.3228,
+      "step": 3405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005961875559723611,
+      "loss": 4.235,
+      "step": 3410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005961744714738146,
+      "loss": 4.2656,
+      "step": 3415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005961613647044656,
+      "loss": 4.404,
+      "step": 3420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005961482356652998,
+      "loss": 4.2853,
+      "step": 3425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005961350843573044,
+      "loss": 4.294,
+      "step": 3430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005961219107814684,
+      "loss": 4.2044,
+      "step": 3435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005961087149387824,
+      "loss": 4.2688,
+      "step": 3440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005960954968302383,
+      "loss": 4.2032,
+      "step": 3445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005960822564568305,
+      "loss": 4.3575,
+      "step": 3450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005960689938195544,
+      "loss": 4.368,
+      "step": 3455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005960557089194072,
+      "loss": 4.2639,
+      "step": 3460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005960424017573881,
+      "loss": 4.3362,
+      "step": 3465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005960290723344974,
+      "loss": 4.2979,
+      "step": 3470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005960157206517376,
+      "loss": 4.3052,
+      "step": 3475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005960023467101128,
+      "loss": 4.3042,
+      "step": 3480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005959889505106284,
+      "loss": 4.2859,
+      "step": 3485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005959755320542919,
+      "loss": 4.3849,
+      "step": 3490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005959620913421123,
+      "loss": 4.3201,
+      "step": 3495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005959486283751001,
+      "loss": 4.41,
+      "step": 3500
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 4.344966411590576,
+      "eval_runtime": 150.7763,
+      "eval_samples_per_second": 12.21,
+      "eval_steps_per_second": 0.769,
+      "step": 3500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005959351431542678,
+      "loss": 4.3984,
+      "step": 3505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005959216356806294,
+      "loss": 4.3778,
+      "step": 3510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005959081059552005,
+      "loss": 4.3246,
+      "step": 3515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005958945539789987,
+      "loss": 4.2603,
+      "step": 3520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005958809797530427,
+      "loss": 4.2545,
+      "step": 3525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005958673832783534,
+      "loss": 4.268,
+      "step": 3530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005958537645559533,
+      "loss": 4.2577,
+      "step": 3535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005958401235868663,
+      "loss": 4.3378,
+      "step": 3540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005958264603721181,
+      "loss": 4.3697,
+      "step": 3545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005958127749127361,
+      "loss": 4.4162,
+      "step": 3550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005957990672097495,
+      "loss": 4.351,
+      "step": 3555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005957853372641891,
+      "loss": 4.3266,
+      "step": 3560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005957715850770871,
+      "loss": 4.2015,
+      "step": 3565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005957578106494778,
+      "loss": 4.1827,
+      "step": 3570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005957440139823968,
+      "loss": 4.4158,
+      "step": 3575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005957301950768816,
+      "loss": 4.2008,
+      "step": 3580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005957163539339714,
+      "loss": 4.1551,
+      "step": 3585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005957024905547068,
+      "loss": 4.2472,
+      "step": 3590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005956886049401305,
+      "loss": 4.2492,
+      "step": 3595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005956746970912863,
+      "loss": 4.2584,
+      "step": 3600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005956607670092204,
+      "loss": 4.3039,
+      "step": 3605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005956468146949799,
+      "loss": 4.2622,
+      "step": 3610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005956328401496141,
+      "loss": 4.2605,
+      "step": 3615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000595618843374174,
+      "loss": 4.2973,
+      "step": 3620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005956048243697117,
+      "loss": 4.218,
+      "step": 3625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005955907831372817,
+      "loss": 4.3062,
+      "step": 3630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005955767196779395,
+      "loss": 4.3318,
+      "step": 3635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005955626339927431,
+      "loss": 4.4257,
+      "step": 3640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000595548526082751,
+      "loss": 4.1505,
+      "step": 3645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005955343959490246,
+      "loss": 4.2879,
+      "step": 3650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005955202435926262,
+      "loss": 4.2549,
+      "step": 3655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00059550606901462,
+      "loss": 4.2972,
+      "step": 3660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005954918722160718,
+      "loss": 4.253,
+      "step": 3665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005954776531980492,
+      "loss": 4.3401,
+      "step": 3670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005954634119616215,
+      "loss": 4.2288,
+      "step": 3675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005954491485078592,
+      "loss": 4.2791,
+      "step": 3680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005954348628378353,
+      "loss": 4.2321,
+      "step": 3685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005954205549526236,
+      "loss": 4.2922,
+      "step": 3690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005954062248533003,
+      "loss": 4.2489,
+      "step": 3695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005953918725409429,
+      "loss": 4.3511,
+      "step": 3700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005953774980166304,
+      "loss": 4.3664,
+      "step": 3705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000595363101281444,
+      "loss": 4.2423,
+      "step": 3710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005953486823364659,
+      "loss": 4.2646,
+      "step": 3715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005953342411827807,
+      "loss": 4.2715,
+      "step": 3720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000595319777821474,
+      "loss": 4.3321,
+      "step": 3725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005953052922536335,
+      "loss": 4.2929,
+      "step": 3730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005952907844803486,
+      "loss": 4.2246,
+      "step": 3735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005952762545027099,
+      "loss": 4.2096,
+      "step": 3740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005952617023218101,
+      "loss": 4.312,
+      "step": 3745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0005952471279387435,
+      "loss": 4.2608,
+      "step": 3750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000595232531354606,
+      "loss": 4.3276,
+      "step": 3755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005952179125704952,
+      "loss": 4.335,
+      "step": 3760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005952032715875105,
+      "loss": 4.2907,
+      "step": 3765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005951886084067525,
+      "loss": 4.3741,
+      "step": 3770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005951739230293239,
+      "loss": 4.3517,
+      "step": 3775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005951592154563292,
+      "loss": 4.277,
+      "step": 3780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000595144485688874,
+      "loss": 4.3368,
+      "step": 3785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005951297337280662,
+      "loss": 4.2605,
+      "step": 3790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005951149595750149,
+      "loss": 4.3184,
+      "step": 3795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005951001632308311,
+      "loss": 4.3056,
+      "step": 3800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005950853446966273,
+      "loss": 4.3516,
+      "step": 3805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000595070503973518,
+      "loss": 4.3533,
+      "step": 3810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005950556410626189,
+      "loss": 4.2064,
+      "step": 3815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005950407559650477,
+      "loss": 4.2559,
+      "step": 3820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005950258486819239,
+      "loss": 4.347,
+      "step": 3825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005950109192143681,
+      "loss": 4.2478,
+      "step": 3830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005949959675635031,
+      "loss": 4.3066,
+      "step": 3835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005949809937304532,
+      "loss": 4.1078,
+      "step": 3840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005949659977163444,
+      "loss": 4.3285,
+      "step": 3845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005949509795223042,
+      "loss": 4.211,
+      "step": 3850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005949359391494619,
+      "loss": 4.2925,
+      "step": 3855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005949208765989485,
+      "loss": 4.2087,
+      "step": 3860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005949057918718966,
+      "loss": 4.3738,
+      "step": 3865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005948906849694406,
+      "loss": 4.234,
+      "step": 3870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005948755558927163,
+      "loss": 4.2358,
+      "step": 3875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005948604046428614,
+      "loss": 4.2694,
+      "step": 3880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005948452312210153,
+      "loss": 4.2782,
+      "step": 3885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005948300356283188,
+      "loss": 4.3455,
+      "step": 3890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005948148178659147,
+      "loss": 4.2975,
+      "step": 3895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005947995779349471,
+      "loss": 4.3128,
+      "step": 3900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005947843158365621,
+      "loss": 4.2444,
+      "step": 3905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005947690315719073,
+      "loss": 4.2566,
+      "step": 3910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005947537251421318,
+      "loss": 4.2086,
+      "step": 3915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000594738396548387,
+      "loss": 4.2687,
+      "step": 3920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005947230457918253,
+      "loss": 4.2127,
+      "step": 3925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005947076728736009,
+      "loss": 4.3346,
+      "step": 3930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005946922777948699,
+      "loss": 4.2569,
+      "step": 3935
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005946768605567898,
+      "loss": 4.327,
+      "step": 3940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005946614211605201,
+      "loss": 4.3121,
+      "step": 3945
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005946459596072216,
+      "loss": 4.1934,
+      "step": 3950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005946304758980569,
+      "loss": 4.0914,
+      "step": 3955
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005946149700341904,
+      "loss": 4.284,
+      "step": 3960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000594599442016788,
+      "loss": 4.2219,
+      "step": 3965
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005945838918470174,
+      "loss": 4.2636,
+      "step": 3970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005945683195260478,
+      "loss": 4.2416,
+      "step": 3975
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005945527250550503,
+      "loss": 4.1594,
+      "step": 3980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005945371084351975,
+      "loss": 4.2529,
+      "step": 3985
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005945214696676635,
+      "loss": 4.2743,
+      "step": 3990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005945058087536243,
+      "loss": 4.1218,
+      "step": 3995
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005944901256942578,
+      "loss": 4.2187,
+      "step": 4000
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 4.265633583068848,
+      "eval_runtime": 150.1723,
+      "eval_samples_per_second": 12.259,
+      "eval_steps_per_second": 0.772,
+      "step": 4000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000594474420490743,
+      "loss": 4.2666,
+      "step": 4005
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005944586931442609,
+      "loss": 4.12,
+      "step": 4010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005944429436559942,
+      "loss": 4.2916,
+      "step": 4015
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005944271720271272,
+      "loss": 4.1823,
+      "step": 4020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005944113782588456,
+      "loss": 4.2687,
+      "step": 4025
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005943955623523373,
+      "loss": 4.2593,
+      "step": 4030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005943797243087915,
+      "loss": 4.2273,
+      "step": 4035
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000594363864129399,
+      "loss": 4.2475,
+      "step": 4040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005943479818153526,
+      "loss": 4.2453,
+      "step": 4045
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005943320773678463,
+      "loss": 4.2077,
+      "step": 4050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005943161507880765,
+      "loss": 4.211,
+      "step": 4055
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005943002020772403,
+      "loss": 4.3064,
+      "step": 4060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005942842312365372,
+      "loss": 4.2258,
+      "step": 4065
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000594268238267168,
+      "loss": 4.1974,
+      "step": 4070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005942522231703357,
+      "loss": 4.3158,
+      "step": 4075
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000594236185947244,
+      "loss": 4.2207,
+      "step": 4080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005942201265990991,
+      "loss": 4.2394,
+      "step": 4085
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005942040451271085,
+      "loss": 4.2613,
+      "step": 4090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005941879415324815,
+      "loss": 4.2127,
+      "step": 4095
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000594171815816429,
+      "loss": 4.2353,
+      "step": 4100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005941556679801636,
+      "loss": 4.2169,
+      "step": 4105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005941394980248994,
+      "loss": 4.1401,
+      "step": 4110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005941233059518524,
+      "loss": 4.2061,
+      "step": 4115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005941070917622401,
+      "loss": 4.1151,
+      "step": 4120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000594090855457282,
+      "loss": 4.1882,
+      "step": 4125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005940745970381985,
+      "loss": 4.1692,
+      "step": 4130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005940583165062126,
+      "loss": 4.2102,
+      "step": 4135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005940420138625482,
+      "loss": 4.1483,
+      "step": 4140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005940256891084313,
+      "loss": 4.2406,
+      "step": 4145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005940093422450895,
+      "loss": 4.2565,
+      "step": 4150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000593992973273752,
+      "loss": 4.316,
+      "step": 4155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005939765821956495,
+      "loss": 4.27,
+      "step": 4160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005939601690120146,
+      "loss": 4.2913,
+      "step": 4165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005939437337240816,
+      "loss": 4.1995,
+      "step": 4170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005939272763330862,
+      "loss": 4.1578,
+      "step": 4175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000593910796840266,
+      "loss": 4.2062,
+      "step": 4180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005938942952468601,
+      "loss": 4.2138,
+      "step": 4185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005938777715541095,
+      "loss": 4.1583,
+      "step": 4190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005938612257632565,
+      "loss": 4.1554,
+      "step": 4195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005938446578755453,
+      "loss": 4.1384,
+      "step": 4200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005938280678922218,
+      "loss": 4.1932,
+      "step": 4205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005938114558145334,
+      "loss": 4.262,
+      "step": 4210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005937948216437294,
+      "loss": 4.3102,
+      "step": 4215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005937781653810603,
+      "loss": 4.1807,
+      "step": 4220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005937614870277788,
+      "loss": 4.1832,
+      "step": 4225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005937447865851391,
+      "loss": 4.2501,
+      "step": 4230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005937280640543968,
+      "loss": 4.2523,
+      "step": 4235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005937113194368095,
+      "loss": 4.1573,
+      "step": 4240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005936945527336362,
+      "loss": 4.1715,
+      "step": 4245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005936777639461375,
+      "loss": 4.2046,
+      "step": 4250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005936609530755761,
+      "loss": 4.2866,
+      "step": 4255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005936441201232162,
+      "loss": 4.1373,
+      "step": 4260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005936272650903232,
+      "loss": 4.2981,
+      "step": 4265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005936103879781647,
+      "loss": 4.1283,
+      "step": 4270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005935934887880099,
+      "loss": 4.2305,
+      "step": 4275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005935765675211292,
+      "loss": 4.1224,
+      "step": 4280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005935596241787953,
+      "loss": 4.2186,
+      "step": 4285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005935426587622821,
+      "loss": 4.2636,
+      "step": 4290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005935256712728653,
+      "loss": 4.1941,
+      "step": 4295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005935086617118224,
+      "loss": 4.1388,
+      "step": 4300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005934916300804322,
+      "loss": 4.1762,
+      "step": 4305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005934745763799756,
+      "loss": 4.2763,
+      "step": 4310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005934575006117349,
+      "loss": 4.2378,
+      "step": 4315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005934404027769941,
+      "loss": 4.2348,
+      "step": 4320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000593423282877039,
+      "loss": 4.203,
+      "step": 4325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0005934061409131567,
+      "loss": 4.2091,
+      "step": 4330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005933889768866362,
+      "loss": 4.1984,
+      "step": 4335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005933717907987682,
+      "loss": 4.2025,
+      "step": 4340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005933545826508453,
+      "loss": 4.2632,
+      "step": 4345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000593337352444161,
+      "loss": 4.2896,
+      "step": 4350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005933201001800111,
+      "loss": 4.2299,
+      "step": 4355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005933028258596931,
+      "loss": 4.1947,
+      "step": 4360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005932855294845055,
+      "loss": 4.176,
+      "step": 4365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005932682110557494,
+      "loss": 4.087,
+      "step": 4370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005932508705747266,
+      "loss": 4.1089,
+      "step": 4375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005932335080427413,
+      "loss": 4.1443,
+      "step": 4380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000593216123461099,
+      "loss": 4.2938,
+      "step": 4385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005931987168311069,
+      "loss": 4.2061,
+      "step": 4390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000593181288154074,
+      "loss": 4.1735,
+      "step": 4395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005931638374313107,
+      "loss": 4.248,
+      "step": 4400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005931463646641292,
+      "loss": 4.1558,
+      "step": 4405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005931288698538434,
+      "loss": 4.2654,
+      "step": 4410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000593111353001769,
+      "loss": 4.0433,
+      "step": 4415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000593093814109223,
+      "loss": 4.217,
+      "step": 4420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005930762531775242,
+      "loss": 4.1692,
+      "step": 4425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005930586702079931,
+      "loss": 4.0659,
+      "step": 4430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000593041065201952,
+      "loss": 4.219,
+      "step": 4435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005930234381607246,
+      "loss": 4.235,
+      "step": 4440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005930057890856364,
+      "loss": 4.1474,
+      "step": 4445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005929881179780144,
+      "loss": 4.2813,
+      "step": 4450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005929704248391875,
+      "loss": 4.1081,
+      "step": 4455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005929527096704862,
+      "loss": 4.1982,
+      "step": 4460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005929349724732424,
+      "loss": 4.2255,
+      "step": 4465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005929172132487899,
+      "loss": 4.1753,
+      "step": 4470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005928994319984643,
+      "loss": 4.2111,
+      "step": 4475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005928816287236023,
+      "loss": 4.3088,
+      "step": 4480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005928638034255429,
+      "loss": 4.3208,
+      "step": 4485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005928459561056264,
+      "loss": 4.2369,
+      "step": 4490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005928280867651947,
+      "loss": 4.252,
+      "step": 4495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005928101954055918,
+      "loss": 4.1711,
+      "step": 4500
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 4.222765922546387,
+      "eval_runtime": 150.6785,
+      "eval_samples_per_second": 12.218,
+      "eval_steps_per_second": 0.77,
+      "step": 4500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005927922820281627,
+      "loss": 4.2227,
+      "step": 4505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005927743466342545,
+      "loss": 4.1118,
+      "step": 4510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000592756389225216,
+      "loss": 4.1699,
+      "step": 4515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005927384098023973,
+      "loss": 4.1542,
+      "step": 4520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005927204083671505,
+      "loss": 4.1589,
+      "step": 4525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005927023849208291,
+      "loss": 4.1576,
+      "step": 4530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005926843394647884,
+      "loss": 4.1702,
+      "step": 4535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005926662720003855,
+      "loss": 4.1987,
+      "step": 4540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005926481825289788,
+      "loss": 4.1527,
+      "step": 4545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005926300710519285,
+      "loss": 4.147,
+      "step": 4550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005926119375705965,
+      "loss": 4.0946,
+      "step": 4555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005925937820863465,
+      "loss": 4.2065,
+      "step": 4560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005925756046005436,
+      "loss": 4.2049,
+      "step": 4565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005925574051145546,
+      "loss": 4.1496,
+      "step": 4570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005925391836297481,
+      "loss": 4.0682,
+      "step": 4575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005925209401474943,
+      "loss": 4.1447,
+      "step": 4580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005925026746691649,
+      "loss": 4.1023,
+      "step": 4585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005924843871961334,
+      "loss": 4.207,
+      "step": 4590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005924660777297749,
+      "loss": 4.1938,
+      "step": 4595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005924477462714664,
+      "loss": 4.1987,
+      "step": 4600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000592429392822586,
+      "loss": 4.2236,
+      "step": 4605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005924110173845139,
+      "loss": 4.2157,
+      "step": 4610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005923926199586319,
+      "loss": 4.0652,
+      "step": 4615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005923742005463235,
+      "loss": 4.1559,
+      "step": 4620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005923557591489735,
+      "loss": 4.1209,
+      "step": 4625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005923372957679688,
+      "loss": 4.0978,
+      "step": 4630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005923188104046975,
+      "loss": 4.1673,
+      "step": 4635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005923003030605499,
+      "loss": 4.1935,
+      "step": 4640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005922817737369175,
+      "loss": 4.0448,
+      "step": 4645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005922632224351936,
+      "loss": 3.9366,
+      "step": 4650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005922446491567733,
+      "loss": 4.1289,
+      "step": 4655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000592226053903053,
+      "loss": 4.1816,
+      "step": 4660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000592207436675431,
+      "loss": 4.1968,
+      "step": 4665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005921887974753075,
+      "loss": 4.2072,
+      "step": 4670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005921701363040837,
+      "loss": 4.3708,
+      "step": 4675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005921514531631631,
+      "loss": 4.1242,
+      "step": 4680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005921327480539505,
+      "loss": 4.2098,
+      "step": 4685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005921140209778524,
+      "loss": 4.0701,
+      "step": 4690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005920952719362769,
+      "loss": 4.2459,
+      "step": 4695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005920765009306341,
+      "loss": 4.1828,
+      "step": 4700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000592057707962335,
+      "loss": 4.0961,
+      "step": 4705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005920388930327932,
+      "loss": 4.1396,
+      "step": 4710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005920200561434234,
+      "loss": 4.191,
+      "step": 4715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005920011972956418,
+      "loss": 4.1557,
+      "step": 4720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005919823164908667,
+      "loss": 4.1885,
+      "step": 4725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005919634137305179,
+      "loss": 4.0797,
+      "step": 4730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005919444890160165,
+      "loss": 4.1649,
+      "step": 4735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005919255423487859,
+      "loss": 4.0826,
+      "step": 4740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005919065737302505,
+      "loss": 4.1142,
+      "step": 4745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005918875831618367,
+      "loss": 4.1676,
+      "step": 4750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005918685706449726,
+      "loss": 4.0972,
+      "step": 4755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005918495361810878,
+      "loss": 4.1348,
+      "step": 4760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005918304797716136,
+      "loss": 4.1432,
+      "step": 4765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005918114014179829,
+      "loss": 4.2454,
+      "step": 4770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005917923011216304,
+      "loss": 4.1003,
+      "step": 4775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005917731788839921,
+      "loss": 4.1819,
+      "step": 4780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005917540347065062,
+      "loss": 4.188,
+      "step": 4785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005917348685906121,
+      "loss": 4.1215,
+      "step": 4790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000591715680537751,
+      "loss": 4.1963,
+      "step": 4795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005916964705493656,
+      "loss": 4.1186,
+      "step": 4800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005916772386269007,
+      "loss": 4.1362,
+      "step": 4805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005916579847718022,
+      "loss": 4.2354,
+      "step": 4810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000591638708985518,
+      "loss": 4.1049,
+      "step": 4815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005916194112694976,
+      "loss": 4.1688,
+      "step": 4820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005916000916251918,
+      "loss": 4.0749,
+      "step": 4825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005915807500540537,
+      "loss": 4.1612,
+      "step": 4830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005915613865575374,
+      "loss": 4.1247,
+      "step": 4835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005915420011370992,
+      "loss": 4.2301,
+      "step": 4840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005915225937941965,
+      "loss": 4.1453,
+      "step": 4845
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005915031645302889,
+      "loss": 4.2323,
+      "step": 4850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005914837133468372,
+      "loss": 4.2572,
+      "step": 4855
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000591464240245304,
+      "loss": 4.1619,
+      "step": 4860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005914447452271537,
+      "loss": 4.1835,
+      "step": 4865
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005914252282938524,
+      "loss": 4.1756,
+      "step": 4870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005914056894468671,
+      "loss": 4.2158,
+      "step": 4875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005913861286876677,
+      "loss": 4.1847,
+      "step": 4880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005913665460177246,
+      "loss": 4.1548,
+      "step": 4885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005913469414385105,
+      "loss": 4.0973,
+      "step": 4890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005913273149514996,
+      "loss": 4.1362,
+      "step": 4895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005913076665581675,
+      "loss": 4.1658,
+      "step": 4900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000591287996259992,
+      "loss": 4.1478,
+      "step": 4905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0005912683040584519,
+      "loss": 4.1785,
+      "step": 4910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000591248589955028,
+      "loss": 4.2769,
+      "step": 4915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005912288539512029,
+      "loss": 4.1097,
+      "step": 4920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005912090960484603,
+      "loss": 4.1856,
+      "step": 4925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005911893162482862,
+      "loss": 4.2468,
+      "step": 4930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005911695145521679,
+      "loss": 4.2657,
+      "step": 4935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005911496909615943,
+      "loss": 4.0184,
+      "step": 4940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000591129845478056,
+      "loss": 4.0345,
+      "step": 4945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005911099781030454,
+      "loss": 4.0501,
+      "step": 4950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005910900888380564,
+      "loss": 4.1289,
+      "step": 4955
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005910701776845845,
+      "loss": 4.1621,
+      "step": 4960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005910502446441269,
+      "loss": 4.0817,
+      "step": 4965
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005910302897181826,
+      "loss": 4.1124,
+      "step": 4970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005910103129082519,
+      "loss": 4.1758,
+      "step": 4975
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005909903142158372,
+      "loss": 4.0607,
+      "step": 4980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005909702936424421,
+      "loss": 4.1467,
+      "step": 4985
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000590950251189572,
+      "loss": 4.2868,
+      "step": 4990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005909301868587343,
+      "loss": 4.1972,
+      "step": 4995
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005909101006514374,
+      "loss": 4.1593,
+      "step": 5000
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 4.196778774261475,
+      "eval_runtime": 150.8836,
+      "eval_samples_per_second": 12.201,
+      "eval_steps_per_second": 0.769,
+      "step": 5000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000590889992569192,
+      "loss": 4.1782,
+      "step": 5005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005908698626135098,
+      "loss": 4.1774,
+      "step": 5010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005908497107859046,
+      "loss": 4.1228,
+      "step": 5015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005908295370878918,
+      "loss": 4.1645,
+      "step": 5020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005908093415209883,
+      "loss": 4.1598,
+      "step": 5025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005907891240867127,
+      "loss": 4.033,
+      "step": 5030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005907688847865851,
+      "loss": 4.1476,
+      "step": 5035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005907486236221278,
+      "loss": 4.0968,
+      "step": 5040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005907283405948638,
+      "loss": 4.273,
+      "step": 5045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005907080357063187,
+      "loss": 4.1311,
+      "step": 5050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000590687708958019,
+      "loss": 3.9782,
+      "step": 5055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005906673603514936,
+      "loss": 4.1972,
+      "step": 5060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005906469898882721,
+      "loss": 4.0807,
+      "step": 5065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005906265975698867,
+      "loss": 4.0946,
+      "step": 5070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005906061833978705,
+      "loss": 4.1808,
+      "step": 5075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005905857473737586,
+      "loss": 4.0355,
+      "step": 5080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005905652894990878,
+      "loss": 4.2127,
+      "step": 5085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005905448097753963,
+      "loss": 4.0982,
+      "step": 5090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005905243082042242,
+      "loss": 4.1078,
+      "step": 5095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005905037847871129,
+      "loss": 4.1049,
+      "step": 5100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000590483239525606,
+      "loss": 4.1008,
+      "step": 5105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005904626724212481,
+      "loss": 4.0823,
+      "step": 5110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005904420834755858,
+      "loss": 4.0588,
+      "step": 5115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005904214726901672,
+      "loss": 4.1389,
+      "step": 5120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005904008400665424,
+      "loss": 4.1278,
+      "step": 5125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005903801856062627,
+      "loss": 4.1523,
+      "step": 5130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005903595093108812,
+      "loss": 4.23,
+      "step": 5135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005903388111819526,
+      "loss": 4.046,
+      "step": 5140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005903180912210335,
+      "loss": 4.0819,
+      "step": 5145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005902973494296816,
+      "loss": 4.1195,
+      "step": 5150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005902765858094569,
+      "loss": 4.0833,
+      "step": 5155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005902558003619205,
+      "loss": 4.2782,
+      "step": 5160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005902349930886357,
+      "loss": 4.1168,
+      "step": 5165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005902141639911666,
+      "loss": 4.1654,
+      "step": 5170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005901933130710797,
+      "loss": 4.0838,
+      "step": 5175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000590172440329943,
+      "loss": 4.0421,
+      "step": 5180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005901515457693258,
+      "loss": 4.1311,
+      "step": 5185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005901306293907995,
+      "loss": 4.0227,
+      "step": 5190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005901096911959365,
+      "loss": 4.2226,
+      "step": 5195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005900887311863118,
+      "loss": 4.1306,
+      "step": 5200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000590067749363501,
+      "loss": 4.0604,
+      "step": 5205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005900467457290821,
+      "loss": 4.0125,
+      "step": 5210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005900257202846344,
+      "loss": 4.1222,
+      "step": 5215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005900046730317388,
+      "loss": 4.1159,
+      "step": 5220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005899836039719782,
+      "loss": 4.1893,
+      "step": 5225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005899625131069367,
+      "loss": 4.1073,
+      "step": 5230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005899414004382001,
+      "loss": 4.0826,
+      "step": 5235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005899202659673563,
+      "loss": 4.0734,
+      "step": 5240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005898991096959941,
+      "loss": 4.1146,
+      "step": 5245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005898779316257047,
+      "loss": 4.0675,
+      "step": 5250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005898567317580804,
+      "loss": 4.0328,
+      "step": 5255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005898355100947154,
+      "loss": 4.0624,
+      "step": 5260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005898142666372055,
+      "loss": 4.0561,
+      "step": 5265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005897930013871478,
+      "loss": 4.0266,
+      "step": 5270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005897717143461418,
+      "loss": 3.9118,
+      "step": 5275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005897504055157878,
+      "loss": 4.1576,
+      "step": 5280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005897290748976883,
+      "loss": 4.0451,
+      "step": 5285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005897077224934471,
+      "loss": 4.2013,
+      "step": 5290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00058968634830467,
+      "loss": 4.1097,
+      "step": 5295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005896649523329642,
+      "loss": 4.1492,
+      "step": 5300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005896435345799383,
+      "loss": 4.0192,
+      "step": 5305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005896220950472031,
+      "loss": 4.1101,
+      "step": 5310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005896006337363707,
+      "loss": 4.1892,
+      "step": 5315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005895791506490548,
+      "loss": 4.1187,
+      "step": 5320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005895576457868708,
+      "loss": 4.0377,
+      "step": 5325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005895361191514358,
+      "loss": 4.0711,
+      "step": 5330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005895145707443684,
+      "loss": 4.0841,
+      "step": 5335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005894930005672892,
+      "loss": 4.0504,
+      "step": 5340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005894714086218199,
+      "loss": 4.1208,
+      "step": 5345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005894497949095841,
+      "loss": 4.1813,
+      "step": 5350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005894281594322073,
+      "loss": 4.1122,
+      "step": 5355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005894065021913161,
+      "loss": 4.0417,
+      "step": 5360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005893848231885391,
+      "loss": 4.0673,
+      "step": 5365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005893631224255066,
+      "loss": 3.9842,
+      "step": 5370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005893413999038501,
+      "loss": 4.0419,
+      "step": 5375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005893196556252034,
+      "loss": 4.0908,
+      "step": 5380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005892978895912011,
+      "loss": 4.1134,
+      "step": 5385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005892761018034803,
+      "loss": 4.0354,
+      "step": 5390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005892542922636791,
+      "loss": 4.1377,
+      "step": 5395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005892324609734373,
+      "loss": 4.1254,
+      "step": 5400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000589210607934397,
+      "loss": 4.0462,
+      "step": 5405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000589188733148201,
+      "loss": 4.1928,
+      "step": 5410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005891668366164945,
+      "loss": 4.109,
+      "step": 5415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005891449183409238,
+      "loss": 3.9687,
+      "step": 5420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005891229783231371,
+      "loss": 4.083,
+      "step": 5425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005891010165647841,
+      "loss": 4.0274,
+      "step": 5430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005890790330675162,
+      "loss": 4.0976,
+      "step": 5435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005890570278329867,
+      "loss": 4.0589,
+      "step": 5440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00058903500086285,
+      "loss": 4.1165,
+      "step": 5445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005890129521587625,
+      "loss": 4.1122,
+      "step": 5450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005889908817223822,
+      "loss": 4.094,
+      "step": 5455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005889687895553687,
+      "loss": 4.0328,
+      "step": 5460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005889466756593831,
+      "loss": 4.0535,
+      "step": 5465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005889245400360883,
+      "loss": 4.1173,
+      "step": 5470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005889023826871489,
+      "loss": 4.1297,
+      "step": 5475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005888802036142308,
+      "loss": 4.0227,
+      "step": 5480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0005888580028190019,
+      "loss": 4.1865,
+      "step": 5485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005888357803031317,
+      "loss": 4.0646,
+      "step": 5490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000588813536068291,
+      "loss": 4.0189,
+      "step": 5495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005887912701161524,
+      "loss": 4.0756,
+      "step": 5500
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 4.124629974365234,
+      "eval_runtime": 150.3809,
+      "eval_samples_per_second": 12.242,
+      "eval_steps_per_second": 0.771,
+      "step": 5500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005887689824483905,
+      "loss": 4.0788,
+      "step": 5505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000588746673066681,
+      "loss": 4.0131,
+      "step": 5510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005887243419727013,
+      "loss": 4.0943,
+      "step": 5515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000588701989168131,
+      "loss": 4.1031,
+      "step": 5520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005886796146546507,
+      "loss": 4.0249,
+      "step": 5525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005886572184339427,
+      "loss": 4.1008,
+      "step": 5530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005886348005076913,
+      "loss": 4.0013,
+      "step": 5535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005886123608775821,
+      "loss": 4.0819,
+      "step": 5540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005885898995453024,
+      "loss": 3.9572,
+      "step": 5545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005885674165125415,
+      "loss": 4.0845,
+      "step": 5550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005885449117809896,
+      "loss": 4.0497,
+      "step": 5555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000588522385352339,
+      "loss": 4.1316,
+      "step": 5560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005884998372282838,
+      "loss": 4.0639,
+      "step": 5565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005884772674105193,
+      "loss": 4.1424,
+      "step": 5570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005884546759007428,
+      "loss": 4.1344,
+      "step": 5575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000588432062700653,
+      "loss": 4.1834,
+      "step": 5580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005884094278119502,
+      "loss": 4.0697,
+      "step": 5585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005883867712363365,
+      "loss": 4.1778,
+      "step": 5590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005883640929755156,
+      "loss": 3.9826,
+      "step": 5595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005883413930311928,
+      "loss": 4.1734,
+      "step": 5600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005883186714050749,
+      "loss": 3.9475,
+      "step": 5605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005882959280988705,
+      "loss": 4.0677,
+      "step": 5610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005882731631142898,
+      "loss": 3.981,
+      "step": 5615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005882503764530447,
+      "loss": 4.1465,
+      "step": 5620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005882275681168485,
+      "loss": 4.1284,
+      "step": 5625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005882047381074163,
+      "loss": 4.128,
+      "step": 5630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005881818864264648,
+      "loss": 3.9485,
+      "step": 5635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005881590130757124,
+      "loss": 4.0819,
+      "step": 5640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000588136118056879,
+      "loss": 4.0794,
+      "step": 5645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005881132013716862,
+      "loss": 4.0057,
+      "step": 5650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005880902630218571,
+      "loss": 4.126,
+      "step": 5655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005880673030091169,
+      "loss": 4.1839,
+      "step": 5660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005880443213351917,
+      "loss": 4.0048,
+      "step": 5665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005880213180018097,
+      "loss": 3.9995,
+      "step": 5670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005879982930107007,
+      "loss": 4.047,
+      "step": 5675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000587975246363596,
+      "loss": 4.0557,
+      "step": 5680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005879521780622287,
+      "loss": 3.9903,
+      "step": 5685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005879290881083333,
+      "loss": 4.0698,
+      "step": 5690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005879059765036461,
+      "loss": 4.0808,
+      "step": 5695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000587882843249905,
+      "loss": 3.9814,
+      "step": 5700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005878596883488493,
+      "loss": 4.0462,
+      "step": 5705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005878365118022205,
+      "loss": 4.057,
+      "step": 5710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000587813313611761,
+      "loss": 4.0648,
+      "step": 5715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005877900937792154,
+      "loss": 4.0972,
+      "step": 5720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005877668523063297,
+      "loss": 4.1129,
+      "step": 5725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005877435891948514,
+      "loss": 4.055,
+      "step": 5730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005877203044465299,
+      "loss": 4.0274,
+      "step": 5735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005876969980631161,
+      "loss": 4.1068,
+      "step": 5740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005876736700463623,
+      "loss": 4.0272,
+      "step": 5745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000587650320398023,
+      "loss": 4.0066,
+      "step": 5750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005876269491198536,
+      "loss": 4.049,
+      "step": 5755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005876035562136119,
+      "loss": 3.9857,
+      "step": 5760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005875801416810566,
+      "loss": 3.9977,
+      "step": 5765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005875567055239485,
+      "loss": 3.978,
+      "step": 5770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005875332477440499,
+      "loss": 3.966,
+      "step": 5775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005875097683431245,
+      "loss": 3.9962,
+      "step": 5780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005874862673229381,
+      "loss": 4.0646,
+      "step": 5785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005874627446852578,
+      "loss": 4.0243,
+      "step": 5790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005874392004318521,
+      "loss": 3.9927,
+      "step": 5795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005874156345644919,
+      "loss": 4.0585,
+      "step": 5800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005873920470849488,
+      "loss": 4.0704,
+      "step": 5805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005873684379949967,
+      "loss": 4.1157,
+      "step": 5810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005873448072964108,
+      "loss": 4.0976,
+      "step": 5815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005873211549909679,
+      "loss": 4.0073,
+      "step": 5820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005872974810804467,
+      "loss": 3.999,
+      "step": 5825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005872737855666273,
+      "loss": 4.0376,
+      "step": 5830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005872500684512915,
+      "loss": 4.0754,
+      "step": 5835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005872263297362227,
+      "loss": 4.0787,
+      "step": 5840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000587202569423206,
+      "loss": 4.0246,
+      "step": 5845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005871787875140278,
+      "loss": 4.0841,
+      "step": 5850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005871549840104766,
+      "loss": 4.0973,
+      "step": 5855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005871311589143424,
+      "loss": 4.1085,
+      "step": 5860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005871073122274162,
+      "loss": 4.0267,
+      "step": 5865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005870834439514919,
+      "loss": 4.0323,
+      "step": 5870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005870595540883638,
+      "loss": 4.0608,
+      "step": 5875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005870356426398283,
+      "loss": 4.0103,
+      "step": 5880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005870117096076835,
+      "loss": 3.9965,
+      "step": 5885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005869877549937291,
+      "loss": 4.087,
+      "step": 5890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005869637787997663,
+      "loss": 4.0467,
+      "step": 5895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000586939781027598,
+      "loss": 4.0311,
+      "step": 5900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005869157616790287,
+      "loss": 4.0099,
+      "step": 5905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005868917207558646,
+      "loss": 4.0817,
+      "step": 5910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005868676582599134,
+      "loss": 3.966,
+      "step": 5915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005868435741929845,
+      "loss": 4.0285,
+      "step": 5920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005868194685568886,
+      "loss": 3.9462,
+      "step": 5925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005867953413534389,
+      "loss": 4.0755,
+      "step": 5930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005867711925844492,
+      "loss": 3.9712,
+      "step": 5935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005867470222517355,
+      "loss": 3.9844,
+      "step": 5940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005867228303571153,
+      "loss": 3.9981,
+      "step": 5945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005866986169024077,
+      "loss": 4.0011,
+      "step": 5950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005866743818894334,
+      "loss": 3.9929,
+      "step": 5955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005866501253200148,
+      "loss": 4.0131,
+      "step": 5960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005866258471959758,
+      "loss": 4.0858,
+      "step": 5965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005866015475191421,
+      "loss": 4.0778,
+      "step": 5970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005865772262913407,
+      "loss": 4.0723,
+      "step": 5975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005865528835144007,
+      "loss": 4.0553,
+      "step": 5980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005865285191901524,
+      "loss": 4.0564,
+      "step": 5985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005865041333204278,
+      "loss": 3.9492,
+      "step": 5990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005864797259070607,
+      "loss": 3.9973,
+      "step": 5995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005864552969518865,
+      "loss": 4.0681,
+      "step": 6000
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 4.073433876037598,
+      "eval_runtime": 149.973,
+      "eval_samples_per_second": 12.276,
+      "eval_steps_per_second": 0.773,
+      "step": 6000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005864308464567419,
+      "loss": 4.0337,
+      "step": 6005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005864063744234657,
+      "loss": 4.0838,
+      "step": 6010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005863818808538978,
+      "loss": 4.003,
+      "step": 6015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005863573657498803,
+      "loss": 4.0435,
+      "step": 6020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005863328291132564,
+      "loss": 4.0459,
+      "step": 6025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005863082709458711,
+      "loss": 4.0842,
+      "step": 6030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005862836912495713,
+      "loss": 4.0807,
+      "step": 6035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005862590900262049,
+      "loss": 4.1169,
+      "step": 6040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005862344672776221,
+      "loss": 4.1037,
+      "step": 6045
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005862098230056743,
+      "loss": 3.976,
+      "step": 6050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005861851572122146,
+      "loss": 4.0173,
+      "step": 6055
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005861604698990978,
+      "loss": 4.0725,
+      "step": 6060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0005861357610681802,
+      "loss": 4.0116,
+      "step": 6065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005861110307213197,
+      "loss": 4.0785,
+      "step": 6070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005860862788603761,
+      "loss": 4.0062,
+      "step": 6075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005860615054872104,
+      "loss": 4.0221,
+      "step": 6080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005860367106036854,
+      "loss": 4.0759,
+      "step": 6085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000586011894211666,
+      "loss": 4.0736,
+      "step": 6090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005859870563130177,
+      "loss": 4.1106,
+      "step": 6095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005859621969096085,
+      "loss": 3.976,
+      "step": 6100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005859373160033076,
+      "loss": 3.9863,
+      "step": 6105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005859124135959861,
+      "loss": 4.0505,
+      "step": 6110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005858874896895163,
+      "loss": 4.0183,
+      "step": 6115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005858625442857722,
+      "loss": 3.9544,
+      "step": 6120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00058583757738663,
+      "loss": 4.0514,
+      "step": 6125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005858125889939669,
+      "loss": 3.9714,
+      "step": 6130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005857875791096617,
+      "loss": 4.0567,
+      "step": 6135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005857625477355954,
+      "loss": 4.0103,
+      "step": 6140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00058573749487365,
+      "loss": 3.9502,
+      "step": 6145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005857124205257091,
+      "loss": 4.0495,
+      "step": 6150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005856873246936588,
+      "loss": 4.1271,
+      "step": 6155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005856622073793856,
+      "loss": 4.0907,
+      "step": 6160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005856370685847784,
+      "loss": 4.0462,
+      "step": 6165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005856119083117276,
+      "loss": 4.026,
+      "step": 6170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005855867265621249,
+      "loss": 4.027,
+      "step": 6175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005855615233378641,
+      "loss": 3.9725,
+      "step": 6180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005855362986408401,
+      "loss": 4.0381,
+      "step": 6185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005855110524729499,
+      "loss": 4.0856,
+      "step": 6190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005854857848360917,
+      "loss": 4.059,
+      "step": 6195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005854604957321657,
+      "loss": 4.0826,
+      "step": 6200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005854351851630732,
+      "loss": 4.0564,
+      "step": 6205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005854098531307177,
+      "loss": 3.918,
+      "step": 6210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000585384499637004,
+      "loss": 4.0235,
+      "step": 6215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005853591246838384,
+      "loss": 4.0396,
+      "step": 6220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005853337282731291,
+      "loss": 4.0262,
+      "step": 6225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005853083104067858,
+      "loss": 4.0257,
+      "step": 6230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005852828710867197,
+      "loss": 3.9727,
+      "step": 6235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005852574103148437,
+      "loss": 4.0742,
+      "step": 6240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005852319280930724,
+      "loss": 4.0149,
+      "step": 6245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000585206424423322,
+      "loss": 4.0074,
+      "step": 6250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00058518089930751,
+      "loss": 4.0104,
+      "step": 6255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000585155352747556,
+      "loss": 4.0947,
+      "step": 6260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005851297847453808,
+      "loss": 4.0355,
+      "step": 6265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000585104195302907,
+      "loss": 4.1322,
+      "step": 6270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000585078584422059,
+      "loss": 4.2945,
+      "step": 6275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005850529521047623,
+      "loss": 4.0278,
+      "step": 6280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005850272983529446,
+      "loss": 4.0641,
+      "step": 6285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005850016231685348,
+      "loss": 4.0774,
+      "step": 6290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005849759265534635,
+      "loss": 4.0438,
+      "step": 6295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000584950208509663,
+      "loss": 4.0762,
+      "step": 6300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000584924469039067,
+      "loss": 3.9799,
+      "step": 6305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005848987081436115,
+      "loss": 4.0686,
+      "step": 6310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005848729258252329,
+      "loss": 3.993,
+      "step": 6315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005848471220858705,
+      "loss": 3.9435,
+      "step": 6320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005848212969274643,
+      "loss": 4.056,
+      "step": 6325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005847954503519561,
+      "loss": 4.0116,
+      "step": 6330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005847695823612897,
+      "loss": 3.9138,
+      "step": 6335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005847436929574101,
+      "loss": 4.0694,
+      "step": 6340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000584717782142264,
+      "loss": 4.1302,
+      "step": 6345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005846918499177998,
+      "loss": 4.1207,
+      "step": 6350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005846658962859677,
+      "loss": 4.0034,
+      "step": 6355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000584639921248719,
+      "loss": 3.8664,
+      "step": 6360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000584613924808007,
+      "loss": 3.914,
+      "step": 6365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005845879069657865,
+      "loss": 3.9682,
+      "step": 6370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005845618677240138,
+      "loss": 3.9832,
+      "step": 6375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000584535807084647,
+      "loss": 3.9809,
+      "step": 6380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005845097250496457,
+      "loss": 3.9714,
+      "step": 6385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005844836216209714,
+      "loss": 3.992,
+      "step": 6390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005844574968005865,
+      "loss": 3.9963,
+      "step": 6395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005844313505904557,
+      "loss": 4.062,
+      "step": 6400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005844051829925451,
+      "loss": 3.9474,
+      "step": 6405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005843789940088223,
+      "loss": 3.9946,
+      "step": 6410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005843527836412565,
+      "loss": 4.0315,
+      "step": 6415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005843265518918185,
+      "loss": 4.0435,
+      "step": 6420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005843002987624812,
+      "loss": 4.0449,
+      "step": 6425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005842740242552182,
+      "loss": 3.9634,
+      "step": 6430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005842477283720058,
+      "loss": 4.1614,
+      "step": 6435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005842214111148207,
+      "loss": 4.0237,
+      "step": 6440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005841950724856422,
+      "loss": 4.0143,
+      "step": 6445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005841687124864506,
+      "loss": 4.0254,
+      "step": 6450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005841423311192283,
+      "loss": 3.9378,
+      "step": 6455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000584115928385959,
+      "loss": 4.0162,
+      "step": 6460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005840895042886278,
+      "loss": 4.0237,
+      "step": 6465
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005840630588292218,
+      "loss": 3.951,
+      "step": 6470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005840365920097296,
+      "loss": 3.9824,
+      "step": 6475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005840101038321414,
+      "loss": 3.8698,
+      "step": 6480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005839835942984488,
+      "loss": 3.9789,
+      "step": 6485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005839570634106455,
+      "loss": 3.9527,
+      "step": 6490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005839305111707262,
+      "loss": 4.0282,
+      "step": 6495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005839039375806875,
+      "loss": 4.0262,
+      "step": 6500
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 4.031991481781006,
+      "eval_runtime": 150.4726,
+      "eval_samples_per_second": 12.235,
+      "eval_steps_per_second": 0.771,
+      "step": 6500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005838773426425277,
+      "loss": 3.972,
+      "step": 6505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005838507263582467,
+      "loss": 3.908,
+      "step": 6510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005838240887298457,
+      "loss": 3.945,
+      "step": 6515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005837974297593278,
+      "loss": 3.9304,
+      "step": 6520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005837707494486976,
+      "loss": 4.0567,
+      "step": 6525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005837440477999613,
+      "loss": 4.0024,
+      "step": 6530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005837173248151269,
+      "loss": 4.0126,
+      "step": 6535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005836905804962036,
+      "loss": 3.9493,
+      "step": 6540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005836638148452027,
+      "loss": 4.0822,
+      "step": 6545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005836370278641365,
+      "loss": 3.9565,
+      "step": 6550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005836102195550195,
+      "loss": 4.0498,
+      "step": 6555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005835833899198675,
+      "loss": 4.138,
+      "step": 6560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005835565389606978,
+      "loss": 4.028,
+      "step": 6565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005835296666795297,
+      "loss": 3.8704,
+      "step": 6570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005835027730783836,
+      "loss": 4.0168,
+      "step": 6575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000583475858159282,
+      "loss": 3.9846,
+      "step": 6580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005834489219242486,
+      "loss": 3.9255,
+      "step": 6585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000583421964375309,
+      "loss": 3.9709,
+      "step": 6590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005833949855144901,
+      "loss": 3.971,
+      "step": 6595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005833679853438208,
+      "loss": 4.0187,
+      "step": 6600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005833409638653311,
+      "loss": 4.0978,
+      "step": 6605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000583313921081053,
+      "loss": 4.0233,
+      "step": 6610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005832868569930201,
+      "loss": 3.9346,
+      "step": 6615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005832597716032673,
+      "loss": 3.9806,
+      "step": 6620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005832326649138315,
+      "loss": 3.9501,
+      "step": 6625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005832055369267507,
+      "loss": 3.8945,
+      "step": 6630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000583178387644065,
+      "loss": 4.0411,
+      "step": 6635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005831512170678157,
+      "loss": 4.033,
+      "step": 6640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0005831240252000461,
+      "loss": 3.9629,
+      "step": 6645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005830968120428008,
+      "loss": 3.9754,
+      "step": 6650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005830695775981261,
+      "loss": 3.961,
+      "step": 6655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005830423218680698,
+      "loss": 3.8399,
+      "step": 6660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005830150448546816,
+      "loss": 3.8657,
+      "step": 6665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005829877465600122,
+      "loss": 3.9245,
+      "step": 6670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005829604269861149,
+      "loss": 4.0341,
+      "step": 6675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005829330861350433,
+      "loss": 3.9792,
+      "step": 6680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005829057240088538,
+      "loss": 4.0263,
+      "step": 6685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005828783406096037,
+      "loss": 3.9451,
+      "step": 6690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005828509359393522,
+      "loss": 3.9552,
+      "step": 6695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005828235100001597,
+      "loss": 4.0922,
+      "step": 6700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005827960627940888,
+      "loss": 3.9879,
+      "step": 6705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005827685943232034,
+      "loss": 4.0245,
+      "step": 6710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005827411045895687,
+      "loss": 4.0227,
+      "step": 6715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005827135935952521,
+      "loss": 3.8635,
+      "step": 6720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005826860613423221,
+      "loss": 3.9943,
+      "step": 6725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005826585078328489,
+      "loss": 4.047,
+      "step": 6730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005826309330689047,
+      "loss": 3.9673,
+      "step": 6735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005826033370525627,
+      "loss": 3.9391,
+      "step": 6740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005825757197858981,
+      "loss": 4.0063,
+      "step": 6745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005825480812709876,
+      "loss": 3.9123,
+      "step": 6750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005825204215099093,
+      "loss": 4.0236,
+      "step": 6755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005824927405047432,
+      "loss": 3.9006,
+      "step": 6760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005824650382575709,
+      "loss": 3.9333,
+      "step": 6765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005824373147704753,
+      "loss": 3.9388,
+      "step": 6770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000582409570045541,
+      "loss": 3.9877,
+      "step": 6775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005823818040848545,
+      "loss": 4.0653,
+      "step": 6780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005823540168905034,
+      "loss": 3.9874,
+      "step": 6785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005823262084645773,
+      "loss": 4.0155,
+      "step": 6790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005822983788091672,
+      "loss": 3.9023,
+      "step": 6795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005822705279263658,
+      "loss": 4.119,
+      "step": 6800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005822426558182674,
+      "loss": 3.99,
+      "step": 6805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005822147624869676,
+      "loss": 4.0494,
+      "step": 6810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000582186847934564,
+      "loss": 3.9168,
+      "step": 6815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005821589121631557,
+      "loss": 3.8806,
+      "step": 6820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005821309551748432,
+      "loss": 3.7961,
+      "step": 6825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005821029769717288,
+      "loss": 3.7826,
+      "step": 6830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005820749775559163,
+      "loss": 3.9815,
+      "step": 6835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000582046956929511,
+      "loss": 3.9875,
+      "step": 6840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005820189150946201,
+      "loss": 4.0475,
+      "step": 6845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005819908520533521,
+      "loss": 3.9757,
+      "step": 6850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005819627678078172,
+      "loss": 4.0318,
+      "step": 6855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005819346623601271,
+      "loss": 3.994,
+      "step": 6860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005819065357123955,
+      "loss": 4.0183,
+      "step": 6865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005818783878667371,
+      "loss": 3.8478,
+      "step": 6870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005818502188252684,
+      "loss": 3.9599,
+      "step": 6875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005818220285901078,
+      "loss": 3.9233,
+      "step": 6880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005817938171633749,
+      "loss": 3.9714,
+      "step": 6885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005817655845471912,
+      "loss": 4.0335,
+      "step": 6890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005817373307436795,
+      "loss": 3.9459,
+      "step": 6895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005817090557549644,
+      "loss": 3.8784,
+      "step": 6900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005816807595831721,
+      "loss": 3.9412,
+      "step": 6905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005816524422304303,
+      "loss": 3.8951,
+      "step": 6910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005816241036988683,
+      "loss": 3.9854,
+      "step": 6915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005815957439906169,
+      "loss": 3.8417,
+      "step": 6920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005815673631078088,
+      "loss": 3.9827,
+      "step": 6925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005815389610525778,
+      "loss": 4.0293,
+      "step": 6930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00058151053782706,
+      "loss": 3.9579,
+      "step": 6935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005814820934333925,
+      "loss": 4.0445,
+      "step": 6940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005814536278737141,
+      "loss": 3.9389,
+      "step": 6945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005814251411501652,
+      "loss": 3.947,
+      "step": 6950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005813966332648881,
+      "loss": 3.9215,
+      "step": 6955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005813681042200262,
+      "loss": 3.8809,
+      "step": 6960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005813395540177249,
+      "loss": 3.9399,
+      "step": 6965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000581310982660131,
+      "loss": 3.906,
+      "step": 6970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000581282390149393,
+      "loss": 3.9693,
+      "step": 6975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005812537764876606,
+      "loss": 3.9784,
+      "step": 6980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005812251416770857,
+      "loss": 4.0167,
+      "step": 6985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005811964857198214,
+      "loss": 4.0349,
+      "step": 6990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005811678086180225,
+      "loss": 3.9012,
+      "step": 6995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005811391103738454,
+      "loss": 4.0148,
+      "step": 7000
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 3.98817777633667,
+      "eval_runtime": 150.0793,
+      "eval_samples_per_second": 12.267,
+      "eval_steps_per_second": 0.773,
+      "step": 7000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000581110390989448,
+      "loss": 3.9029,
+      "step": 7005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005810816504669897,
+      "loss": 4.0553,
+      "step": 7010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000581052888808632,
+      "loss": 4.097,
+      "step": 7015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005810241060165374,
+      "loss": 3.9799,
+      "step": 7020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005809953020928703,
+      "loss": 3.9113,
+      "step": 7025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005809664770397964,
+      "loss": 3.9452,
+      "step": 7030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005809376308594835,
+      "loss": 3.9655,
+      "step": 7035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005809087635541005,
+      "loss": 3.8256,
+      "step": 7040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000580879875125818,
+      "loss": 3.9917,
+      "step": 7045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005808509655768086,
+      "loss": 3.9427,
+      "step": 7050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005808220349092458,
+      "loss": 4.0305,
+      "step": 7055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005807930831253052,
+      "loss": 3.9139,
+      "step": 7060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005807641102271637,
+      "loss": 3.9324,
+      "step": 7065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005807351162170001,
+      "loss": 3.9821,
+      "step": 7070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005807061010969944,
+      "loss": 3.9877,
+      "step": 7075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005806770648693286,
+      "loss": 4.0239,
+      "step": 7080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000580648007536186,
+      "loss": 3.9869,
+      "step": 7085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005806189290997514,
+      "loss": 3.9419,
+      "step": 7090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005805898295622115,
+      "loss": 4.0456,
+      "step": 7095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005805607089257545,
+      "loss": 3.9479,
+      "step": 7100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00058053156719257,
+      "loss": 3.9563,
+      "step": 7105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005805024043648493,
+      "loss": 3.9586,
+      "step": 7110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005804732204447854,
+      "loss": 3.9508,
+      "step": 7115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005804440154345728,
+      "loss": 3.9228,
+      "step": 7120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005804147893364073,
+      "loss": 3.9427,
+      "step": 7125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000580385542152487,
+      "loss": 3.9727,
+      "step": 7130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005803562738850107,
+      "loss": 3.8936,
+      "step": 7135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005803269845361794,
+      "loss": 3.8108,
+      "step": 7140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005802976741081956,
+      "loss": 3.943,
+      "step": 7145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005802683426032631,
+      "loss": 3.9068,
+      "step": 7150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005802389900235878,
+      "loss": 3.931,
+      "step": 7155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005802096163713764,
+      "loss": 4.0288,
+      "step": 7160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005801802216488381,
+      "loss": 3.8147,
+      "step": 7165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005801508058581829,
+      "loss": 3.8442,
+      "step": 7170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005801213690016228,
+      "loss": 3.9372,
+      "step": 7175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005800919110813716,
+      "loss": 4.0047,
+      "step": 7180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005800624320996439,
+      "loss": 3.9978,
+      "step": 7185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005800329320586568,
+      "loss": 3.9016,
+      "step": 7190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005800034109606282,
+      "loss": 3.9061,
+      "step": 7195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005799738688077781,
+      "loss": 3.9114,
+      "step": 7200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005799443056023279,
+      "loss": 3.8848,
+      "step": 7205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005799147213465006,
+      "loss": 3.9432,
+      "step": 7210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005798851160425209,
+      "loss": 3.9281,
+      "step": 7215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0005798554896926149,
+      "loss": 3.8136,
+      "step": 7220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005798258422990101,
+      "loss": 3.9658,
+      "step": 7225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005797961738639362,
+      "loss": 3.9506,
+      "step": 7230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000579766484389624,
+      "loss": 4.0028,
+      "step": 7235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005797367738783059,
+      "loss": 3.9895,
+      "step": 7240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000579707042332216,
+      "loss": 3.9246,
+      "step": 7245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00057967728975359,
+      "loss": 3.9121,
+      "step": 7250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005796475161446651,
+      "loss": 3.8231,
+      "step": 7255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005796177215076801,
+      "loss": 3.9446,
+      "step": 7260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005795879058448756,
+      "loss": 3.9487,
+      "step": 7265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005795580691584934,
+      "loss": 3.9777,
+      "step": 7270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005795282114507772,
+      "loss": 3.8589,
+      "step": 7275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005794983327239719,
+      "loss": 3.9822,
+      "step": 7280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005794684329803244,
+      "loss": 3.9954,
+      "step": 7285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000579438512222083,
+      "loss": 3.9695,
+      "step": 7290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005794085704514977,
+      "loss": 3.9504,
+      "step": 7295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005793786076708198,
+      "loss": 3.9779,
+      "step": 7300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005793486238823024,
+      "loss": 3.8648,
+      "step": 7305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005793186190882001,
+      "loss": 3.9049,
+      "step": 7310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000579288593290769,
+      "loss": 3.9022,
+      "step": 7315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005792585464922673,
+      "loss": 3.9056,
+      "step": 7320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005792284786949538,
+      "loss": 3.9115,
+      "step": 7325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005791983899010899,
+      "loss": 3.953,
+      "step": 7330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005791682801129379,
+      "loss": 3.9481,
+      "step": 7335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000579138149332762,
+      "loss": 3.8981,
+      "step": 7340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005791079975628277,
+      "loss": 4.0269,
+      "step": 7345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005790778248054025,
+      "loss": 3.9883,
+      "step": 7350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005790476310627551,
+      "loss": 3.9876,
+      "step": 7355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000579017416337156,
+      "loss": 3.988,
+      "step": 7360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000578987180630877,
+      "loss": 3.9376,
+      "step": 7365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005789569239461918,
+      "loss": 3.9417,
+      "step": 7370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005789266462853757,
+      "loss": 4.0789,
+      "step": 7375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005788963476507052,
+      "loss": 3.9385,
+      "step": 7380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005788660280444586,
+      "loss": 3.929,
+      "step": 7385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005788356874689159,
+      "loss": 4.0291,
+      "step": 7390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005788053259263587,
+      "loss": 3.8872,
+      "step": 7395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005787749434190695,
+      "loss": 3.8675,
+      "step": 7400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005787445399493335,
+      "loss": 4.0703,
+      "step": 7405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005787141155194365,
+      "loss": 3.9236,
+      "step": 7410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005786836701316665,
+      "loss": 3.9127,
+      "step": 7415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005786532037883128,
+      "loss": 3.8897,
+      "step": 7420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005786227164916662,
+      "loss": 3.9272,
+      "step": 7425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005785922082440192,
+      "loss": 3.9871,
+      "step": 7430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005785616790476659,
+      "loss": 3.9186,
+      "step": 7435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005785311289049019,
+      "loss": 3.974,
+      "step": 7440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005785005578180245,
+      "loss": 3.9538,
+      "step": 7445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005784699657893325,
+      "loss": 4.0123,
+      "step": 7450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005784393528211261,
+      "loss": 3.9068,
+      "step": 7455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005784087189157074,
+      "loss": 3.7432,
+      "step": 7460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005783780640753798,
+      "loss": 3.9067,
+      "step": 7465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005783473883024486,
+      "loss": 3.9339,
+      "step": 7470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005783166915992202,
+      "loss": 3.8964,
+      "step": 7475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000578285973968003,
+      "loss": 3.8972,
+      "step": 7480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005782552354111067,
+      "loss": 3.9025,
+      "step": 7485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005782244759308427,
+      "loss": 3.979,
+      "step": 7490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005781936955295241,
+      "loss": 3.9776,
+      "step": 7495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005781628942094653,
+      "loss": 3.9969,
+      "step": 7500
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 3.961434841156006,
+      "eval_runtime": 149.9847,
+      "eval_samples_per_second": 12.275,
+      "eval_steps_per_second": 0.773,
+      "step": 7500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005781320719729822,
+      "loss": 3.9575,
+      "step": 7505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005781012288223929,
+      "loss": 3.8435,
+      "step": 7510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005780703647600165,
+      "loss": 3.9905,
+      "step": 7515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005780394797881735,
+      "loss": 3.8025,
+      "step": 7520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005780085739091867,
+      "loss": 3.9212,
+      "step": 7525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00057797764712538,
+      "loss": 3.8277,
+      "step": 7530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005779466994390788,
+      "loss": 3.8959,
+      "step": 7535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005779157308526102,
+      "loss": 3.9607,
+      "step": 7540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000577884741368303,
+      "loss": 3.9904,
+      "step": 7545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005778537309884873,
+      "loss": 4.0796,
+      "step": 7550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000577822699715495,
+      "loss": 3.893,
+      "step": 7555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005777916475516597,
+      "loss": 3.953,
+      "step": 7560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005777605744993159,
+      "loss": 3.9167,
+      "step": 7565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005777294805608006,
+      "loss": 3.8872,
+      "step": 7570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005776983657384515,
+      "loss": 3.9342,
+      "step": 7575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005776672300346086,
+      "loss": 3.9996,
+      "step": 7580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000577636073451613,
+      "loss": 3.9381,
+      "step": 7585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005776048959918076,
+      "loss": 4.0558,
+      "step": 7590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005775736976575367,
+      "loss": 3.904,
+      "step": 7595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005775424784511462,
+      "loss": 3.9933,
+      "step": 7600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005775112383749839,
+      "loss": 3.995,
+      "step": 7605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005774799774313985,
+      "loss": 3.9822,
+      "step": 7610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000577448695622741,
+      "loss": 3.7759,
+      "step": 7615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005774173929513633,
+      "loss": 3.7732,
+      "step": 7620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005773860694196195,
+      "loss": 3.8441,
+      "step": 7625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000577354725029865,
+      "loss": 3.9408,
+      "step": 7630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005773233597844564,
+      "loss": 3.9715,
+      "step": 7635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005772919736857524,
+      "loss": 3.8511,
+      "step": 7640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005772605667361131,
+      "loss": 3.9275,
+      "step": 7645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005772291389379002,
+      "loss": 3.8361,
+      "step": 7650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005771976902934768,
+      "loss": 3.8592,
+      "step": 7655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005771662208052077,
+      "loss": 3.8605,
+      "step": 7660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005771347304754593,
+      "loss": 3.885,
+      "step": 7665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005771032193065994,
+      "loss": 3.8989,
+      "step": 7670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005770716873009975,
+      "loss": 3.9576,
+      "step": 7675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005770401344610248,
+      "loss": 3.8623,
+      "step": 7680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005770085607890537,
+      "loss": 3.8787,
+      "step": 7685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005769769662874585,
+      "loss": 3.9269,
+      "step": 7690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000576945350958615,
+      "loss": 3.947,
+      "step": 7695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005769137148049002,
+      "loss": 3.869,
+      "step": 7700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005768820578286935,
+      "loss": 3.9183,
+      "step": 7705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005768503800323748,
+      "loss": 3.94,
+      "step": 7710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005768186814183265,
+      "loss": 3.7839,
+      "step": 7715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000576786961988932,
+      "loss": 3.8936,
+      "step": 7720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005767552217465765,
+      "loss": 3.9323,
+      "step": 7725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005767234606936465,
+      "loss": 3.8951,
+      "step": 7730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005766916788325306,
+      "loss": 3.8801,
+      "step": 7735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005766598761656186,
+      "loss": 3.9485,
+      "step": 7740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005766280526953016,
+      "loss": 3.9617,
+      "step": 7745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005765962084239727,
+      "loss": 4.004,
+      "step": 7750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005765643433540265,
+      "loss": 3.9779,
+      "step": 7755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005765324574878592,
+      "loss": 3.8105,
+      "step": 7760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005765005508278681,
+      "loss": 3.9296,
+      "step": 7765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005764686233764527,
+      "loss": 3.9916,
+      "step": 7770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005764366751360137,
+      "loss": 3.8613,
+      "step": 7775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005764047061089534,
+      "loss": 3.8955,
+      "step": 7780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005763727162976758,
+      "loss": 3.9348,
+      "step": 7785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005763407057045863,
+      "loss": 3.9172,
+      "step": 7790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000576308674332092,
+      "loss": 3.8628,
+      "step": 7795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0005762766221826015,
+      "loss": 3.9669,
+      "step": 7800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005762445492585247,
+      "loss": 3.9162,
+      "step": 7805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005762124555622737,
+      "loss": 3.9174,
+      "step": 7810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005761803410962615,
+      "loss": 3.7882,
+      "step": 7815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005761482058629032,
+      "loss": 3.9603,
+      "step": 7820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005761160498646149,
+      "loss": 3.8465,
+      "step": 7825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005760838731038147,
+      "loss": 3.8924,
+      "step": 7830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005760516755829225,
+      "loss": 3.9661,
+      "step": 7835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005760194573043586,
+      "loss": 3.8856,
+      "step": 7840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005759872182705463,
+      "loss": 3.9037,
+      "step": 7845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005759549584839096,
+      "loss": 3.8099,
+      "step": 7850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005759226779468742,
+      "loss": 3.9186,
+      "step": 7855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005758903766618676,
+      "loss": 3.8689,
+      "step": 7860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005758580546313185,
+      "loss": 3.9651,
+      "step": 7865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005758257118576574,
+      "loss": 3.8181,
+      "step": 7870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005757933483433165,
+      "loss": 3.9611,
+      "step": 7875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000575760964090729,
+      "loss": 3.8883,
+      "step": 7880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005757285591023305,
+      "loss": 3.9339,
+      "step": 7885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005756961333805574,
+      "loss": 3.9402,
+      "step": 7890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005756636869278478,
+      "loss": 3.9722,
+      "step": 7895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005756312197466418,
+      "loss": 3.8141,
+      "step": 7900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005755987318393807,
+      "loss": 3.851,
+      "step": 7905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005755662232085075,
+      "loss": 3.9061,
+      "step": 7910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005755336938564663,
+      "loss": 3.846,
+      "step": 7915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005755011437857037,
+      "loss": 3.8962,
+      "step": 7920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005754685729986668,
+      "loss": 3.9421,
+      "step": 7925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000575435981497805,
+      "loss": 3.8796,
+      "step": 7930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005754033692855691,
+      "loss": 3.8234,
+      "step": 7935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005753707363644111,
+      "loss": 3.8632,
+      "step": 7940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000575338082736785,
+      "loss": 3.8991,
+      "step": 7945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005753054084051462,
+      "loss": 4.0148,
+      "step": 7950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005752727133719516,
+      "loss": 3.9543,
+      "step": 7955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005752399976396596,
+      "loss": 3.8603,
+      "step": 7960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005752072612107306,
+      "loss": 3.8823,
+      "step": 7965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005751745040876258,
+      "loss": 3.8419,
+      "step": 7970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005751417262728085,
+      "loss": 3.9414,
+      "step": 7975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005751089277687436,
+      "loss": 3.911,
+      "step": 7980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005750761085778972,
+      "loss": 3.9677,
+      "step": 7985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005750432687027372,
+      "loss": 3.9097,
+      "step": 7990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000575010408145733,
+      "loss": 3.827,
+      "step": 7995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005749775269093554,
+      "loss": 3.8829,
+      "step": 8000
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 3.9269142150878906,
+      "eval_runtime": 150.3729,
+      "eval_samples_per_second": 12.243,
+      "eval_steps_per_second": 0.771,
+      "step": 8000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000574944624996077,
+      "loss": 3.8017,
+      "step": 8005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000574911702408372,
+      "loss": 3.9319,
+      "step": 8010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005748787591487158,
+      "loss": 3.9119,
+      "step": 8015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005748457952195856,
+      "loss": 3.9751,
+      "step": 8020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005748128106234602,
+      "loss": 3.8432,
+      "step": 8025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005747798053628198,
+      "loss": 3.9297,
+      "step": 8030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005747467794401464,
+      "loss": 3.8792,
+      "step": 8035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005747137328579231,
+      "loss": 3.8893,
+      "step": 8040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000574680665618635,
+      "loss": 3.8906,
+      "step": 8045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005746475777247686,
+      "loss": 3.8385,
+      "step": 8050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005746144691788119,
+      "loss": 3.7674,
+      "step": 8055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005745813399832544,
+      "loss": 3.8042,
+      "step": 8060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005745481901405874,
+      "loss": 3.9017,
+      "step": 8065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005745150196533037,
+      "loss": 3.8173,
+      "step": 8070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005744818285238972,
+      "loss": 3.8445,
+      "step": 8075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000574448616754864,
+      "loss": 3.9037,
+      "step": 8080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005744153843487012,
+      "loss": 3.8863,
+      "step": 8085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000574382131307908,
+      "loss": 3.8032,
+      "step": 8090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005743488576349846,
+      "loss": 3.8524,
+      "step": 8095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005743155633324332,
+      "loss": 3.7371,
+      "step": 8100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005742822484027573,
+      "loss": 3.8358,
+      "step": 8105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005742489128484619,
+      "loss": 4.0695,
+      "step": 8110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005742155566720538,
+      "loss": 3.8984,
+      "step": 8115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005741821798760412,
+      "loss": 3.8755,
+      "step": 8120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005741487824629339,
+      "loss": 3.9444,
+      "step": 8125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000574115364435243,
+      "loss": 3.9359,
+      "step": 8130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005740819257954814,
+      "loss": 3.9182,
+      "step": 8135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005740484665461639,
+      "loss": 3.9721,
+      "step": 8140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005740149866898061,
+      "loss": 3.8413,
+      "step": 8145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005739814862289256,
+      "loss": 3.9575,
+      "step": 8150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005739479651660415,
+      "loss": 3.8612,
+      "step": 8155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005739144235036744,
+      "loss": 3.8002,
+      "step": 8160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005738808612443464,
+      "loss": 3.8984,
+      "step": 8165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005738472783905811,
+      "loss": 3.8158,
+      "step": 8170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005738136749449041,
+      "loss": 3.8328,
+      "step": 8175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000573780050909842,
+      "loss": 3.8829,
+      "step": 8180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005737464062879231,
+      "loss": 3.7717,
+      "step": 8185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005737127410816775,
+      "loss": 3.879,
+      "step": 8190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005736790552936364,
+      "loss": 3.9169,
+      "step": 8195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005736453489263329,
+      "loss": 3.8645,
+      "step": 8200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005736116219823016,
+      "loss": 3.8988,
+      "step": 8205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005735778744640787,
+      "loss": 3.8545,
+      "step": 8210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005735441063742016,
+      "loss": 3.7853,
+      "step": 8215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005735103177152096,
+      "loss": 3.8307,
+      "step": 8220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005734765084896435,
+      "loss": 3.9213,
+      "step": 8225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005734426787000453,
+      "loss": 3.9432,
+      "step": 8230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005734088283489593,
+      "loss": 3.868,
+      "step": 8235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005733749574389304,
+      "loss": 3.827,
+      "step": 8240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005733410659725058,
+      "loss": 3.9695,
+      "step": 8245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005733071539522339,
+      "loss": 3.8528,
+      "step": 8250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005732732213806647,
+      "loss": 3.8285,
+      "step": 8255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005732392682603497,
+      "loss": 3.9186,
+      "step": 8260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005732052945938421,
+      "loss": 3.8465,
+      "step": 8265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005731713003836965,
+      "loss": 3.8373,
+      "step": 8270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005731372856324691,
+      "loss": 3.9738,
+      "step": 8275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005731032503427176,
+      "loss": 3.8417,
+      "step": 8280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005730691945170013,
+      "loss": 3.9056,
+      "step": 8285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005730351181578811,
+      "loss": 3.9564,
+      "step": 8290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005730010212679192,
+      "loss": 3.8577,
+      "step": 8295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005729669038496796,
+      "loss": 3.8578,
+      "step": 8300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005729327659057279,
+      "loss": 3.9018,
+      "step": 8305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005728986074386309,
+      "loss": 3.9108,
+      "step": 8310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005728644284509572,
+      "loss": 3.8183,
+      "step": 8315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005728302289452769,
+      "loss": 3.8654,
+      "step": 8320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005727960089241615,
+      "loss": 3.8433,
+      "step": 8325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005727617683901845,
+      "loss": 3.8515,
+      "step": 8330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005727275073459202,
+      "loss": 3.8286,
+      "step": 8335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005726932257939452,
+      "loss": 3.8778,
+      "step": 8340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000572658923736837,
+      "loss": 3.7871,
+      "step": 8345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005726246011771752,
+      "loss": 3.9097,
+      "step": 8350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005725902581175406,
+      "loss": 3.837,
+      "step": 8355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005725558945605155,
+      "loss": 3.7465,
+      "step": 8360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005725215105086839,
+      "loss": 3.9166,
+      "step": 8365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005724871059646316,
+      "loss": 3.9286,
+      "step": 8370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005724526809309452,
+      "loss": 3.785,
+      "step": 8375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005724182354102136,
+      "loss": 3.8503,
+      "step": 8380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005723837694050267,
+      "loss": 3.8255,
+      "step": 8385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005723492829179764,
+      "loss": 3.887,
+      "step": 8390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005723147759516558,
+      "loss": 3.7949,
+      "step": 8395
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005722802485086595,
+      "loss": 3.8919,
+      "step": 8400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005722457005915841,
+      "loss": 3.8261,
+      "step": 8405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000572211132203027,
+      "loss": 3.8836,
+      "step": 8410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000572176543345588,
+      "loss": 3.7968,
+      "step": 8415
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005721419340218678,
+      "loss": 3.8791,
+      "step": 8420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000572107304234469,
+      "loss": 3.8896,
+      "step": 8425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005720726539859953,
+      "loss": 3.8461,
+      "step": 8430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005720379832790524,
+      "loss": 3.8805,
+      "step": 8435
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005720032921162473,
+      "loss": 3.7594,
+      "step": 8440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005719685805001886,
+      "loss": 3.906,
+      "step": 8445
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005719338484334866,
+      "loss": 3.8963,
+      "step": 8450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005718990959187527,
+      "loss": 3.8806,
+      "step": 8455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005718643229586002,
+      "loss": 3.9259,
+      "step": 8460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005718295295556441,
+      "loss": 3.8151,
+      "step": 8465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005717947157125004,
+      "loss": 3.8314,
+      "step": 8470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000571759881431787,
+      "loss": 3.8986,
+      "step": 8475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005717250267161232,
+      "loss": 3.8939,
+      "step": 8480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00057169015156813,
+      "loss": 3.9307,
+      "step": 8485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005716552559904297,
+      "loss": 3.7868,
+      "step": 8490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005716203399856466,
+      "loss": 3.8096,
+      "step": 8495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005715854035564058,
+      "loss": 3.8729,
+      "step": 8500
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 3.8896238803863525,
+      "eval_runtime": 150.0738,
+      "eval_samples_per_second": 12.267,
+      "eval_steps_per_second": 0.773,
+      "step": 8500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005715504467053345,
+      "loss": 3.7995,
+      "step": 8505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005715154694350613,
+      "loss": 3.8546,
+      "step": 8510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005714804717482163,
+      "loss": 3.8225,
+      "step": 8515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005714454536474311,
+      "loss": 3.866,
+      "step": 8520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000571410415135339,
+      "loss": 3.8313,
+      "step": 8525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005713753562145746,
+      "loss": 3.8833,
+      "step": 8530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005713402768877743,
+      "loss": 3.8299,
+      "step": 8535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005713051771575756,
+      "loss": 3.9089,
+      "step": 8540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000571270057026618,
+      "loss": 3.8661,
+      "step": 8545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005712349164975424,
+      "loss": 3.7924,
+      "step": 8550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005711997555729912,
+      "loss": 3.9294,
+      "step": 8555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005711645742556082,
+      "loss": 3.7973,
+      "step": 8560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005711293725480389,
+      "loss": 3.7566,
+      "step": 8565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005710941504529304,
+      "loss": 3.9148,
+      "step": 8570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005710589079729309,
+      "loss": 3.8453,
+      "step": 8575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005710236451106909,
+      "loss": 3.8899,
+      "step": 8580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005709883618688617,
+      "loss": 3.8985,
+      "step": 8585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005709530582500964,
+      "loss": 3.8628,
+      "step": 8590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005709177342570498,
+      "loss": 3.7169,
+      "step": 8595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005708823898923779,
+      "loss": 3.8679,
+      "step": 8600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005708470251587387,
+      "loss": 3.9674,
+      "step": 8605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005708116400587912,
+      "loss": 3.9565,
+      "step": 8610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005707762345951963,
+      "loss": 3.786,
+      "step": 8615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005707408087706163,
+      "loss": 3.8328,
+      "step": 8620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005707053625877149,
+      "loss": 3.8651,
+      "step": 8625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005706698960491576,
+      "loss": 3.8354,
+      "step": 8630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005706344091576113,
+      "loss": 3.9148,
+      "step": 8635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005705989019157443,
+      "loss": 3.743,
+      "step": 8640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005705633743262269,
+      "loss": 3.9586,
+      "step": 8645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005705278263917301,
+      "loss": 3.9242,
+      "step": 8650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005704922581149275,
+      "loss": 3.7486,
+      "step": 8655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005704566694984931,
+      "loss": 3.8465,
+      "step": 8660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005704210605451033,
+      "loss": 3.9023,
+      "step": 8665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005703854312574357,
+      "loss": 3.9225,
+      "step": 8670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005703497816381692,
+      "loss": 3.9354,
+      "step": 8675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005703141116899847,
+      "loss": 3.8554,
+      "step": 8680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005702784214155645,
+      "loss": 3.8595,
+      "step": 8685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000570242710817592,
+      "loss": 3.8911,
+      "step": 8690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005702069798987526,
+      "loss": 3.9046,
+      "step": 8695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005701712286617332,
+      "loss": 3.9157,
+      "step": 8700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005701354571092219,
+      "loss": 3.7783,
+      "step": 8705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005700996652439088,
+      "loss": 3.8195,
+      "step": 8710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000570063853068485,
+      "loss": 3.8212,
+      "step": 8715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005700280205856435,
+      "loss": 3.8575,
+      "step": 8720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005699921677980788,
+      "loss": 3.8246,
+      "step": 8725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005699562947084867,
+      "loss": 3.7655,
+      "step": 8730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005699204013195648,
+      "loss": 3.6899,
+      "step": 8735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000569884487634012,
+      "loss": 3.8449,
+      "step": 8740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005698485536545289,
+      "loss": 3.8651,
+      "step": 8745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005698125993838176,
+      "loss": 3.8758,
+      "step": 8750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005697766248245813,
+      "loss": 3.7971,
+      "step": 8755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005697406299795257,
+      "loss": 3.8458,
+      "step": 8760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000569704614851357,
+      "loss": 3.8787,
+      "step": 8765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005696685794427835,
+      "loss": 3.7689,
+      "step": 8770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005696325237565148,
+      "loss": 3.8765,
+      "step": 8775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005695964477952623,
+      "loss": 3.8841,
+      "step": 8780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005695603515617384,
+      "loss": 3.8682,
+      "step": 8785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005695242350586576,
+      "loss": 3.9042,
+      "step": 8790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005694880982887356,
+      "loss": 3.8366,
+      "step": 8795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005694519412546899,
+      "loss": 3.72,
+      "step": 8800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005694157639592388,
+      "loss": 3.8754,
+      "step": 8805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005693795664051032,
+      "loss": 3.7116,
+      "step": 8810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005693433485950048,
+      "loss": 3.8559,
+      "step": 8815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005693071105316668,
+      "loss": 3.8147,
+      "step": 8820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005692708522178144,
+      "loss": 3.8115,
+      "step": 8825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005692345736561738,
+      "loss": 3.8211,
+      "step": 8830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005691982748494732,
+      "loss": 3.7904,
+      "step": 8835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005691619558004418,
+      "loss": 3.725,
+      "step": 8840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005691256165118109,
+      "loss": 3.7322,
+      "step": 8845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005690892569863127,
+      "loss": 3.8021,
+      "step": 8850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005690528772266816,
+      "loss": 3.8315,
+      "step": 8855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005690164772356529,
+      "loss": 3.8825,
+      "step": 8860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005689800570159639,
+      "loss": 3.9128,
+      "step": 8865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005689436165703531,
+      "loss": 3.7504,
+      "step": 8870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005689071559015605,
+      "loss": 3.8452,
+      "step": 8875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000568870675012328,
+      "loss": 3.8525,
+      "step": 8880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005688341739053988,
+      "loss": 3.8366,
+      "step": 8885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005687976525835173,
+      "loss": 3.9315,
+      "step": 8890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005687611110494299,
+      "loss": 3.863,
+      "step": 8895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005687245493058845,
+      "loss": 3.9057,
+      "step": 8900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00056868796735563,
+      "loss": 3.8194,
+      "step": 8905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005686513652014175,
+      "loss": 3.8638,
+      "step": 8910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005686147428459991,
+      "loss": 3.8555,
+      "step": 8915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005685781002921286,
+      "loss": 3.7245,
+      "step": 8920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005685414375425615,
+      "loss": 3.8012,
+      "step": 8925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005685047546000544,
+      "loss": 3.8829,
+      "step": 8930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000568468051467366,
+      "loss": 3.7756,
+      "step": 8935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005684313281472559,
+      "loss": 3.8913,
+      "step": 8940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005683945846424856,
+      "loss": 3.7455,
+      "step": 8945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005683578209558181,
+      "loss": 3.7563,
+      "step": 8950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005683210370900177,
+      "loss": 3.8759,
+      "step": 8955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005682842330478505,
+      "loss": 3.8912,
+      "step": 8960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005682474088320839,
+      "loss": 3.9026,
+      "step": 8965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005682105644454869,
+      "loss": 3.8464,
+      "step": 8970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00056817369989083,
+      "loss": 3.8083,
+      "step": 8975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005681368151708852,
+      "loss": 3.8982,
+      "step": 8980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005680999102884261,
+      "loss": 3.8016,
+      "step": 8985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005680629852462278,
+      "loss": 3.9751,
+      "step": 8990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005680260400470668,
+      "loss": 3.8363,
+      "step": 8995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005679890746937211,
+      "loss": 3.8684,
+      "step": 9000
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 3.870915412902832,
+      "eval_runtime": 150.275,
+      "eval_samples_per_second": 12.251,
+      "eval_steps_per_second": 0.772,
+      "step": 9000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005679520891889705,
+      "loss": 3.9171,
+      "step": 9005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000567915083535596,
+      "loss": 3.9465,
+      "step": 9010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005678780577363802,
+      "loss": 3.9283,
+      "step": 9015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005678410117941075,
+      "loss": 3.8472,
+      "step": 9020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005678039457115632,
+      "loss": 3.9198,
+      "step": 9025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005677668594915347,
+      "loss": 3.8925,
+      "step": 9030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005677297531368106,
+      "loss": 3.893,
+      "step": 9035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005676926266501812,
+      "loss": 3.8164,
+      "step": 9040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005676554800344382,
+      "loss": 3.8225,
+      "step": 9045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005676183132923748,
+      "loss": 3.8328,
+      "step": 9050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005675811264267858,
+      "loss": 3.8067,
+      "step": 9055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005675439194404673,
+      "loss": 3.8443,
+      "step": 9060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005675066923362174,
+      "loss": 3.7139,
+      "step": 9065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000567469445116835,
+      "loss": 3.733,
+      "step": 9070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005674321777851211,
+      "loss": 3.7259,
+      "step": 9075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005673948903438781,
+      "loss": 3.8794,
+      "step": 9080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005673575827959097,
+      "loss": 3.7507,
+      "step": 9085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005673202551440212,
+      "loss": 3.8077,
+      "step": 9090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005672829073910196,
+      "loss": 3.778,
+      "step": 9095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005672455395397132,
+      "loss": 3.8871,
+      "step": 9100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005672081515929118,
+      "loss": 3.8326,
+      "step": 9105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005671707435534267,
+      "loss": 3.8089,
+      "step": 9110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005671333154240709,
+      "loss": 3.8371,
+      "step": 9115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000567095867207659,
+      "loss": 3.7742,
+      "step": 9120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005670583989070066,
+      "loss": 3.8002,
+      "step": 9125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005670209105249313,
+      "loss": 3.7759,
+      "step": 9130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000566983402064252,
+      "loss": 3.8101,
+      "step": 9135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005669458735277891,
+      "loss": 3.8326,
+      "step": 9140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005669083249183646,
+      "loss": 3.9168,
+      "step": 9145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000566870756238802,
+      "loss": 3.8423,
+      "step": 9150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005668331674919261,
+      "loss": 3.7503,
+      "step": 9155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005667955586805635,
+      "loss": 3.9171,
+      "step": 9160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005667579298075424,
+      "loss": 3.844,
+      "step": 9165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005667202808756919,
+      "loss": 3.9111,
+      "step": 9170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005666826118878432,
+      "loss": 3.7965,
+      "step": 9175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005666449228468288,
+      "loss": 3.8016,
+      "step": 9180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005666072137554829,
+      "loss": 3.7283,
+      "step": 9185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005665694846166408,
+      "loss": 3.7984,
+      "step": 9190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005665317354331395,
+      "loss": 3.7922,
+      "step": 9195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005664939662078178,
+      "loss": 3.7183,
+      "step": 9200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005664561769435155,
+      "loss": 3.9538,
+      "step": 9205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005664183676430743,
+      "loss": 3.7491,
+      "step": 9210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005663805383093372,
+      "loss": 3.8107,
+      "step": 9215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005663426889451488,
+      "loss": 3.8527,
+      "step": 9220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005663048195533552,
+      "loss": 3.8206,
+      "step": 9225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000566266930136804,
+      "loss": 3.8189,
+      "step": 9230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005662290206983441,
+      "loss": 3.7953,
+      "step": 9235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005661910912408264,
+      "loss": 3.8083,
+      "step": 9240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005661531417671027,
+      "loss": 3.7971,
+      "step": 9245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005661151722800269,
+      "loss": 3.6886,
+      "step": 9250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005660771827824539,
+      "loss": 3.7777,
+      "step": 9255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005660391732772404,
+      "loss": 3.6483,
+      "step": 9260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005660011437672445,
+      "loss": 3.8698,
+      "step": 9265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005659630942553258,
+      "loss": 3.8082,
+      "step": 9270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005659250247443454,
+      "loss": 3.8567,
+      "step": 9275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005658869352371661,
+      "loss": 3.8349,
+      "step": 9280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005658488257366518,
+      "loss": 3.8353,
+      "step": 9285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005658106962456684,
+      "loss": 3.7528,
+      "step": 9290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005657725467670828,
+      "loss": 3.7924,
+      "step": 9295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005657343773037637,
+      "loss": 3.8372,
+      "step": 9300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005656961878585813,
+      "loss": 3.7955,
+      "step": 9305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005656579784344074,
+      "loss": 3.7235,
+      "step": 9310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005656197490341149,
+      "loss": 3.8506,
+      "step": 9315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005655814996605785,
+      "loss": 3.844,
+      "step": 9320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005655432303166744,
+      "loss": 3.7618,
+      "step": 9325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005655049410052804,
+      "loss": 3.7935,
+      "step": 9330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005654666317292754,
+      "loss": 3.9014,
+      "step": 9335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005654283024915403,
+      "loss": 3.829,
+      "step": 9340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005653899532949571,
+      "loss": 3.7477,
+      "step": 9345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005653515841424095,
+      "loss": 3.9187,
+      "step": 9350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005653131950367826,
+      "loss": 3.8836,
+      "step": 9355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005652747859809632,
+      "loss": 3.8617,
+      "step": 9360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005652363569778395,
+      "loss": 3.8054,
+      "step": 9365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005651979080303009,
+      "loss": 3.8272,
+      "step": 9370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000565159439141239,
+      "loss": 3.8475,
+      "step": 9375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000565120950313546,
+      "loss": 3.823,
+      "step": 9380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005650824415501163,
+      "loss": 3.9115,
+      "step": 9385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005650439128538455,
+      "loss": 3.837,
+      "step": 9390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005650053642276307,
+      "loss": 3.7632,
+      "step": 9395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005649667956743708,
+      "loss": 3.7894,
+      "step": 9400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005649282071969658,
+      "loss": 3.8228,
+      "step": 9405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005648895987983172,
+      "loss": 3.8639,
+      "step": 9410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005648509704813284,
+      "loss": 3.8556,
+      "step": 9415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000564812322248904,
+      "loss": 3.856,
+      "step": 9420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00056477365410395,
+      "loss": 3.8072,
+      "step": 9425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005647349660493743,
+      "loss": 3.8082,
+      "step": 9430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005646962580880858,
+      "loss": 3.7869,
+      "step": 9435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005646575302229953,
+      "loss": 3.8706,
+      "step": 9440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005646187824570148,
+      "loss": 3.7245,
+      "step": 9445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000564580014793058,
+      "loss": 3.7688,
+      "step": 9450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005645412272340401,
+      "loss": 3.7546,
+      "step": 9455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005645024197828777,
+      "loss": 3.7613,
+      "step": 9460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005644635924424888,
+      "loss": 3.7864,
+      "step": 9465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005644247452157932,
+      "loss": 3.8333,
+      "step": 9470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005643858781057118,
+      "loss": 3.8384,
+      "step": 9475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005643469911151673,
+      "loss": 3.8734,
+      "step": 9480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005643080842470839,
+      "loss": 3.8705,
+      "step": 9485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005642691575043872,
+      "loss": 3.7985,
+      "step": 9490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005642302108900041,
+      "loss": 3.8142,
+      "step": 9495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005641912444068634,
+      "loss": 3.8861,
+      "step": 9500
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 3.8443562984466553,
+      "eval_runtime": 150.875,
+      "eval_samples_per_second": 12.202,
+      "eval_steps_per_second": 0.769,
+      "step": 9500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005641522580578949,
+      "loss": 3.8243,
+      "step": 9505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005641132518460305,
+      "loss": 3.817,
+      "step": 9510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005640742257742031,
+      "loss": 3.7974,
+      "step": 9515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005640351798453472,
+      "loss": 3.8722,
+      "step": 9520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000563996114062399,
+      "loss": 3.8327,
+      "step": 9525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000563957028428296,
+      "loss": 3.7176,
+      "step": 9530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005639179229459772,
+      "loss": 3.8109,
+      "step": 9535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005638787976183831,
+      "loss": 3.7644,
+      "step": 9540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005638396524484558,
+      "loss": 3.8023,
+      "step": 9545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005638004874391389,
+      "loss": 3.7181,
+      "step": 9550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005637613025933772,
+      "loss": 3.8218,
+      "step": 9555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005637220979141174,
+      "loss": 3.821,
+      "step": 9560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005636828734043072,
+      "loss": 3.7989,
+      "step": 9565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005636436290668965,
+      "loss": 3.8229,
+      "step": 9570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005636043649048359,
+      "loss": 3.7486,
+      "step": 9575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005635650809210781,
+      "loss": 3.7537,
+      "step": 9580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005635257771185768,
+      "loss": 3.8646,
+      "step": 9585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005634864535002878,
+      "loss": 3.8951,
+      "step": 9590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005634471100691678,
+      "loss": 3.8565,
+      "step": 9595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005634077468281752,
+      "loss": 3.865,
+      "step": 9600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00056336836378027,
+      "loss": 3.7137,
+      "step": 9605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005633289609284135,
+      "loss": 3.875,
+      "step": 9610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005632895382755689,
+      "loss": 3.7517,
+      "step": 9615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005632500958247002,
+      "loss": 3.8388,
+      "step": 9620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005632106335787734,
+      "loss": 3.7177,
+      "step": 9625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005631711515407561,
+      "loss": 3.9332,
+      "step": 9630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005631316497136167,
+      "loss": 3.7879,
+      "step": 9635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005630921281003259,
+      "loss": 3.8254,
+      "step": 9640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005630525867038553,
+      "loss": 3.7538,
+      "step": 9645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005630130255271784,
+      "loss": 3.8595,
+      "step": 9650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005629734445732698,
+      "loss": 3.8395,
+      "step": 9655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005629338438451059,
+      "loss": 3.8212,
+      "step": 9660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005628942233456645,
+      "loss": 3.6969,
+      "step": 9665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005628545830779248,
+      "loss": 3.8934,
+      "step": 9670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005628149230448677,
+      "loss": 3.784,
+      "step": 9675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005627752432494751,
+      "loss": 3.7458,
+      "step": 9680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005627355436947309,
+      "loss": 3.7764,
+      "step": 9685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005626958243836205,
+      "loss": 3.8557,
+      "step": 9690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005626560853191302,
+      "loss": 3.7871,
+      "step": 9695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005626163265042485,
+      "loss": 3.8723,
+      "step": 9700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005625765479419649,
+      "loss": 3.7628,
+      "step": 9705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005625367496352706,
+      "loss": 3.7876,
+      "step": 9710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005624969315871581,
+      "loss": 3.8288,
+      "step": 9715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005624570938006217,
+      "loss": 3.7054,
+      "step": 9720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000562417236278657,
+      "loss": 3.8978,
+      "step": 9725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005623773590242608,
+      "loss": 3.7314,
+      "step": 9730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000562337462040432,
+      "loss": 3.844,
+      "step": 9735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005622975453301706,
+      "loss": 3.8081,
+      "step": 9740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005622576088964779,
+      "loss": 3.7213,
+      "step": 9745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005622176527423571,
+      "loss": 3.7476,
+      "step": 9750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005621776768708126,
+      "loss": 3.8561,
+      "step": 9755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005621376812848505,
+      "loss": 3.8914,
+      "step": 9760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005620976659874782,
+      "loss": 3.706,
+      "step": 9765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005620576309817046,
+      "loss": 3.7814,
+      "step": 9770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005620175762705402,
+      "loss": 3.7574,
+      "step": 9775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000561977501856997,
+      "loss": 3.8686,
+      "step": 9780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005619374077440882,
+      "loss": 3.8096,
+      "step": 9785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005618972939348286,
+      "loss": 3.8072,
+      "step": 9790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005618571604322349,
+      "loss": 3.8532,
+      "step": 9795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005618170072393246,
+      "loss": 3.7328,
+      "step": 9800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005617768343591172,
+      "loss": 3.8532,
+      "step": 9805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005617366417946335,
+      "loss": 3.8652,
+      "step": 9810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005616964295488957,
+      "loss": 3.7735,
+      "step": 9815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005616561976249275,
+      "loss": 3.7548,
+      "step": 9820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005616159460257542,
+      "loss": 3.764,
+      "step": 9825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005615756747544026,
+      "loss": 3.7759,
+      "step": 9830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005615353838139006,
+      "loss": 3.7218,
+      "step": 9835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005614950732072784,
+      "loss": 3.8552,
+      "step": 9840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005614547429375667,
+      "loss": 3.7065,
+      "step": 9845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005614143930077982,
+      "loss": 3.8019,
+      "step": 9850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005613740234210072,
+      "loss": 3.7688,
+      "step": 9855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005613336341802291,
+      "loss": 3.7568,
+      "step": 9860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000561293225288501,
+      "loss": 3.7632,
+      "step": 9865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005612527967488614,
+      "loss": 3.6562,
+      "step": 9870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005612123485643505,
+      "loss": 3.8351,
+      "step": 9875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005611718807380096,
+      "loss": 3.8006,
+      "step": 9880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005611313932728818,
+      "loss": 3.6692,
+      "step": 9885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005610908861720114,
+      "loss": 3.7825,
+      "step": 9890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005610503594384446,
+      "loss": 3.7926,
+      "step": 9895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005610098130752286,
+      "loss": 3.8703,
+      "step": 9900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005609692470854122,
+      "loss": 3.7963,
+      "step": 9905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005609286614720459,
+      "loss": 3.8204,
+      "step": 9910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005608880562381815,
+      "loss": 3.8013,
+      "step": 9915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005608474313868724,
+      "loss": 3.7682,
+      "step": 9920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005608067869211733,
+      "loss": 3.7958,
+      "step": 9925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005607661228441402,
+      "loss": 3.7584,
+      "step": 9930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005607254391588312,
+      "loss": 3.8284,
+      "step": 9935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005606847358683054,
+      "loss": 3.7367,
+      "step": 9940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005606440129756234,
+      "loss": 3.7325,
+      "step": 9945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005606032704838475,
+      "loss": 3.6778,
+      "step": 9950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005605625083960412,
+      "loss": 3.7351,
+      "step": 9955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005605217267152696,
+      "loss": 3.812,
+      "step": 9960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005604809254445992,
+      "loss": 3.7126,
+      "step": 9965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005604401045870983,
+      "loss": 3.8023,
+      "step": 9970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005603992641458362,
+      "loss": 3.7676,
+      "step": 9975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000560358404123884,
+      "loss": 3.8314,
+      "step": 9980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005603175245243141,
+      "loss": 3.7735,
+      "step": 9985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005602766253502005,
+      "loss": 3.752,
+      "step": 9990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005602357066046185,
+      "loss": 3.8366,
+      "step": 9995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005601947682906452,
+      "loss": 3.7659,
+      "step": 10000
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 3.824028253555298,
+      "eval_runtime": 150.1823,
+      "eval_samples_per_second": 12.258,
+      "eval_steps_per_second": 0.772,
+      "step": 10000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005601538104113585,
+      "loss": 3.7734,
+      "step": 10005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005601128329698388,
+      "loss": 3.7532,
+      "step": 10010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000560071835969167,
+      "loss": 3.828,
+      "step": 10015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000560030819412426,
+      "loss": 3.8327,
+      "step": 10020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005599897833027,
+      "loss": 3.9457,
+      "step": 10025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005599487276430749,
+      "loss": 3.7388,
+      "step": 10030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005599076524366375,
+      "loss": 3.8464,
+      "step": 10035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005598665576864768,
+      "loss": 3.7601,
+      "step": 10040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005598254433956826,
+      "loss": 3.7211,
+      "step": 10045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005597843095673466,
+      "loss": 3.7223,
+      "step": 10050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000559743156204562,
+      "loss": 3.6862,
+      "step": 10055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005597019833104232,
+      "loss": 3.8612,
+      "step": 10060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005596607908880263,
+      "loss": 3.7786,
+      "step": 10065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005596195789404685,
+      "loss": 3.9116,
+      "step": 10070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000559578347470849,
+      "loss": 3.8694,
+      "step": 10075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005595370964822681,
+      "loss": 3.7191,
+      "step": 10080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005594958259778276,
+      "loss": 3.8221,
+      "step": 10085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005594545359606308,
+      "loss": 3.8228,
+      "step": 10090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005594132264337827,
+      "loss": 3.8175,
+      "step": 10095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005593718974003895,
+      "loss": 3.7765,
+      "step": 10100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005593305488635587,
+      "loss": 3.7567,
+      "step": 10105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005592891808263998,
+      "loss": 3.8263,
+      "step": 10110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005592477932920232,
+      "loss": 3.664,
+      "step": 10115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005592063862635414,
+      "loss": 3.873,
+      "step": 10120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005591649597440675,
+      "loss": 3.7406,
+      "step": 10125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000559123513736717,
+      "loss": 3.7542,
+      "step": 10130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005590820482446063,
+      "loss": 3.8525,
+      "step": 10135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005590405632708533,
+      "loss": 3.7978,
+      "step": 10140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005589990588185773,
+      "loss": 3.7052,
+      "step": 10145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005589575348908996,
+      "loss": 3.8114,
+      "step": 10150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005589159914909425,
+      "loss": 3.7365,
+      "step": 10155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005588744286218297,
+      "loss": 3.7698,
+      "step": 10160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005588328462866864,
+      "loss": 3.7538,
+      "step": 10165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005587912444886397,
+      "loss": 3.7885,
+      "step": 10170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005587496232308177,
+      "loss": 3.761,
+      "step": 10175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005587079825163502,
+      "loss": 3.7494,
+      "step": 10180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005586663223483681,
+      "loss": 3.7903,
+      "step": 10185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005586246427300043,
+      "loss": 3.7783,
+      "step": 10190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005585829436643928,
+      "loss": 3.7313,
+      "step": 10195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005585412251546691,
+      "loss": 3.8423,
+      "step": 10200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005584994872039704,
+      "loss": 3.8002,
+      "step": 10205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005584577298154349,
+      "loss": 3.7193,
+      "step": 10210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005584159529922028,
+      "loss": 3.7501,
+      "step": 10215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005583741567374153,
+      "loss": 3.8655,
+      "step": 10220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005583323410542154,
+      "loss": 3.7713,
+      "step": 10225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005582905059457473,
+      "loss": 3.7608,
+      "step": 10230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000558248651415157,
+      "loss": 3.7848,
+      "step": 10235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005582067774655916,
+      "loss": 3.8751,
+      "step": 10240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005581648841001997,
+      "loss": 3.8157,
+      "step": 10245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005581229713221318,
+      "loss": 3.7949,
+      "step": 10250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005580810391345391,
+      "loss": 3.7647,
+      "step": 10255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000558039087540575,
+      "loss": 3.7398,
+      "step": 10260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005579971165433939,
+      "loss": 3.7206,
+      "step": 10265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000557955126146152,
+      "loss": 3.9106,
+      "step": 10270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005579131163520064,
+      "loss": 3.7303,
+      "step": 10275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005578710871641165,
+      "loss": 3.6985,
+      "step": 10280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005578290385856421,
+      "loss": 3.7783,
+      "step": 10285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005577869706197456,
+      "loss": 3.7213,
+      "step": 10290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00055774488326959,
+      "loss": 3.8223,
+      "step": 10295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00055770277653834,
+      "loss": 3.8479,
+      "step": 10300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000557660650429162,
+      "loss": 3.7619,
+      "step": 10305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005576185049452235,
+      "loss": 3.7465,
+      "step": 10310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005575763400896938,
+      "loss": 3.8019,
+      "step": 10315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005575341558657433,
+      "loss": 3.7722,
+      "step": 10320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005574919522765441,
+      "loss": 3.7072,
+      "step": 10325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005574497293252698,
+      "loss": 3.8018,
+      "step": 10330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005574074870150952,
+      "loss": 3.7442,
+      "step": 10335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005573652253491968,
+      "loss": 3.7484,
+      "step": 10340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005573229443307524,
+      "loss": 3.7656,
+      "step": 10345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005572806439629414,
+      "loss": 3.7769,
+      "step": 10350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005572383242489445,
+      "loss": 3.7266,
+      "step": 10355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005571959851919439,
+      "loss": 3.8326,
+      "step": 10360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005571536267951234,
+      "loss": 3.7286,
+      "step": 10365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000557111249061668,
+      "loss": 3.7353,
+      "step": 10370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005570688519947643,
+      "loss": 3.7861,
+      "step": 10375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005570264355976005,
+      "loss": 3.746,
+      "step": 10380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005569839998733659,
+      "loss": 3.8178,
+      "step": 10385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005569415448252516,
+      "loss": 3.7118,
+      "step": 10390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005568990704564498,
+      "loss": 3.8499,
+      "step": 10395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005568565767701546,
+      "loss": 3.7467,
+      "step": 10400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005568140637695611,
+      "loss": 3.7022,
+      "step": 10405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005567715314578664,
+      "loss": 3.7888,
+      "step": 10410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005567289798382682,
+      "loss": 3.661,
+      "step": 10415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005566864089139665,
+      "loss": 3.7638,
+      "step": 10420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005566438186881624,
+      "loss": 3.7379,
+      "step": 10425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005566012091640585,
+      "loss": 3.7724,
+      "step": 10430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005565585803448585,
+      "loss": 3.8414,
+      "step": 10435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005565159322337682,
+      "loss": 3.7841,
+      "step": 10440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005564732648339945,
+      "loss": 3.7352,
+      "step": 10445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005564305781487456,
+      "loss": 3.796,
+      "step": 10450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005563878721812314,
+      "loss": 3.681,
+      "step": 10455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005563451469346632,
+      "loss": 3.6562,
+      "step": 10460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005563024024122537,
+      "loss": 3.7886,
+      "step": 10465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005562596386172172,
+      "loss": 3.7527,
+      "step": 10470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005562168555527689,
+      "loss": 3.7763,
+      "step": 10475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005561740532221265,
+      "loss": 3.7259,
+      "step": 10480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005561312316285079,
+      "loss": 3.7692,
+      "step": 10485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005560883907751334,
+      "loss": 3.753,
+      "step": 10490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005560455306652243,
+      "loss": 3.7146,
+      "step": 10495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005560026513020036,
+      "loss": 3.7457,
+      "step": 10500
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 3.808900833129883,
+      "eval_runtime": 150.2733,
+      "eval_samples_per_second": 12.251,
+      "eval_steps_per_second": 0.772,
+      "step": 10500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005559597526886955,
+      "loss": 3.7659,
+      "step": 10505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005559168348285258,
+      "loss": 3.7797,
+      "step": 10510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005558738977247216,
+      "loss": 3.7639,
+      "step": 10515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005558309413805117,
+      "loss": 3.7582,
+      "step": 10520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005557879657991261,
+      "loss": 3.7758,
+      "step": 10525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005557449709837963,
+      "loss": 3.6406,
+      "step": 10530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005557019569377555,
+      "loss": 3.7329,
+      "step": 10535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005556589236642378,
+      "loss": 3.762,
+      "step": 10540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005556158711664793,
+      "loss": 3.7226,
+      "step": 10545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005555727994477175,
+      "loss": 3.7183,
+      "step": 10550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005555297085111907,
+      "loss": 3.7314,
+      "step": 10555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005554865983601396,
+      "loss": 3.829,
+      "step": 10560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005554434689978056,
+      "loss": 3.7891,
+      "step": 10565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005554003204274318,
+      "loss": 3.765,
+      "step": 10570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005553571526522628,
+      "loss": 3.7701,
+      "step": 10575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005553139656755446,
+      "loss": 3.7329,
+      "step": 10580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005552707595005247,
+      "loss": 3.8377,
+      "step": 10585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000555227534130452,
+      "loss": 3.6951,
+      "step": 10590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005551842895685766,
+      "loss": 3.7439,
+      "step": 10595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005551410258181506,
+      "loss": 3.764,
+      "step": 10600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005550977428824269,
+      "loss": 3.7233,
+      "step": 10605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005550544407646603,
+      "loss": 3.8349,
+      "step": 10610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000555011119468107,
+      "loss": 3.606,
+      "step": 10615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005549677789960242,
+      "loss": 3.7585,
+      "step": 10620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005549244193516713,
+      "loss": 3.6967,
+      "step": 10625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005548810405383084,
+      "loss": 3.7009,
+      "step": 10630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005548376425591976,
+      "loss": 3.7488,
+      "step": 10635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005547942254176021,
+      "loss": 3.8379,
+      "step": 10640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005547507891167866,
+      "loss": 3.6371,
+      "step": 10645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005547073336600174,
+      "loss": 3.7803,
+      "step": 10650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000554663859050562,
+      "loss": 3.7893,
+      "step": 10655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005546203652916895,
+      "loss": 3.7123,
+      "step": 10660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005545768523866706,
+      "loss": 3.7559,
+      "step": 10665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005545333203387768,
+      "loss": 3.7527,
+      "step": 10670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005544897691512821,
+      "loss": 3.747,
+      "step": 10675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005544461988274608,
+      "loss": 3.7641,
+      "step": 10680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0005544026093705894,
+      "loss": 3.8162,
+      "step": 10685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005543590007839455,
+      "loss": 3.8019,
+      "step": 10690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005543153730708084,
+      "loss": 3.8212,
+      "step": 10695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005542717262344586,
+      "loss": 3.6983,
+      "step": 10700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005542280602781782,
+      "loss": 3.7003,
+      "step": 10705
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005541843752052504,
+      "loss": 3.8489,
+      "step": 10710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005541406710189603,
+      "loss": 3.7563,
+      "step": 10715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005540969477225942,
+      "loss": 3.8294,
+      "step": 10720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005540532053194399,
+      "loss": 3.786,
+      "step": 10725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005540094438127865,
+      "loss": 3.7697,
+      "step": 10730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005539656632059247,
+      "loss": 3.7351,
+      "step": 10735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005539218635021466,
+      "loss": 3.8407,
+      "step": 10740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005538780447047457,
+      "loss": 3.7196,
+      "step": 10745
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005538342068170169,
+      "loss": 3.6508,
+      "step": 10750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005537903498422567,
+      "loss": 3.7522,
+      "step": 10755
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005537464737837628,
+      "loss": 3.6545,
+      "step": 10760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005537025786448346,
+      "loss": 3.7622,
+      "step": 10765
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005536586644287726,
+      "loss": 3.7217,
+      "step": 10770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000553614731138879,
+      "loss": 3.6806,
+      "step": 10775
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005535707787784574,
+      "loss": 3.8162,
+      "step": 10780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005535268073508128,
+      "loss": 3.7956,
+      "step": 10785
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005534828168592515,
+      "loss": 3.7212,
+      "step": 10790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005534388073070815,
+      "loss": 3.8734,
+      "step": 10795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005533947786976121,
+      "loss": 3.7601,
+      "step": 10800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005533507310341539,
+      "loss": 3.7531,
+      "step": 10805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005533066643200191,
+      "loss": 3.8357,
+      "step": 10810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005532625785585214,
+      "loss": 3.7786,
+      "step": 10815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005532184737529757,
+      "loss": 3.8007,
+      "step": 10820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005531743499066985,
+      "loss": 3.8569,
+      "step": 10825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005531302070230077,
+      "loss": 3.729,
+      "step": 10830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005530860451052226,
+      "loss": 3.7075,
+      "step": 10835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005530418641566639,
+      "loss": 3.8074,
+      "step": 10840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005529976641806539,
+      "loss": 3.7291,
+      "step": 10845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005529534451805162,
+      "loss": 3.8529,
+      "step": 10850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005529092071595756,
+      "loss": 3.7,
+      "step": 10855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000552864950121159,
+      "loss": 3.65,
+      "step": 10860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005528206740685938,
+      "loss": 3.729,
+      "step": 10865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005527763790052099,
+      "loss": 3.822,
+      "step": 10870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005527320649343375,
+      "loss": 3.7941,
+      "step": 10875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005526877318593092,
+      "loss": 3.7147,
+      "step": 10880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005526433797834584,
+      "loss": 3.6646,
+      "step": 10885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005525990087101202,
+      "loss": 3.827,
+      "step": 10890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005525546186426311,
+      "loss": 3.7701,
+      "step": 10895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000552510209584329,
+      "loss": 3.6971,
+      "step": 10900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005524657815385532,
+      "loss": 3.712,
+      "step": 10905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005524213345086446,
+      "loss": 3.7302,
+      "step": 10910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005523768684979453,
+      "loss": 3.7991,
+      "step": 10915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005523323835097988,
+      "loss": 3.8046,
+      "step": 10920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005522878795475503,
+      "loss": 3.746,
+      "step": 10925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005522433566145462,
+      "loss": 3.6934,
+      "step": 10930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005521988147141343,
+      "loss": 3.7359,
+      "step": 10935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005521542538496642,
+      "loss": 3.6942,
+      "step": 10940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005521096740244864,
+      "loss": 3.6852,
+      "step": 10945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005520650752419531,
+      "loss": 3.7642,
+      "step": 10950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005520204575054182,
+      "loss": 3.7516,
+      "step": 10955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005519758208182362,
+      "loss": 3.8863,
+      "step": 10960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005519311651837639,
+      "loss": 3.8518,
+      "step": 10965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005518864906053592,
+      "loss": 3.6166,
+      "step": 10970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005518417970863812,
+      "loss": 3.6787,
+      "step": 10975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005517970846301908,
+      "loss": 3.6608,
+      "step": 10980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00055175235324015,
+      "loss": 3.7437,
+      "step": 10985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005517076029196225,
+      "loss": 3.7547,
+      "step": 10990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005516628336719732,
+      "loss": 3.8382,
+      "step": 10995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005516180455005686,
+      "loss": 3.9065,
+      "step": 11000
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 3.82816481590271,
+      "eval_runtime": 150.3797,
+      "eval_samples_per_second": 12.242,
+      "eval_steps_per_second": 0.771,
+      "step": 11000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005515732384087765,
+      "loss": 3.8359,
+      "step": 11005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005515284123999662,
+      "loss": 3.7338,
+      "step": 11010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005514835674775083,
+      "loss": 3.6409,
+      "step": 11015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000551438703644775,
+      "loss": 3.6348,
+      "step": 11020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005513938209051396,
+      "loss": 3.8714,
+      "step": 11025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005513489192619775,
+      "loss": 3.7836,
+      "step": 11030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005513039987186647,
+      "loss": 3.6187,
+      "step": 11035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005512590592785791,
+      "loss": 3.7686,
+      "step": 11040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005512141009451,
+      "loss": 3.7887,
+      "step": 11045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005511691237216078,
+      "loss": 3.7462,
+      "step": 11050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005511241276114849,
+      "loss": 3.7589,
+      "step": 11055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005510791126181145,
+      "loss": 3.6582,
+      "step": 11060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005510340787448815,
+      "loss": 3.6191,
+      "step": 11065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005509890259951725,
+      "loss": 3.7391,
+      "step": 11070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000550943954372375,
+      "loss": 3.7873,
+      "step": 11075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005508988638798781,
+      "loss": 3.58,
+      "step": 11080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005508537545210726,
+      "loss": 3.7628,
+      "step": 11085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005508086262993502,
+      "loss": 3.684,
+      "step": 11090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005507634792181046,
+      "loss": 3.808,
+      "step": 11095
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005507183132807304,
+      "loss": 3.6461,
+      "step": 11100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005506731284906242,
+      "loss": 3.7431,
+      "step": 11105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005506279248511832,
+      "loss": 3.6644,
+      "step": 11110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005505827023658067,
+      "loss": 3.8145,
+      "step": 11115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005505374610378953,
+      "loss": 3.7549,
+      "step": 11120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005504922008708507,
+      "loss": 3.6777,
+      "step": 11125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005504469218680763,
+      "loss": 3.6589,
+      "step": 11130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005504016240329772,
+      "loss": 3.7951,
+      "step": 11135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000550356307368959,
+      "loss": 3.6839,
+      "step": 11140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005503109718794296,
+      "loss": 3.8131,
+      "step": 11145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005502656175677979,
+      "loss": 3.7538,
+      "step": 11150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005502202444374744,
+      "loss": 3.5535,
+      "step": 11155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005501748524918709,
+      "loss": 3.6548,
+      "step": 11160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005501294417344006,
+      "loss": 3.7523,
+      "step": 11165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005500840121684781,
+      "loss": 3.778,
+      "step": 11170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005500385637975196,
+      "loss": 3.6966,
+      "step": 11175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005499930966249426,
+      "loss": 3.7497,
+      "step": 11180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005499476106541659,
+      "loss": 3.7591,
+      "step": 11185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005499021058886096,
+      "loss": 3.7704,
+      "step": 11190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005498565823316959,
+      "loss": 3.8197,
+      "step": 11195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005498110399868477,
+      "loss": 3.6668,
+      "step": 11200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005497654788574896,
+      "loss": 3.7984,
+      "step": 11205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005497198989470474,
+      "loss": 3.7465,
+      "step": 11210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005496743002589487,
+      "loss": 3.6981,
+      "step": 11215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000549628682796622,
+      "loss": 3.8666,
+      "step": 11220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005495830465634978,
+      "loss": 3.7241,
+      "step": 11225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005495373915630076,
+      "loss": 3.7521,
+      "step": 11230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005494917177985844,
+      "loss": 3.717,
+      "step": 11235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005494460252736627,
+      "loss": 3.8369,
+      "step": 11240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005494003139916782,
+      "loss": 3.6984,
+      "step": 11245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005493545839560682,
+      "loss": 3.7683,
+      "step": 11250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005493088351702716,
+      "loss": 3.753,
+      "step": 11255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005492630676377281,
+      "loss": 3.6862,
+      "step": 11260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0005492172813618795,
+      "loss": 3.7831,
+      "step": 11265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005491714763461684,
+      "loss": 3.6696,
+      "step": 11270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005491256525940395,
+      "loss": 3.792,
+      "step": 11275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000549079810108938,
+      "loss": 3.7118,
+      "step": 11280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005490339488943115,
+      "loss": 3.7159,
+      "step": 11285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005489880689536083,
+      "loss": 3.8581,
+      "step": 11290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005489421702902783,
+      "loss": 3.7271,
+      "step": 11295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005488962529077728,
+      "loss": 3.7899,
+      "step": 11300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005488503168095449,
+      "loss": 3.7817,
+      "step": 11305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005488043619990483,
+      "loss": 3.7262,
+      "step": 11310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005487583884797389,
+      "loss": 3.7328,
+      "step": 11315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005487123962550735,
+      "loss": 3.8268,
+      "step": 11320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005486663853285106,
+      "loss": 3.6743,
+      "step": 11325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005486203557035098,
+      "loss": 3.7256,
+      "step": 11330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005485743073835325,
+      "loss": 3.6811,
+      "step": 11335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005485282403720411,
+      "loss": 3.6535,
+      "step": 11340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005484821546724998,
+      "loss": 3.7042,
+      "step": 11345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000548436050288374,
+      "loss": 3.7035,
+      "step": 11350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005483899272231303,
+      "loss": 3.6523,
+      "step": 11355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000548343785480237,
+      "loss": 3.6676,
+      "step": 11360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005482976250631639,
+      "loss": 3.7115,
+      "step": 11365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000548251445975382,
+      "loss": 3.7758,
+      "step": 11370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005482052482203634,
+      "loss": 3.7202,
+      "step": 11375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005481590318015823,
+      "loss": 3.7921,
+      "step": 11380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005481127967225137,
+      "loss": 3.5875,
+      "step": 11385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005480665429866344,
+      "loss": 3.872,
+      "step": 11390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005480202705974225,
+      "loss": 3.7499,
+      "step": 11395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005479739795583573,
+      "loss": 3.8315,
+      "step": 11400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005479276698729195,
+      "loss": 3.7723,
+      "step": 11405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005478813415445916,
+      "loss": 3.7126,
+      "step": 11410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005478349945768574,
+      "loss": 3.8067,
+      "step": 11415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005477886289732015,
+      "loss": 3.6816,
+      "step": 11420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005477422447371107,
+      "loss": 3.8123,
+      "step": 11425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005476958418720728,
+      "loss": 3.6074,
+      "step": 11430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000547649420381577,
+      "loss": 3.7801,
+      "step": 11435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005476029802691138,
+      "loss": 3.7547,
+      "step": 11440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005475565215381755,
+      "loss": 3.7165,
+      "step": 11445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005475100441922556,
+      "loss": 3.721,
+      "step": 11450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005474635482348488,
+      "loss": 3.6334,
+      "step": 11455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005474170336694514,
+      "loss": 3.7961,
+      "step": 11460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000547370500499561,
+      "loss": 3.7415,
+      "step": 11465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005473239487286767,
+      "loss": 3.6964,
+      "step": 11470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005472773783602991,
+      "loss": 3.7686,
+      "step": 11475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005472307893979298,
+      "loss": 3.7813,
+      "step": 11480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005471841818450722,
+      "loss": 3.8297,
+      "step": 11485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000547137555705231,
+      "loss": 3.6398,
+      "step": 11490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005470909109819122,
+      "loss": 3.8471,
+      "step": 11495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005470442476786231,
+      "loss": 3.636,
+      "step": 11500
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 3.7674803733825684,
+      "eval_runtime": 150.2732,
+      "eval_samples_per_second": 12.251,
+      "eval_steps_per_second": 0.772,
+      "step": 11500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005469975657988727,
+      "loss": 3.7453,
+      "step": 11505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005469508653461713,
+      "loss": 3.6952,
+      "step": 11510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005469041463240302,
+      "loss": 3.6867,
+      "step": 11515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005468574087359628,
+      "loss": 3.7732,
+      "step": 11520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005468106525854834,
+      "loss": 3.7456,
+      "step": 11525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005467638778761078,
+      "loss": 3.7178,
+      "step": 11530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005467170846113532,
+      "loss": 3.6693,
+      "step": 11535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005466702727947383,
+      "loss": 3.7435,
+      "step": 11540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005466234424297829,
+      "loss": 3.7004,
+      "step": 11545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005465765935200086,
+      "loss": 3.8126,
+      "step": 11550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005465297260689383,
+      "loss": 3.7205,
+      "step": 11555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005464828400800959,
+      "loss": 3.7958,
+      "step": 11560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005464359355570071,
+      "loss": 3.745,
+      "step": 11565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005463890125031991,
+      "loss": 3.681,
+      "step": 11570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005463420709222,
+      "loss": 3.6648,
+      "step": 11575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005462951108175396,
+      "loss": 3.7188,
+      "step": 11580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005462481321927492,
+      "loss": 3.6209,
+      "step": 11585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005462011350513611,
+      "loss": 3.8103,
+      "step": 11590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005461541193969095,
+      "loss": 3.7334,
+      "step": 11595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005461070852329296,
+      "loss": 3.8183,
+      "step": 11600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005460600325629582,
+      "loss": 3.7361,
+      "step": 11605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005460129613905334,
+      "loss": 3.6082,
+      "step": 11610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005459658717191948,
+      "loss": 3.7858,
+      "step": 11615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005459187635524831,
+      "loss": 3.7071,
+      "step": 11620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005458716368939407,
+      "loss": 3.8105,
+      "step": 11625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005458244917471112,
+      "loss": 3.7418,
+      "step": 11630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005457773281155398,
+      "loss": 3.7896,
+      "step": 11635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000545730146002773,
+      "loss": 3.7541,
+      "step": 11640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005456829454123585,
+      "loss": 3.7075,
+      "step": 11645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005456357263478456,
+      "loss": 3.7605,
+      "step": 11650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000545588488812785,
+      "loss": 3.7757,
+      "step": 11655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005455412328107287,
+      "loss": 3.7652,
+      "step": 11660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005454939583452299,
+      "loss": 3.7985,
+      "step": 11665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005454466654198437,
+      "loss": 3.7654,
+      "step": 11670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005453993540381261,
+      "loss": 3.6536,
+      "step": 11675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005453520242036348,
+      "loss": 3.7901,
+      "step": 11680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005453046759199286,
+      "loss": 3.6965,
+      "step": 11685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000545257309190568,
+      "loss": 3.7089,
+      "step": 11690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005452099240191146,
+      "loss": 3.7599,
+      "step": 11695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005451625204091317,
+      "loss": 3.6116,
+      "step": 11700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005451150983641835,
+      "loss": 3.7526,
+      "step": 11705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005450676578878362,
+      "loss": 3.7258,
+      "step": 11710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005450201989836571,
+      "loss": 3.7765,
+      "step": 11715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005449727216552147,
+      "loss": 3.7091,
+      "step": 11720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000544925225906079,
+      "loss": 3.6662,
+      "step": 11725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005448777117398216,
+      "loss": 3.6813,
+      "step": 11730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005448301791600152,
+      "loss": 3.6477,
+      "step": 11735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005447826281702342,
+      "loss": 3.6587,
+      "step": 11740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000544735058774054,
+      "loss": 3.7099,
+      "step": 11745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005446874709750516,
+      "loss": 3.7078,
+      "step": 11750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005446398647768053,
+      "loss": 3.7051,
+      "step": 11755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000544592240182895,
+      "loss": 3.7258,
+      "step": 11760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005445445971969017,
+      "loss": 3.6229,
+      "step": 11765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005444969358224082,
+      "loss": 3.8427,
+      "step": 11770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005444492560629979,
+      "loss": 3.6114,
+      "step": 11775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005444015579222564,
+      "loss": 3.7371,
+      "step": 11780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005443538414037702,
+      "loss": 3.704,
+      "step": 11785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005443061065111275,
+      "loss": 3.7573,
+      "step": 11790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005442583532479177,
+      "loss": 3.7781,
+      "step": 11795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005442105816177315,
+      "loss": 3.7478,
+      "step": 11800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005441627916241611,
+      "loss": 3.6679,
+      "step": 11805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005441149832708,
+      "loss": 3.5633,
+      "step": 11810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005440671565612434,
+      "loss": 3.6967,
+      "step": 11815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005440193114990873,
+      "loss": 3.6898,
+      "step": 11820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005439714480879297,
+      "loss": 3.6566,
+      "step": 11825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005439235663313695,
+      "loss": 3.6392,
+      "step": 11830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005438756662330071,
+      "loss": 3.7343,
+      "step": 11835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005438277477964445,
+      "loss": 3.7082,
+      "step": 11840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005437798110252849,
+      "loss": 3.6861,
+      "step": 11845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005437318559231328,
+      "loss": 3.8254,
+      "step": 11850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005436838824935943,
+      "loss": 3.7302,
+      "step": 11855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005436358907402767,
+      "loss": 3.7656,
+      "step": 11860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005435878806667886,
+      "loss": 3.7166,
+      "step": 11865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005435398522767404,
+      "loss": 3.6894,
+      "step": 11870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005434918055737434,
+      "loss": 3.7122,
+      "step": 11875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005434437405614104,
+      "loss": 3.7014,
+      "step": 11880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005433956572433558,
+      "loss": 3.7079,
+      "step": 11885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005433475556231951,
+      "loss": 3.6731,
+      "step": 11890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005432994357045455,
+      "loss": 3.7536,
+      "step": 11895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005432512974910252,
+      "loss": 3.58,
+      "step": 11900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005432031409862538,
+      "loss": 3.7737,
+      "step": 11905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005431549661938527,
+      "loss": 3.6355,
+      "step": 11910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005431067731174443,
+      "loss": 3.7166,
+      "step": 11915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005430585617606524,
+      "loss": 3.7184,
+      "step": 11920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005430103321271025,
+      "loss": 3.7067,
+      "step": 11925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005429620842204209,
+      "loss": 3.8456,
+      "step": 11930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005429138180442358,
+      "loss": 3.6251,
+      "step": 11935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005428655336021763,
+      "loss": 3.7128,
+      "step": 11940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005428172308978735,
+      "loss": 3.7401,
+      "step": 11945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005427689099349593,
+      "loss": 3.7788,
+      "step": 11950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005427205707170674,
+      "loss": 3.759,
+      "step": 11955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005426722132478323,
+      "loss": 3.78,
+      "step": 11960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005426238375308906,
+      "loss": 3.7712,
+      "step": 11965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005425754435698796,
+      "loss": 3.7476,
+      "step": 11970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005425270313684385,
+      "loss": 3.7033,
+      "step": 11975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005424786009302075,
+      "loss": 3.6609,
+      "step": 11980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005424301522588285,
+      "loss": 3.7616,
+      "step": 11985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005423816853579444,
+      "loss": 3.6577,
+      "step": 11990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005423332002311998,
+      "loss": 3.7224,
+      "step": 11995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005422846968822404,
+      "loss": 3.6866,
+      "step": 12000
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 3.7599782943725586,
+      "eval_runtime": 150.9725,
+      "eval_samples_per_second": 12.194,
+      "eval_steps_per_second": 0.768,
+      "step": 12000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005422361753147136,
+      "loss": 3.7354,
+      "step": 12005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005421876355322677,
+      "loss": 3.7213,
+      "step": 12010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000542139077538553,
+      "loss": 3.7977,
+      "step": 12015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005420905013372205,
+      "loss": 3.786,
+      "step": 12020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000542041906931923,
+      "loss": 3.7917,
+      "step": 12025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005419932943263146,
+      "loss": 3.7851,
+      "step": 12030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005419446635240505,
+      "loss": 3.7055,
+      "step": 12035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005418960145287878,
+      "loss": 3.7075,
+      "step": 12040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005418473473441845,
+      "loss": 3.77,
+      "step": 12045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005417986619739001,
+      "loss": 3.6612,
+      "step": 12050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005417499584215955,
+      "loss": 3.5958,
+      "step": 12055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005417012366909329,
+      "loss": 3.7396,
+      "step": 12060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005416524967855761,
+      "loss": 3.6948,
+      "step": 12065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00054160373870919,
+      "loss": 3.7575,
+      "step": 12070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005415549624654409,
+      "loss": 3.7255,
+      "step": 12075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005415061680579966,
+      "loss": 3.751,
+      "step": 12080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005414573554905262,
+      "loss": 3.7404,
+      "step": 12085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005414085247667002,
+      "loss": 3.8004,
+      "step": 12090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005413596758901901,
+      "loss": 3.649,
+      "step": 12095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005413108088646695,
+      "loss": 3.7088,
+      "step": 12100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005412619236938126,
+      "loss": 3.6278,
+      "step": 12105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005412130203812957,
+      "loss": 3.7007,
+      "step": 12110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005411640989307958,
+      "loss": 3.7458,
+      "step": 12115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005411151593459915,
+      "loss": 3.7283,
+      "step": 12120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005410662016305629,
+      "loss": 3.7408,
+      "step": 12125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005410172257881915,
+      "loss": 3.7432,
+      "step": 12130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005409682318225599,
+      "loss": 3.7822,
+      "step": 12135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005409192197373522,
+      "loss": 3.7222,
+      "step": 12140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005408701895362538,
+      "loss": 3.741,
+      "step": 12145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005408211412229517,
+      "loss": 3.5775,
+      "step": 12150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005407720748011338,
+      "loss": 3.7984,
+      "step": 12155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00054072299027449,
+      "loss": 3.752,
+      "step": 12160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005406738876467109,
+      "loss": 3.706,
+      "step": 12165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000540624766921489,
+      "loss": 3.7555,
+      "step": 12170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005405756281025176,
+      "loss": 3.7497,
+      "step": 12175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005405264711934922,
+      "loss": 3.6543,
+      "step": 12180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005404772961981086,
+      "loss": 3.6224,
+      "step": 12185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005404281031200648,
+      "loss": 3.6435,
+      "step": 12190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005403788919630598,
+      "loss": 3.7747,
+      "step": 12195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005403296627307942,
+      "loss": 3.5874,
+      "step": 12200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005402804154269695,
+      "loss": 3.5728,
+      "step": 12205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005402311500552889,
+      "loss": 3.7743,
+      "step": 12210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005401818666194571,
+      "loss": 3.6319,
+      "step": 12215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005401325651231798,
+      "loss": 3.6402,
+      "step": 12220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005400832455701644,
+      "loss": 3.6527,
+      "step": 12225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005400339079641192,
+      "loss": 3.7487,
+      "step": 12230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005399845523087543,
+      "loss": 3.6764,
+      "step": 12235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000539935178607781,
+      "loss": 3.756,
+      "step": 12240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005398857868649118,
+      "loss": 3.5799,
+      "step": 12245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005398363770838609,
+      "loss": 3.6562,
+      "step": 12250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005397869492683436,
+      "loss": 3.6615,
+      "step": 12255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005397375034220767,
+      "loss": 3.7017,
+      "step": 12260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005396880395487781,
+      "loss": 3.7083,
+      "step": 12265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005396385576521672,
+      "loss": 3.7206,
+      "step": 12270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000539589057735965,
+      "loss": 3.6814,
+      "step": 12275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005395395398038935,
+      "loss": 3.5883,
+      "step": 12280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005394900038596763,
+      "loss": 3.6049,
+      "step": 12285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005394404499070381,
+      "loss": 3.7757,
+      "step": 12290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005393908779497052,
+      "loss": 3.7323,
+      "step": 12295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005393412879914051,
+      "loss": 3.7704,
+      "step": 12300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005392916800358667,
+      "loss": 3.7201,
+      "step": 12305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005392420540868205,
+      "loss": 3.7927,
+      "step": 12310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005391924101479978,
+      "loss": 3.656,
+      "step": 12315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005391427482231317,
+      "loss": 3.7216,
+      "step": 12320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005390930683159566,
+      "loss": 3.7133,
+      "step": 12325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000539043370430208,
+      "loss": 3.6312,
+      "step": 12330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000538993654569623,
+      "loss": 3.643,
+      "step": 12335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00053894392073794,
+      "loss": 3.5749,
+      "step": 12340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005388941689388988,
+      "loss": 3.5516,
+      "step": 12345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005388443991762405,
+      "loss": 3.7479,
+      "step": 12350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005387946114537072,
+      "loss": 3.6775,
+      "step": 12355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005387448057750431,
+      "loss": 3.6478,
+      "step": 12360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005386949821439932,
+      "loss": 3.7257,
+      "step": 12365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005386451405643038,
+      "loss": 3.6918,
+      "step": 12370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000538595281039723,
+      "loss": 3.6817,
+      "step": 12375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005385454035739996,
+      "loss": 3.6628,
+      "step": 12380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005384955081708847,
+      "loss": 3.6527,
+      "step": 12385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005384455948341296,
+      "loss": 3.7236,
+      "step": 12390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005383956635674878,
+      "loss": 3.5499,
+      "step": 12395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005383457143747141,
+      "loss": 3.7324,
+      "step": 12400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005382957472595638,
+      "loss": 3.6974,
+      "step": 12405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005382457622257949,
+      "loss": 3.6924,
+      "step": 12410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005381957592771654,
+      "loss": 3.5955,
+      "step": 12415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005381457384174358,
+      "loss": 3.6657,
+      "step": 12420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000538095699650367,
+      "loss": 3.7176,
+      "step": 12425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005380456429797218,
+      "loss": 3.6794,
+      "step": 12430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005379955684092643,
+      "loss": 3.7586,
+      "step": 12435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005379454759427597,
+      "loss": 3.7462,
+      "step": 12440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005378953655839747,
+      "loss": 3.6993,
+      "step": 12445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005378452373366776,
+      "loss": 3.6243,
+      "step": 12450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005377950912046374,
+      "loss": 3.6997,
+      "step": 12455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005377449271916251,
+      "loss": 3.6549,
+      "step": 12460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005376947453014126,
+      "loss": 3.7064,
+      "step": 12465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005376445455377736,
+      "loss": 3.6477,
+      "step": 12470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005375943279044825,
+      "loss": 3.7082,
+      "step": 12475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005375440924053156,
+      "loss": 3.6758,
+      "step": 12480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005374938390440505,
+      "loss": 3.6828,
+      "step": 12485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005374435678244658,
+      "loss": 3.6934,
+      "step": 12490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005373932787503417,
+      "loss": 3.6921,
+      "step": 12495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005373429718254595,
+      "loss": 3.6586,
+      "step": 12500
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 3.7146527767181396,
+      "eval_runtime": 150.1722,
+      "eval_samples_per_second": 12.259,
+      "eval_steps_per_second": 0.772,
+      "step": 12500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005372926470536023,
+      "loss": 3.759,
+      "step": 12505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005372423044385541,
+      "loss": 3.6515,
+      "step": 12510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005371919439841005,
+      "loss": 3.8031,
+      "step": 12515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005371415656940283,
+      "loss": 3.6402,
+      "step": 12520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005370911695721256,
+      "loss": 3.7482,
+      "step": 12525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005370407556221821,
+      "loss": 3.7288,
+      "step": 12530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005369903238479886,
+      "loss": 3.6749,
+      "step": 12535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005369398742533373,
+      "loss": 3.6862,
+      "step": 12540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005368894068420217,
+      "loss": 3.6132,
+      "step": 12545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005368389216178368,
+      "loss": 3.666,
+      "step": 12550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005367884185845787,
+      "loss": 3.5704,
+      "step": 12555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005367378977460452,
+      "loss": 3.7454,
+      "step": 12560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005366873591060349,
+      "loss": 3.7277,
+      "step": 12565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005366368026683483,
+      "loss": 3.7656,
+      "step": 12570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005365862284367869,
+      "loss": 3.8027,
+      "step": 12575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005365356364151535,
+      "loss": 3.6515,
+      "step": 12580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005364850266072525,
+      "loss": 3.6866,
+      "step": 12585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005364343990168895,
+      "loss": 3.6147,
+      "step": 12590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005363837536478714,
+      "loss": 3.6964,
+      "step": 12595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005363330905040064,
+      "loss": 3.6638,
+      "step": 12600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005362824095891043,
+      "loss": 3.6581,
+      "step": 12605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005362317109069758,
+      "loss": 3.7257,
+      "step": 12610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005361809944614334,
+      "loss": 3.6829,
+      "step": 12615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005361302602562907,
+      "loss": 3.6478,
+      "step": 12620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005360795082953624,
+      "loss": 3.6036,
+      "step": 12625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005360287385824651,
+      "loss": 3.6498,
+      "step": 12630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005359779511214162,
+      "loss": 3.749,
+      "step": 12635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005359271459160347,
+      "loss": 3.6121,
+      "step": 12640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005358763229701411,
+      "loss": 3.6755,
+      "step": 12645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005358254822875567,
+      "loss": 3.6591,
+      "step": 12650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005357746238721047,
+      "loss": 3.6631,
+      "step": 12655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005357237477276093,
+      "loss": 3.6449,
+      "step": 12660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005356728538578961,
+      "loss": 3.6601,
+      "step": 12665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005356219422667922,
+      "loss": 3.6639,
+      "step": 12670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005355710129581257,
+      "loss": 3.693,
+      "step": 12675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005355200659357262,
+      "loss": 3.6938,
+      "step": 12680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005354691012034249,
+      "loss": 3.715,
+      "step": 12685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000535418118765054,
+      "loss": 3.521,
+      "step": 12690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000535367118624447,
+      "loss": 3.692,
+      "step": 12695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005353161007854389,
+      "loss": 3.6524,
+      "step": 12700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005352650652518659,
+      "loss": 3.6092,
+      "step": 12705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005352140120275659,
+      "loss": 3.712,
+      "step": 12710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005351629411163776,
+      "loss": 3.7175,
+      "step": 12715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005351118525221412,
+      "loss": 3.6494,
+      "step": 12720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005350607462486985,
+      "loss": 3.7582,
+      "step": 12725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005350096222998924,
+      "loss": 3.6472,
+      "step": 12730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005349584806795671,
+      "loss": 3.6468,
+      "step": 12735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005349073213915681,
+      "loss": 3.6471,
+      "step": 12740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005348561444397425,
+      "loss": 3.7198,
+      "step": 12745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005348049498279384,
+      "loss": 3.5422,
+      "step": 12750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005347537375600055,
+      "loss": 3.6621,
+      "step": 12755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005347025076397945,
+      "loss": 3.6559,
+      "step": 12760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005346512600711579,
+      "loss": 3.6488,
+      "step": 12765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000534599994857949,
+      "loss": 3.6929,
+      "step": 12770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005345487120040229,
+      "loss": 3.605,
+      "step": 12775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005344974115132356,
+      "loss": 3.7029,
+      "step": 12780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005344460933894448,
+      "loss": 3.5666,
+      "step": 12785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005343947576365093,
+      "loss": 3.6138,
+      "step": 12790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005343434042582892,
+      "loss": 3.6309,
+      "step": 12795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005342920332586462,
+      "loss": 3.6865,
+      "step": 12800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005342406446414429,
+      "loss": 3.5163,
+      "step": 12805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005341892384105436,
+      "loss": 3.7673,
+      "step": 12810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005341378145698139,
+      "loss": 3.7516,
+      "step": 12815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005340863731231204,
+      "loss": 3.6601,
+      "step": 12820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005340349140743314,
+      "loss": 3.6487,
+      "step": 12825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005339834374273162,
+      "loss": 3.594,
+      "step": 12830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005339319431859455,
+      "loss": 3.6715,
+      "step": 12835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005338804313540918,
+      "loss": 3.6531,
+      "step": 12840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005338289019356283,
+      "loss": 3.6891,
+      "step": 12845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005337773549344294,
+      "loss": 3.722,
+      "step": 12850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005337257903543719,
+      "loss": 3.6754,
+      "step": 12855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005336742081993325,
+      "loss": 3.723,
+      "step": 12860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005336226084731903,
+      "loss": 3.5262,
+      "step": 12865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005335709911798253,
+      "loss": 3.6648,
+      "step": 12870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005335193563231188,
+      "loss": 3.755,
+      "step": 12875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005334677039069534,
+      "loss": 3.6505,
+      "step": 12880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005334160339352131,
+      "loss": 3.5853,
+      "step": 12885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005333643464117834,
+      "loss": 3.6742,
+      "step": 12890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005333126413405508,
+      "loss": 3.7451,
+      "step": 12895
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005332609187254033,
+      "loss": 3.5786,
+      "step": 12900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005332091785702301,
+      "loss": 3.6852,
+      "step": 12905
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005331574208789219,
+      "loss": 3.7323,
+      "step": 12910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005331056456553705,
+      "loss": 3.5534,
+      "step": 12915
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005330538529034692,
+      "loss": 3.6558,
+      "step": 12920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005330020426271126,
+      "loss": 3.5388,
+      "step": 12925
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005329502148301965,
+      "loss": 3.7418,
+      "step": 12930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005328983695166181,
+      "loss": 3.6543,
+      "step": 12935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005328465066902759,
+      "loss": 3.6607,
+      "step": 12940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005327946263550698,
+      "loss": 3.796,
+      "step": 12945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005327427285149006,
+      "loss": 3.6999,
+      "step": 12950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005326908131736712,
+      "loss": 3.5925,
+      "step": 12955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005326388803352851,
+      "loss": 3.675,
+      "step": 12960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005325869300036475,
+      "loss": 3.72,
+      "step": 12965
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005325349621826648,
+      "loss": 3.7205,
+      "step": 12970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005324829768762445,
+      "loss": 3.6847,
+      "step": 12975
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000532430974088296,
+      "loss": 3.7384,
+      "step": 12980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005323789538227293,
+      "loss": 3.6207,
+      "step": 12985
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005323269160834562,
+      "loss": 3.6368,
+      "step": 12990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005322748608743897,
+      "loss": 3.5439,
+      "step": 12995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005322227881994441,
+      "loss": 3.6824,
+      "step": 13000
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 3.6955907344818115,
+      "eval_runtime": 150.2757,
+      "eval_samples_per_second": 12.251,
+      "eval_steps_per_second": 0.772,
+      "step": 13000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005321706980625349,
+      "loss": 3.6233,
+      "step": 13005
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005321185904675791,
+      "loss": 3.6698,
+      "step": 13010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005320664654184947,
+      "loss": 3.693,
+      "step": 13015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005320143229192016,
+      "loss": 3.6562,
+      "step": 13020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005319621629736205,
+      "loss": 3.7259,
+      "step": 13025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005319099855856734,
+      "loss": 3.6133,
+      "step": 13030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005318577907592841,
+      "loss": 3.626,
+      "step": 13035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000531805578498377,
+      "loss": 3.6176,
+      "step": 13040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005317533488068784,
+      "loss": 3.7022,
+      "step": 13045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005317011016887157,
+      "loss": 3.5367,
+      "step": 13050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005316488371478177,
+      "loss": 3.6384,
+      "step": 13055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005315965551881143,
+      "loss": 3.6755,
+      "step": 13060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005315442558135368,
+      "loss": 3.7178,
+      "step": 13065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005314919390280179,
+      "loss": 3.6934,
+      "step": 13070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005314396048354916,
+      "loss": 3.634,
+      "step": 13075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005313872532398932,
+      "loss": 3.5333,
+      "step": 13080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000531334884245159,
+      "loss": 3.6866,
+      "step": 13085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005312824978552272,
+      "loss": 3.6536,
+      "step": 13090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005312300940740369,
+      "loss": 3.7084,
+      "step": 13095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005311776729055286,
+      "loss": 3.637,
+      "step": 13100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005311252343536439,
+      "loss": 3.7074,
+      "step": 13105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005310727784223261,
+      "loss": 3.7197,
+      "step": 13110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005310203051155198,
+      "loss": 3.6676,
+      "step": 13115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005309678144371703,
+      "loss": 3.5821,
+      "step": 13120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000530915306391225,
+      "loss": 3.6428,
+      "step": 13125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005308627809816319,
+      "loss": 3.7832,
+      "step": 13130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000530810238212341,
+      "loss": 3.6682,
+      "step": 13135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000530757678087303,
+      "loss": 3.6524,
+      "step": 13140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005307051006104701,
+      "loss": 3.6169,
+      "step": 13145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005306525057857961,
+      "loss": 3.6496,
+      "step": 13150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005305998936172357,
+      "loss": 3.6896,
+      "step": 13155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005305472641087452,
+      "loss": 3.6579,
+      "step": 13160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000530494617264282,
+      "loss": 3.7858,
+      "step": 13165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005304419530878046,
+      "loss": 3.7619,
+      "step": 13170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005303892715832736,
+      "loss": 3.784,
+      "step": 13175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005303365727546499,
+      "loss": 3.7082,
+      "step": 13180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005302838566058964,
+      "loss": 3.6019,
+      "step": 13185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005302311231409773,
+      "loss": 3.641,
+      "step": 13190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005301783723638574,
+      "loss": 3.6698,
+      "step": 13195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005301256042785036,
+      "loss": 3.6359,
+      "step": 13200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005300728188888838,
+      "loss": 3.7043,
+      "step": 13205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005300200161989672,
+      "loss": 3.7608,
+      "step": 13210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000529967196212724,
+      "loss": 3.6878,
+      "step": 13215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005299143589341263,
+      "loss": 3.6146,
+      "step": 13220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005298615043671472,
+      "loss": 3.7112,
+      "step": 13225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005298086325157609,
+      "loss": 3.6229,
+      "step": 13230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005297557433839432,
+      "loss": 3.6452,
+      "step": 13235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005297028369756711,
+      "loss": 3.6943,
+      "step": 13240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005296499132949227,
+      "loss": 3.6575,
+      "step": 13245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005295969723456781,
+      "loss": 3.5772,
+      "step": 13250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005295440141319176,
+      "loss": 3.6112,
+      "step": 13255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005294910386576237,
+      "loss": 3.7853,
+      "step": 13260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005294380459267797,
+      "loss": 3.6779,
+      "step": 13265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005293850359433706,
+      "loss": 3.7071,
+      "step": 13270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005293320087113824,
+      "loss": 3.7668,
+      "step": 13275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005292789642348024,
+      "loss": 3.7479,
+      "step": 13280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005292259025176194,
+      "loss": 3.6091,
+      "step": 13285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005291728235638232,
+      "loss": 3.6423,
+      "step": 13290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005291197273774053,
+      "loss": 3.631,
+      "step": 13295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000529066613962358,
+      "loss": 3.6896,
+      "step": 13300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005290134833226752,
+      "loss": 3.726,
+      "step": 13305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005289603354623523,
+      "loss": 3.6359,
+      "step": 13310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005289071703853855,
+      "loss": 3.6416,
+      "step": 13315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005288539880957727,
+      "loss": 3.7228,
+      "step": 13320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005288007885975127,
+      "loss": 3.6365,
+      "step": 13325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005287475718946061,
+      "loss": 3.7192,
+      "step": 13330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005286943379910543,
+      "loss": 3.7571,
+      "step": 13335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005286410868908604,
+      "loss": 3.6936,
+      "step": 13340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005285878185980285,
+      "loss": 3.6531,
+      "step": 13345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005285345331165641,
+      "loss": 3.6293,
+      "step": 13350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005284812304504742,
+      "loss": 3.7047,
+      "step": 13355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005284279106037666,
+      "loss": 3.6494,
+      "step": 13360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005283745735804508,
+      "loss": 3.7135,
+      "step": 13365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005283212193845374,
+      "loss": 3.6306,
+      "step": 13370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005282678480200385,
+      "loss": 3.5678,
+      "step": 13375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005282144594909674,
+      "loss": 3.6169,
+      "step": 13380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005281610538013384,
+      "loss": 3.6236,
+      "step": 13385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005281076309551674,
+      "loss": 3.6551,
+      "step": 13390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005280541909564717,
+      "loss": 3.7224,
+      "step": 13395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005280007338092696,
+      "loss": 3.5341,
+      "step": 13400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005279472595175807,
+      "loss": 3.6211,
+      "step": 13405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005278937680854262,
+      "loss": 3.5654,
+      "step": 13410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005278402595168282,
+      "loss": 3.7164,
+      "step": 13415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005277867338158104,
+      "loss": 3.7283,
+      "step": 13420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005277331909863976,
+      "loss": 3.6716,
+      "step": 13425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000527679631032616,
+      "loss": 3.7008,
+      "step": 13430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005276260539584928,
+      "loss": 3.6196,
+      "step": 13435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005275724597680571,
+      "loss": 3.6388,
+      "step": 13440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005275188484653386,
+      "loss": 3.6706,
+      "step": 13445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005274652200543688,
+      "loss": 3.6409,
+      "step": 13450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005274115745391801,
+      "loss": 3.5821,
+      "step": 13455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005273579119238065,
+      "loss": 3.6809,
+      "step": 13460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005273042322122832,
+      "loss": 3.7153,
+      "step": 13465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005272505354086465,
+      "loss": 3.6738,
+      "step": 13470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005271968215169341,
+      "loss": 3.6512,
+      "step": 13475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005271430905411851,
+      "loss": 3.6398,
+      "step": 13480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005270893424854399,
+      "loss": 3.5679,
+      "step": 13485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005270355773537398,
+      "loss": 3.6169,
+      "step": 13490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000526981795150128,
+      "loss": 3.6797,
+      "step": 13495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005269279958786484,
+      "loss": 3.565,
+      "step": 13500
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 3.6858224868774414,
+      "eval_runtime": 149.7698,
+      "eval_samples_per_second": 12.292,
+      "eval_steps_per_second": 0.775,
+      "step": 13500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005268741795433464,
+      "loss": 3.7488,
+      "step": 13505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005268203461482689,
+      "loss": 3.6822,
+      "step": 13510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005267664956974638,
+      "loss": 3.5031,
+      "step": 13515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005267126281949804,
+      "loss": 3.6234,
+      "step": 13520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005266587436448692,
+      "loss": 3.6111,
+      "step": 13525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000526604842051182,
+      "loss": 3.5746,
+      "step": 13530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005265509234179722,
+      "loss": 3.6438,
+      "step": 13535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005264969877492939,
+      "loss": 3.6441,
+      "step": 13540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005264430350492029,
+      "loss": 3.5713,
+      "step": 13545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005263890653217562,
+      "loss": 3.604,
+      "step": 13550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000526335078571012,
+      "loss": 3.6513,
+      "step": 13555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005262810748010298,
+      "loss": 3.6257,
+      "step": 13560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005262270540158705,
+      "loss": 3.6804,
+      "step": 13565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005261730162195962,
+      "loss": 3.7044,
+      "step": 13570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005261189614162701,
+      "loss": 3.6778,
+      "step": 13575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000526064889609957,
+      "loss": 3.6329,
+      "step": 13580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005260108008047227,
+      "loss": 3.6702,
+      "step": 13585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005259566950046344,
+      "loss": 3.6102,
+      "step": 13590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005259025722137608,
+      "loss": 3.5786,
+      "step": 13595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005258484324361715,
+      "loss": 3.599,
+      "step": 13600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005257942756759375,
+      "loss": 3.6527,
+      "step": 13605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005257401019371312,
+      "loss": 3.7808,
+      "step": 13610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005256859112238259,
+      "loss": 3.5285,
+      "step": 13615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000525631703540097,
+      "loss": 3.6049,
+      "step": 13620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005255774788900201,
+      "loss": 3.431,
+      "step": 13625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000525523237277673,
+      "loss": 3.5476,
+      "step": 13630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005254689787071342,
+      "loss": 3.5949,
+      "step": 13635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005254147031824837,
+      "loss": 3.7066,
+      "step": 13640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005253604107078027,
+      "loss": 3.6354,
+      "step": 13645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005253061012871738,
+      "loss": 3.5766,
+      "step": 13650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005252517749246809,
+      "loss": 3.7445,
+      "step": 13655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005251974316244088,
+      "loss": 3.7481,
+      "step": 13660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000525143071390444,
+      "loss": 3.521,
+      "step": 13665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000525088694226874,
+      "loss": 3.7559,
+      "step": 13670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005250343001377877,
+      "loss": 3.6996,
+      "step": 13675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005249798891272755,
+      "loss": 3.6476,
+      "step": 13680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005249254611994286,
+      "loss": 3.7529,
+      "step": 13685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005248710163583396,
+      "loss": 3.7192,
+      "step": 13690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005248165546081027,
+      "loss": 3.6813,
+      "step": 13695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005247620759528131,
+      "loss": 3.6612,
+      "step": 13700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005247075803965672,
+      "loss": 3.5626,
+      "step": 13705
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000524653067943463,
+      "loss": 3.6624,
+      "step": 13710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005245985385975993,
+      "loss": 3.671,
+      "step": 13715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005245439923630765,
+      "loss": 3.5639,
+      "step": 13720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005244894292439963,
+      "loss": 3.5779,
+      "step": 13725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005244348492444615,
+      "loss": 3.6789,
+      "step": 13730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005243802523685763,
+      "loss": 3.6376,
+      "step": 13735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005243256386204461,
+      "loss": 3.5759,
+      "step": 13740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005242710080041774,
+      "loss": 3.6651,
+      "step": 13745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005242163605238783,
+      "loss": 3.7021,
+      "step": 13750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000524161696183658,
+      "loss": 3.5692,
+      "step": 13755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005241070149876269,
+      "loss": 3.6352,
+      "step": 13760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000524052316939897,
+      "loss": 3.799,
+      "step": 13765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000523997602044581,
+      "loss": 3.6014,
+      "step": 13770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005239428703057933,
+      "loss": 3.5783,
+      "step": 13775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005238881217276493,
+      "loss": 3.6207,
+      "step": 13780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005238333563142661,
+      "loss": 3.652,
+      "step": 13785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005237785740697617,
+      "loss": 3.5884,
+      "step": 13790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005237237749982552,
+      "loss": 3.6441,
+      "step": 13795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005236689591038674,
+      "loss": 3.6444,
+      "step": 13800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005236141263907203,
+      "loss": 3.5998,
+      "step": 13805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005235592768629368,
+      "loss": 3.6522,
+      "step": 13810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005235044105246415,
+      "loss": 3.6656,
+      "step": 13815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005234495273799598,
+      "loss": 3.6451,
+      "step": 13820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000523394627433019,
+      "loss": 3.7121,
+      "step": 13825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005233397106879469,
+      "loss": 4.2102,
+      "step": 13830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005232847771488733,
+      "loss": 4.3941,
+      "step": 13835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005232298268199288,
+      "loss": 4.4938,
+      "step": 13840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005231748597052452,
+      "loss": 4.4764,
+      "step": 13845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000523119875808956,
+      "loss": 4.1145,
+      "step": 13850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005230648751351956,
+      "loss": 3.8547,
+      "step": 13855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005230098576880997,
+      "loss": 3.8067,
+      "step": 13860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005229548234718056,
+      "loss": 3.6504,
+      "step": 13865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005228997724904512,
+      "loss": 3.7026,
+      "step": 13870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005228447047481763,
+      "loss": 3.7336,
+      "step": 13875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005227896202491218,
+      "loss": 3.6148,
+      "step": 13880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005227345189974295,
+      "loss": 3.7376,
+      "step": 13885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000522679400997243,
+      "loss": 3.7209,
+      "step": 13890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005226242662527066,
+      "loss": 3.7498,
+      "step": 13895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005225691147679664,
+      "loss": 3.7111,
+      "step": 13900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005225139465471694,
+      "loss": 3.6576,
+      "step": 13905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000522458761594464,
+      "loss": 3.6185,
+      "step": 13910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005224035599139997,
+      "loss": 3.641,
+      "step": 13915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005223483415099276,
+      "loss": 3.704,
+      "step": 13920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005222931063863998,
+      "loss": 3.6244,
+      "step": 13925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005222378545475695,
+      "loss": 3.5334,
+      "step": 13930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005221825859975915,
+      "loss": 3.651,
+      "step": 13935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005221273007406218,
+      "loss": 3.5782,
+      "step": 13940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005220719987808174,
+      "loss": 3.6259,
+      "step": 13945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005220166801223369,
+      "loss": 3.6128,
+      "step": 13950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005219613447693396,
+      "loss": 3.6453,
+      "step": 13955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005219059927259869,
+      "loss": 3.7525,
+      "step": 13960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005218506239964409,
+      "loss": 3.6357,
+      "step": 13965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005217952385848648,
+      "loss": 3.6617,
+      "step": 13970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005217398364954234,
+      "loss": 3.5647,
+      "step": 13975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005216844177322828,
+      "loss": 3.6491,
+      "step": 13980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00052162898229961,
+      "loss": 3.644,
+      "step": 13985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005215735302015737,
+      "loss": 3.7162,
+      "step": 13990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005215180614423434,
+      "loss": 3.5794,
+      "step": 13995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005214625760260901,
+      "loss": 3.5723,
+      "step": 14000
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 3.667832851409912,
+      "eval_runtime": 149.7768,
+      "eval_samples_per_second": 12.292,
+      "eval_steps_per_second": 0.774,
+      "step": 14000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005214070739569861,
+      "loss": 3.5753,
+      "step": 14005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005213515552392048,
+      "loss": 3.6657,
+      "step": 14010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005212960198769209,
+      "loss": 3.6619,
+      "step": 14015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005212404678743105,
+      "loss": 3.6746,
+      "step": 14020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000521184899235551,
+      "loss": 3.6758,
+      "step": 14025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005211293139648203,
+      "loss": 3.5972,
+      "step": 14030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005210737120662985,
+      "loss": 3.5256,
+      "step": 14035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005210180935441667,
+      "loss": 3.7113,
+      "step": 14040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005209624584026068,
+      "loss": 3.5753,
+      "step": 14045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005209068066458026,
+      "loss": 3.6638,
+      "step": 14050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005208511382779385,
+      "loss": 3.6401,
+      "step": 14055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005207954533032008,
+      "loss": 3.659,
+      "step": 14060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005207397517257765,
+      "loss": 3.7119,
+      "step": 14065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005206840335498541,
+      "loss": 3.6507,
+      "step": 14070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005206282987796234,
+      "loss": 3.5722,
+      "step": 14075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005205725474192753,
+      "loss": 3.6486,
+      "step": 14080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005205167794730022,
+      "loss": 3.772,
+      "step": 14085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005204609949449972,
+      "loss": 3.6892,
+      "step": 14090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005204051938394554,
+      "loss": 3.7151,
+      "step": 14095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005203493761605724,
+      "loss": 3.6335,
+      "step": 14100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005202935419125458,
+      "loss": 3.68,
+      "step": 14105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005202376910995736,
+      "loss": 3.6781,
+      "step": 14110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005201818237258558,
+      "loss": 3.7159,
+      "step": 14115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005201259397955934,
+      "loss": 3.6356,
+      "step": 14120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005200700393129884,
+      "loss": 3.5929,
+      "step": 14125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005200141222822443,
+      "loss": 3.702,
+      "step": 14130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005199581887075657,
+      "loss": 3.572,
+      "step": 14135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005199022385931585,
+      "loss": 3.5301,
+      "step": 14140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005198462719432301,
+      "loss": 3.7308,
+      "step": 14145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005197902887619887,
+      "loss": 3.6635,
+      "step": 14150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005197342890536441,
+      "loss": 3.6117,
+      "step": 14155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005196782728224069,
+      "loss": 3.6756,
+      "step": 14160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005196222400724894,
+      "loss": 3.6662,
+      "step": 14165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005195661908081052,
+      "loss": 3.5442,
+      "step": 14170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005195101250334684,
+      "loss": 3.6221,
+      "step": 14175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005194540427527955,
+      "loss": 3.7254,
+      "step": 14180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005193979439703031,
+      "loss": 3.6972,
+      "step": 14185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005193418286902098,
+      "loss": 3.5639,
+      "step": 14190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005192856969167349,
+      "loss": 3.7267,
+      "step": 14195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005192295486540995,
+      "loss": 3.6168,
+      "step": 14200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005191733839065257,
+      "loss": 3.5785,
+      "step": 14205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005191172026782365,
+      "loss": 3.6425,
+      "step": 14210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005190610049734568,
+      "loss": 3.6147,
+      "step": 14215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005190047907964121,
+      "loss": 3.6009,
+      "step": 14220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005189485601513295,
+      "loss": 3.6146,
+      "step": 14225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005188923130424372,
+      "loss": 3.6774,
+      "step": 14230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005188360494739649,
+      "loss": 3.6255,
+      "step": 14235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005187797694501431,
+      "loss": 3.4911,
+      "step": 14240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005187234729752039,
+      "loss": 3.5676,
+      "step": 14245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005186671600533805,
+      "loss": 3.6448,
+      "step": 14250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005186108306889073,
+      "loss": 3.5538,
+      "step": 14255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00051855448488602,
+      "loss": 3.4816,
+      "step": 14260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005184981226489555,
+      "loss": 3.5797,
+      "step": 14265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005184417439819519,
+      "loss": 3.4676,
+      "step": 14270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005183853488892488,
+      "loss": 3.6309,
+      "step": 14275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005183289373750866,
+      "loss": 3.5128,
+      "step": 14280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005182725094437072,
+      "loss": 3.6644,
+      "step": 14285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005182160650993537,
+      "loss": 3.6486,
+      "step": 14290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005181596043462704,
+      "loss": 3.6108,
+      "step": 14295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000518103127188703,
+      "loss": 3.5354,
+      "step": 14300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005180466336308983,
+      "loss": 3.5014,
+      "step": 14305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000517990123677104,
+      "loss": 3.416,
+      "step": 14310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005179335973315697,
+      "loss": 3.5127,
+      "step": 14315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005178770545985458,
+      "loss": 3.6262,
+      "step": 14320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005178204954822839,
+      "loss": 3.6022,
+      "step": 14325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000517763919987037,
+      "loss": 3.6538,
+      "step": 14330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005177073281170595,
+      "loss": 3.6462,
+      "step": 14335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005176507198766066,
+      "loss": 3.5705,
+      "step": 14340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005175940952699351,
+      "loss": 3.5812,
+      "step": 14345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005175374543013025,
+      "loss": 3.6557,
+      "step": 14350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005174807969749685,
+      "loss": 3.6208,
+      "step": 14355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000517424123295193,
+      "loss": 3.5958,
+      "step": 14360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005173674332662377,
+      "loss": 3.5134,
+      "step": 14365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005173107268923655,
+      "loss": 3.477,
+      "step": 14370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005172540041778403,
+      "loss": 3.5057,
+      "step": 14375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005171972651269273,
+      "loss": 3.6122,
+      "step": 14380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005171405097438933,
+      "loss": 3.6689,
+      "step": 14385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005170837380330056,
+      "loss": 3.5244,
+      "step": 14390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005170269499985334,
+      "loss": 3.6451,
+      "step": 14395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005169701456447468,
+      "loss": 3.6317,
+      "step": 14400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005169133249759171,
+      "loss": 3.5989,
+      "step": 14405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005168564879963172,
+      "loss": 3.6657,
+      "step": 14410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005167996347102207,
+      "loss": 3.6195,
+      "step": 14415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005167427651219027,
+      "loss": 3.6496,
+      "step": 14420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005166858792356396,
+      "loss": 3.719,
+      "step": 14425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005166289770557089,
+      "loss": 3.5693,
+      "step": 14430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005165720585863893,
+      "loss": 3.5917,
+      "step": 14435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005165151238319608,
+      "loss": 3.5493,
+      "step": 14440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005164581727967046,
+      "loss": 3.5514,
+      "step": 14445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005164012054849032,
+      "loss": 3.5368,
+      "step": 14450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005163442219008402,
+      "loss": 3.5879,
+      "step": 14455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005162872220488004,
+      "loss": 3.6387,
+      "step": 14460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00051623020593307,
+      "loss": 3.5715,
+      "step": 14465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005161731735579361,
+      "loss": 3.6949,
+      "step": 14470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005161161249276876,
+      "loss": 3.6781,
+      "step": 14475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005160590600466141,
+      "loss": 3.537,
+      "step": 14480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005160019789190065,
+      "loss": 3.664,
+      "step": 14485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005159448815491571,
+      "loss": 3.6726,
+      "step": 14490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005158877679413593,
+      "loss": 3.6492,
+      "step": 14495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005158306380999078,
+      "loss": 3.5562,
+      "step": 14500
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 3.648855447769165,
+      "eval_runtime": 149.8786,
+      "eval_samples_per_second": 12.283,
+      "eval_steps_per_second": 0.774,
+      "step": 14500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005157734920290984,
+      "loss": 3.634,
+      "step": 14505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005157163297332282,
+      "loss": 3.5898,
+      "step": 14510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005156591512165955,
+      "loss": 3.6621,
+      "step": 14515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005156019564834999,
+      "loss": 3.6402,
+      "step": 14520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005155447455382421,
+      "loss": 3.6102,
+      "step": 14525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005154875183851241,
+      "loss": 3.624,
+      "step": 14530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005154302750284491,
+      "loss": 3.6164,
+      "step": 14535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005153730154725214,
+      "loss": 3.4685,
+      "step": 14540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005153157397216468,
+      "loss": 3.6154,
+      "step": 14545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000515258447780132,
+      "loss": 3.6433,
+      "step": 14550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005152011396522851,
+      "loss": 3.5996,
+      "step": 14555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005151438153424155,
+      "loss": 3.6255,
+      "step": 14560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005150864748548335,
+      "loss": 3.5906,
+      "step": 14565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005150291181938508,
+      "loss": 3.6043,
+      "step": 14570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005149717453637805,
+      "loss": 3.4846,
+      "step": 14575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005149143563689367,
+      "loss": 3.5817,
+      "step": 14580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005148569512136346,
+      "loss": 3.5475,
+      "step": 14585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005147995299021911,
+      "loss": 3.6521,
+      "step": 14590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005147420924389236,
+      "loss": 3.6627,
+      "step": 14595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005146846388281514,
+      "loss": 3.6745,
+      "step": 14600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005146271690741945,
+      "loss": 3.6828,
+      "step": 14605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005145696831813744,
+      "loss": 3.6828,
+      "step": 14610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005145121811540139,
+      "loss": 3.5683,
+      "step": 14615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005144546629964367,
+      "loss": 3.6632,
+      "step": 14620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005143971287129679,
+      "loss": 3.6587,
+      "step": 14625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005143395783079337,
+      "loss": 3.603,
+      "step": 14630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005142820117856618,
+      "loss": 3.6555,
+      "step": 14635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005142244291504808,
+      "loss": 3.7154,
+      "step": 14640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005141668304067205,
+      "loss": 3.6161,
+      "step": 14645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005141092155587123,
+      "loss": 3.6059,
+      "step": 14650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005140515846107882,
+      "loss": 3.6758,
+      "step": 14655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000513993937567282,
+      "loss": 3.6174,
+      "step": 14660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005139362744325285,
+      "loss": 3.5476,
+      "step": 14665
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005138785952108635,
+      "loss": 3.6128,
+      "step": 14670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005138208999066242,
+      "loss": 3.5876,
+      "step": 14675
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005137631885241491,
+      "loss": 3.5896,
+      "step": 14680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005137054610677777,
+      "loss": 3.5601,
+      "step": 14685
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005136477175418508,
+      "loss": 3.6276,
+      "step": 14690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005135899579507106,
+      "loss": 3.5822,
+      "step": 14695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005135321822987001,
+      "loss": 3.4708,
+      "step": 14700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005134743905901639,
+      "loss": 3.6286,
+      "step": 14705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005134165828294475,
+      "loss": 3.5722,
+      "step": 14710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005133587590208979,
+      "loss": 3.6132,
+      "step": 14715
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005133009191688629,
+      "loss": 3.7012,
+      "step": 14720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000513243063277692,
+      "loss": 3.6159,
+      "step": 14725
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005131851913517358,
+      "loss": 3.6005,
+      "step": 14730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005131273033953456,
+      "loss": 3.592,
+      "step": 14735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005130693994128744,
+      "loss": 3.6199,
+      "step": 14740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005130114794086763,
+      "loss": 3.5501,
+      "step": 14745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005129535433871067,
+      "loss": 3.6375,
+      "step": 14750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000512895591352522,
+      "loss": 3.5636,
+      "step": 14755
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00051283762330928,
+      "loss": 3.6531,
+      "step": 14760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005127796392617393,
+      "loss": 3.5518,
+      "step": 14765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005127216392142604,
+      "loss": 3.5923,
+      "step": 14770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005126636231712042,
+      "loss": 3.5783,
+      "step": 14775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005126055911369335,
+      "loss": 3.69,
+      "step": 14780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005125475431158121,
+      "loss": 3.7556,
+      "step": 14785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005124894791122045,
+      "loss": 3.6912,
+      "step": 14790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005124313991304773,
+      "loss": 3.6166,
+      "step": 14795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005123733031749974,
+      "loss": 3.5407,
+      "step": 14800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005123151912501336,
+      "loss": 3.6813,
+      "step": 14805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005122570633602554,
+      "loss": 3.6641,
+      "step": 14810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005121989195097339,
+      "loss": 3.6209,
+      "step": 14815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005121407597029413,
+      "loss": 3.6123,
+      "step": 14820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005120825839442506,
+      "loss": 3.5358,
+      "step": 14825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005120243922380366,
+      "loss": 3.6142,
+      "step": 14830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005119661845886747,
+      "loss": 3.5662,
+      "step": 14835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005119079610005422,
+      "loss": 3.6078,
+      "step": 14840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005118497214780171,
+      "loss": 3.6417,
+      "step": 14845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005117914660254785,
+      "loss": 3.5625,
+      "step": 14850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005117331946473071,
+      "loss": 3.619,
+      "step": 14855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005116749073478846,
+      "loss": 3.616,
+      "step": 14860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005116166041315939,
+      "loss": 3.5866,
+      "step": 14865
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005115582850028192,
+      "loss": 3.6377,
+      "step": 14870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005114999499659454,
+      "loss": 3.6135,
+      "step": 14875
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005114415990253595,
+      "loss": 3.5777,
+      "step": 14880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005113832321854491,
+      "loss": 3.5029,
+      "step": 14885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005113248494506027,
+      "loss": 3.5864,
+      "step": 14890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005112664508252107,
+      "loss": 3.5529,
+      "step": 14895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005112080363136644,
+      "loss": 3.588,
+      "step": 14900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005111496059203561,
+      "loss": 3.614,
+      "step": 14905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005110911596496797,
+      "loss": 3.6759,
+      "step": 14910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005110326975060298,
+      "loss": 3.4662,
+      "step": 14915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005109742194938026,
+      "loss": 3.6857,
+      "step": 14920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005109157256173954,
+      "loss": 3.6664,
+      "step": 14925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005108572158812064,
+      "loss": 3.5142,
+      "step": 14930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005107986902896355,
+      "loss": 3.5656,
+      "step": 14935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005107401488470834,
+      "loss": 3.7301,
+      "step": 14940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005106815915579522,
+      "loss": 3.5186,
+      "step": 14945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005106230184266448,
+      "loss": 3.65,
+      "step": 14950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005105644294575661,
+      "loss": 3.6077,
+      "step": 14955
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005105058246551214,
+      "loss": 3.6377,
+      "step": 14960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005104472040237174,
+      "loss": 3.5337,
+      "step": 14965
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005103885675677622,
+      "loss": 3.5585,
+      "step": 14970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005103299152916651,
+      "loss": 3.6477,
+      "step": 14975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005102712471998361,
+      "loss": 3.5803,
+      "step": 14980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000510212563296687,
+      "loss": 3.5321,
+      "step": 14985
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005101538635866304,
+      "loss": 3.6679,
+      "step": 14990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005100951480740804,
+      "loss": 3.6513,
+      "step": 14995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005100364167634518,
+      "loss": 3.5916,
+      "step": 15000
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 3.638385534286499,
+      "eval_runtime": 149.9801,
+      "eval_samples_per_second": 12.275,
+      "eval_steps_per_second": 0.773,
+      "step": 15000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005099776696591613,
+      "loss": 3.6363,
+      "step": 15005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005099189067656261,
+      "loss": 3.5853,
+      "step": 15010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005098601280872649,
+      "loss": 3.6406,
+      "step": 15015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005098013336284975,
+      "loss": 3.5502,
+      "step": 15020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005097425233937451,
+      "loss": 3.6303,
+      "step": 15025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005096836973874298,
+      "loss": 3.5962,
+      "step": 15030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005096248556139753,
+      "loss": 3.4167,
+      "step": 15035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005095659980778056,
+      "loss": 3.5678,
+      "step": 15040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005095071247833473,
+      "loss": 3.5376,
+      "step": 15045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005094482357350266,
+      "loss": 3.6403,
+      "step": 15050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000509389330937272,
+      "loss": 3.6371,
+      "step": 15055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000509330410394513,
+      "loss": 3.5036,
+      "step": 15060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005092714741111797,
+      "loss": 3.6479,
+      "step": 15065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005092125220917042,
+      "loss": 3.6069,
+      "step": 15070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005091535543405192,
+      "loss": 3.6093,
+      "step": 15075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005090945708620588,
+      "loss": 3.6439,
+      "step": 15080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005090355716607582,
+      "loss": 3.5337,
+      "step": 15085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005089765567410539,
+      "loss": 3.604,
+      "step": 15090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005089175261073837,
+      "loss": 3.5584,
+      "step": 15095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005088584797641861,
+      "loss": 3.5309,
+      "step": 15100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005087994177159012,
+      "loss": 3.6083,
+      "step": 15105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005087403399669702,
+      "loss": 3.5913,
+      "step": 15110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005086812465218354,
+      "loss": 3.5442,
+      "step": 15115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005086221373849403,
+      "loss": 3.6318,
+      "step": 15120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005085630125607297,
+      "loss": 3.6189,
+      "step": 15125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005085038720536493,
+      "loss": 3.5056,
+      "step": 15130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005084447158681465,
+      "loss": 3.5779,
+      "step": 15135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005083855440086691,
+      "loss": 3.5665,
+      "step": 15140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005083263564796669,
+      "loss": 3.5793,
+      "step": 15145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005082671532855903,
+      "loss": 3.5825,
+      "step": 15150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005082079344308912,
+      "loss": 3.7259,
+      "step": 15155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005081486999200224,
+      "loss": 3.6756,
+      "step": 15160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005080894497574381,
+      "loss": 3.5817,
+      "step": 15165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005080301839475936,
+      "loss": 3.518,
+      "step": 15170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005079709024949455,
+      "loss": 3.5779,
+      "step": 15175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005079116054039513,
+      "loss": 3.7219,
+      "step": 15180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005078522926790698,
+      "loss": 3.5903,
+      "step": 15185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005077929643247613,
+      "loss": 3.5368,
+      "step": 15190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005077336203454867,
+      "loss": 3.6035,
+      "step": 15195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005076742607457086,
+      "loss": 3.6253,
+      "step": 15200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005076148855298903,
+      "loss": 3.6759,
+      "step": 15205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005075554947024967,
+      "loss": 3.6473,
+      "step": 15210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005074960882679935,
+      "loss": 3.5805,
+      "step": 15215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005074366662308478,
+      "loss": 3.5368,
+      "step": 15220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000507377228595528,
+      "loss": 3.5171,
+      "step": 15225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005073177753665034,
+      "loss": 3.6021,
+      "step": 15230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005072583065482446,
+      "loss": 3.6194,
+      "step": 15235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005071988221452233,
+      "loss": 3.6315,
+      "step": 15240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005071393221619124,
+      "loss": 3.6864,
+      "step": 15245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005070798066027862,
+      "loss": 3.5701,
+      "step": 15250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005070202754723196,
+      "loss": 3.6381,
+      "step": 15255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005069607287749893,
+      "loss": 3.572,
+      "step": 15260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000506901166515273,
+      "loss": 3.6789,
+      "step": 15265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005068415886976493,
+      "loss": 3.536,
+      "step": 15270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000506781995326598,
+      "loss": 3.5427,
+      "step": 15275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005067223864066005,
+      "loss": 3.5554,
+      "step": 15280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000506662761942139,
+      "loss": 3.5994,
+      "step": 15285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000506603121937697,
+      "loss": 3.6482,
+      "step": 15290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000506543466397759,
+      "loss": 3.6878,
+      "step": 15295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005064837953268108,
+      "loss": 3.489,
+      "step": 15300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005064241087293393,
+      "loss": 3.4993,
+      "step": 15305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005063644066098329,
+      "loss": 3.5804,
+      "step": 15310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005063046889727807,
+      "loss": 3.5341,
+      "step": 15315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005062449558226732,
+      "loss": 3.5108,
+      "step": 15320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005061852071640019,
+      "loss": 3.6813,
+      "step": 15325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005061254430012598,
+      "loss": 3.4356,
+      "step": 15330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005060656633389406,
+      "loss": 3.6614,
+      "step": 15335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005060058681815399,
+      "loss": 3.6952,
+      "step": 15340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005059460575335534,
+      "loss": 3.6666,
+      "step": 15345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000505886231399479,
+      "loss": 3.5999,
+      "step": 15350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000505826389783815,
+      "loss": 3.6896,
+      "step": 15355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005057665326910614,
+      "loss": 3.5638,
+      "step": 15360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005057066601257191,
+      "loss": 3.5589,
+      "step": 15365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005056467720922903,
+      "loss": 3.6379,
+      "step": 15370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000505586868595278,
+      "loss": 3.5751,
+      "step": 15375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005055269496391868,
+      "loss": 3.5328,
+      "step": 15380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005054670152285223,
+      "loss": 3.5601,
+      "step": 15385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005054070653677913,
+      "loss": 3.5779,
+      "step": 15390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005053471000615017,
+      "loss": 3.637,
+      "step": 15395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005052871193141625,
+      "loss": 3.5866,
+      "step": 15400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000505227123130284,
+      "loss": 3.6035,
+      "step": 15405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005051671115143776,
+      "loss": 3.6226,
+      "step": 15410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000505107084470956,
+      "loss": 3.5384,
+      "step": 15415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005050470420045327,
+      "loss": 3.512,
+      "step": 15420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005049869841196228,
+      "loss": 3.6033,
+      "step": 15425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005049269108207421,
+      "loss": 3.5151,
+      "step": 15430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005048668221124081,
+      "loss": 3.4719,
+      "step": 15435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005048067179991389,
+      "loss": 3.6271,
+      "step": 15440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000504746598485454,
+      "loss": 3.7077,
+      "step": 15445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005046864635758744,
+      "loss": 3.6447,
+      "step": 15450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005046263132749217,
+      "loss": 3.601,
+      "step": 15455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005045661475871189,
+      "loss": 3.6165,
+      "step": 15460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005045059665169902,
+      "loss": 3.6336,
+      "step": 15465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005044457700690609,
+      "loss": 3.4781,
+      "step": 15470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005043855582478574,
+      "loss": 3.5317,
+      "step": 15475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005043253310579075,
+      "loss": 3.5082,
+      "step": 15480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005042650885037397,
+      "loss": 3.6004,
+      "step": 15485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005042048305898843,
+      "loss": 3.6375,
+      "step": 15490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000504144557320872,
+      "loss": 3.4989,
+      "step": 15495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005040842687012351,
+      "loss": 3.6445,
+      "step": 15500
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 3.6320202350616455,
+      "eval_runtime": 150.4615,
+      "eval_samples_per_second": 12.236,
+      "eval_steps_per_second": 0.771,
+      "step": 15500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005040239647355073,
+      "loss": 3.521,
+      "step": 15505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000503963645428223,
+      "loss": 3.5892,
+      "step": 15510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005039033107839177,
+      "loss": 3.5733,
+      "step": 15515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005038429608071285,
+      "loss": 3.6047,
+      "step": 15520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005037825955023933,
+      "loss": 3.6021,
+      "step": 15525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005037222148742514,
+      "loss": 3.6138,
+      "step": 15530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005036618189272428,
+      "loss": 3.6236,
+      "step": 15535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005036014076659094,
+      "loss": 3.5677,
+      "step": 15540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005035409810947934,
+      "loss": 3.498,
+      "step": 15545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005034805392184389,
+      "loss": 3.5847,
+      "step": 15550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005034200820413906,
+      "loss": 3.6563,
+      "step": 15555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005033596095681946,
+      "loss": 3.6285,
+      "step": 15560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005032991218033981,
+      "loss": 3.524,
+      "step": 15565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005032386187515497,
+      "loss": 3.5815,
+      "step": 15570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005031781004171986,
+      "loss": 3.5129,
+      "step": 15575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005031175668048957,
+      "loss": 3.5887,
+      "step": 15580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005030570179191927,
+      "loss": 3.5245,
+      "step": 15585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005029964537646427,
+      "loss": 3.6023,
+      "step": 15590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005029358743457997,
+      "loss": 3.5239,
+      "step": 15595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005028752796672189,
+      "loss": 3.5323,
+      "step": 15600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005028146697334568,
+      "loss": 3.5991,
+      "step": 15605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000502754044549071,
+      "loss": 3.4643,
+      "step": 15610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005026934041186201,
+      "loss": 3.5993,
+      "step": 15615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005026327484466641,
+      "loss": 3.5821,
+      "step": 15620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005025720775377637,
+      "loss": 3.498,
+      "step": 15625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005025113913964813,
+      "loss": 3.5981,
+      "step": 15630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005024506900273803,
+      "loss": 3.5101,
+      "step": 15635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005023899734350248,
+      "loss": 3.5864,
+      "step": 15640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005023292416239805,
+      "loss": 3.5627,
+      "step": 15645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005022684945988141,
+      "loss": 3.5497,
+      "step": 15650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005022077323640937,
+      "loss": 3.6516,
+      "step": 15655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005021469549243881,
+      "loss": 3.4684,
+      "step": 15660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005020861622842673,
+      "loss": 3.5867,
+      "step": 15665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000502025354448303,
+      "loss": 3.5082,
+      "step": 15670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005019645314210672,
+      "loss": 3.6077,
+      "step": 15675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005019036932071338,
+      "loss": 3.5891,
+      "step": 15680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005018428398110774,
+      "loss": 3.6244,
+      "step": 15685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005017819712374739,
+      "loss": 3.4976,
+      "step": 15690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005017210874909003,
+      "loss": 3.6083,
+      "step": 15695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005016601885759346,
+      "loss": 3.5849,
+      "step": 15700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005015992744971562,
+      "loss": 3.6031,
+      "step": 15705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005015383452591456,
+      "loss": 3.5034,
+      "step": 15710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005014774008664844,
+      "loss": 3.5678,
+      "step": 15715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005014164413237552,
+      "loss": 3.6184,
+      "step": 15720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005013554666355417,
+      "loss": 3.6015,
+      "step": 15725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005012944768064291,
+      "loss": 3.6218,
+      "step": 15730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005012334718410036,
+      "loss": 3.8092,
+      "step": 15735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005011724517438521,
+      "loss": 4.6015,
+      "step": 15740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005011114165195634,
+      "loss": 4.2182,
+      "step": 15745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005010503661727268,
+      "loss": 3.7612,
+      "step": 15750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005009893007079329,
+      "loss": 3.7264,
+      "step": 15755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005009282201297739,
+      "loss": 3.6667,
+      "step": 15760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005008671244428423,
+      "loss": 3.6342,
+      "step": 15765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005008060136517324,
+      "loss": 3.6282,
+      "step": 15770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005007448877610395,
+      "loss": 3.6548,
+      "step": 15775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005006837467753597,
+      "loss": 3.6451,
+      "step": 15780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005006225906992906,
+      "loss": 3.7284,
+      "step": 15785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000500561419537431,
+      "loss": 3.6631,
+      "step": 15790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005005002332943804,
+      "loss": 3.6146,
+      "step": 15795
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005004390319747399,
+      "loss": 3.6857,
+      "step": 15800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005003778155831113,
+      "loss": 3.5865,
+      "step": 15805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000500316584124098,
+      "loss": 3.7437,
+      "step": 15810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005002553376023042,
+      "loss": 3.5996,
+      "step": 15815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005001940760223352,
+      "loss": 3.7551,
+      "step": 15820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005001327993887977,
+      "loss": 3.6479,
+      "step": 15825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005000715077062994,
+      "loss": 3.6632,
+      "step": 15830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005000102009794491,
+      "loss": 3.5554,
+      "step": 15835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004999488792128567,
+      "loss": 3.6261,
+      "step": 15840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004998875424111332,
+      "loss": 3.5204,
+      "step": 15845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004998261905788909,
+      "loss": 3.6914,
+      "step": 15850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004997648237207433,
+      "loss": 3.4878,
+      "step": 15855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004997034418413046,
+      "loss": 3.5916,
+      "step": 15860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004996420449451907,
+      "loss": 3.6535,
+      "step": 15865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000499580633037018,
+      "loss": 3.6224,
+      "step": 15870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004995192061214047,
+      "loss": 3.5223,
+      "step": 15875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004994577642029695,
+      "loss": 3.563,
+      "step": 15880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004993963072863326,
+      "loss": 3.561,
+      "step": 15885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004993348353761154,
+      "loss": 3.6209,
+      "step": 15890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004992733484769401,
+      "loss": 3.3646,
+      "step": 15895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004992118465934302,
+      "loss": 3.5317,
+      "step": 15900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004991503297302105,
+      "loss": 3.5534,
+      "step": 15905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004990887978919066,
+      "loss": 3.6341,
+      "step": 15910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004990272510831454,
+      "loss": 3.5312,
+      "step": 15915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004989656893085549,
+      "loss": 3.5533,
+      "step": 15920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004989041125727643,
+      "loss": 3.5574,
+      "step": 15925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004988425208804037,
+      "loss": 3.5899,
+      "step": 15930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004987809142361046,
+      "loss": 3.5567,
+      "step": 15935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004987192926444995,
+      "loss": 3.5264,
+      "step": 15940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000498657656110222,
+      "loss": 3.6206,
+      "step": 15945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004985960046379069,
+      "loss": 3.5102,
+      "step": 15950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004985343382321901,
+      "loss": 3.5516,
+      "step": 15955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004984726568977083,
+      "loss": 3.6478,
+      "step": 15960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004984109606391002,
+      "loss": 3.6414,
+      "step": 15965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004983492494610045,
+      "loss": 3.5679,
+      "step": 15970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004982875233680618,
+      "loss": 3.5373,
+      "step": 15975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004982257823649136,
+      "loss": 3.6789,
+      "step": 15980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004981640264562023,
+      "loss": 3.642,
+      "step": 15985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004981022556465719,
+      "loss": 3.5777,
+      "step": 15990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004980404699406672,
+      "loss": 3.6623,
+      "step": 15995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000497978669343134,
+      "loss": 3.6401,
+      "step": 16000
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 3.6124560832977295,
+      "eval_runtime": 150.0796,
+      "eval_samples_per_second": 12.267,
+      "eval_steps_per_second": 0.773,
+      "step": 16000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004979168538586195,
+      "loss": 3.5841,
+      "step": 16005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004978550234917719,
+      "loss": 3.595,
+      "step": 16010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004977931782472405,
+      "loss": 3.5558,
+      "step": 16015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004977313181296759,
+      "loss": 3.592,
+      "step": 16020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004976694431437295,
+      "loss": 3.5335,
+      "step": 16025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004976075532940539,
+      "loss": 3.4794,
+      "step": 16030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000497545648585303,
+      "loss": 3.6644,
+      "step": 16035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004974837290221318,
+      "loss": 3.5437,
+      "step": 16040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004974217946091962,
+      "loss": 3.6994,
+      "step": 16045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004973598453511535,
+      "loss": 3.6072,
+      "step": 16050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004972978812526619,
+      "loss": 3.5775,
+      "step": 16055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004972359023183808,
+      "loss": 3.5683,
+      "step": 16060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004971739085529705,
+      "loss": 3.4989,
+      "step": 16065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004971118999610929,
+      "loss": 3.5584,
+      "step": 16070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004970498765474106,
+      "loss": 3.5787,
+      "step": 16075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004969878383165873,
+      "loss": 3.6137,
+      "step": 16080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004969257852732883,
+      "loss": 3.7059,
+      "step": 16085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004968637174221794,
+      "loss": 3.6148,
+      "step": 16090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000496801634767928,
+      "loss": 3.5649,
+      "step": 16095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000496739537315202,
+      "loss": 3.6258,
+      "step": 16100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004966774250686713,
+      "loss": 3.5316,
+      "step": 16105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000496615298033006,
+      "loss": 3.5973,
+      "step": 16110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000496553156212878,
+      "loss": 3.6305,
+      "step": 16115
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004964909996129599,
+      "loss": 3.5401,
+      "step": 16120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004964288282379258,
+      "loss": 3.5479,
+      "step": 16125
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004963666420924504,
+      "loss": 3.5814,
+      "step": 16130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004963044411812099,
+      "loss": 3.4288,
+      "step": 16135
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004962422255088815,
+      "loss": 3.5357,
+      "step": 16140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004961799950801434,
+      "loss": 3.5193,
+      "step": 16145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000496117749899675,
+      "loss": 3.5514,
+      "step": 16150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000496055489972157,
+      "loss": 3.5557,
+      "step": 16155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004959932153022709,
+      "loss": 3.5929,
+      "step": 16160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004959309258946996,
+      "loss": 3.542,
+      "step": 16165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004958686217541266,
+      "loss": 3.5837,
+      "step": 16170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004958063028852372,
+      "loss": 3.4923,
+      "step": 16175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004957439692927173,
+      "loss": 3.6609,
+      "step": 16180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004956816209812539,
+      "loss": 3.4934,
+      "step": 16185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004956192579555356,
+      "loss": 3.5329,
+      "step": 16190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004955568802202517,
+      "loss": 3.5227,
+      "step": 16195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004954944877800924,
+      "loss": 3.5705,
+      "step": 16200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004954320806397497,
+      "loss": 3.5531,
+      "step": 16205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000495369658803916,
+      "loss": 3.4774,
+      "step": 16210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004953072222772853,
+      "loss": 3.5052,
+      "step": 16215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004952447710645524,
+      "loss": 3.5444,
+      "step": 16220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004951823051704134,
+      "loss": 3.5683,
+      "step": 16225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004951198245995652,
+      "loss": 3.4055,
+      "step": 16230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004950573293567062,
+      "loss": 3.6428,
+      "step": 16235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004949948194465357,
+      "loss": 3.6984,
+      "step": 16240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004949322948737541,
+      "loss": 3.5885,
+      "step": 16245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004948697556430631,
+      "loss": 3.4286,
+      "step": 16250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004948072017591649,
+      "loss": 3.565,
+      "step": 16255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004947446332267636,
+      "loss": 3.5906,
+      "step": 16260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000494682050050564,
+      "loss": 3.634,
+      "step": 16265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000494619452235272,
+      "loss": 3.5922,
+      "step": 16270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004945568397855945,
+      "loss": 3.538,
+      "step": 16275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004944942127062397,
+      "loss": 3.572,
+      "step": 16280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004944315710019169,
+      "loss": 3.6377,
+      "step": 16285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004943689146773365,
+      "loss": 3.6228,
+      "step": 16290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004943062437372098,
+      "loss": 3.5855,
+      "step": 16295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004942435581862493,
+      "loss": 3.5372,
+      "step": 16300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004941808580291688,
+      "loss": 3.5609,
+      "step": 16305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000494118143270683,
+      "loss": 3.6014,
+      "step": 16310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004940554139155074,
+      "loss": 3.5558,
+      "step": 16315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004939926699683594,
+      "loss": 3.5162,
+      "step": 16320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004939299114339568,
+      "loss": 3.6057,
+      "step": 16325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004938671383170187,
+      "loss": 3.6958,
+      "step": 16330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004938043506222653,
+      "loss": 3.6499,
+      "step": 16335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004937415483544181,
+      "loss": 3.5561,
+      "step": 16340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004936787315181993,
+      "loss": 3.5672,
+      "step": 16345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004936159001183325,
+      "loss": 3.5341,
+      "step": 16350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004935530541595422,
+      "loss": 3.5887,
+      "step": 16355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004934901936465543,
+      "loss": 3.5734,
+      "step": 16360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004934273185840953,
+      "loss": 3.5032,
+      "step": 16365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004933644289768933,
+      "loss": 3.6115,
+      "step": 16370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004933015248296772,
+      "loss": 3.5705,
+      "step": 16375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004932386061471772,
+      "loss": 3.6443,
+      "step": 16380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004931756729341242,
+      "loss": 3.4791,
+      "step": 16385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004931127251952507,
+      "loss": 3.6197,
+      "step": 16390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00049304976293529,
+      "loss": 3.4558,
+      "step": 16395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004929867861589764,
+      "loss": 3.5077,
+      "step": 16400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004929237948710455,
+      "loss": 3.5552,
+      "step": 16405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004928607890762341,
+      "loss": 3.5842,
+      "step": 16410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004927977687792796,
+      "loss": 3.5561,
+      "step": 16415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000492734733984921,
+      "loss": 3.4731,
+      "step": 16420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004926716846978982,
+      "loss": 3.5924,
+      "step": 16425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004926086209229521,
+      "loss": 3.4558,
+      "step": 16430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004925455426648248,
+      "loss": 3.457,
+      "step": 16435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004924824499282595,
+      "loss": 3.5853,
+      "step": 16440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004924193427180004,
+      "loss": 3.6165,
+      "step": 16445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004923562210387928,
+      "loss": 3.4844,
+      "step": 16450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004922930848953834,
+      "loss": 3.629,
+      "step": 16455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004922299342925193,
+      "loss": 3.4659,
+      "step": 16460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004921667692349493,
+      "loss": 3.6049,
+      "step": 16465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004921035897274232,
+      "loss": 3.5527,
+      "step": 16470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004920403957746916,
+      "loss": 3.5113,
+      "step": 16475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004919771873815065,
+      "loss": 3.5445,
+      "step": 16480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004919139645526207,
+      "loss": 3.5608,
+      "step": 16485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004918507272927883,
+      "loss": 3.5982,
+      "step": 16490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004917874756067644,
+      "loss": 3.5369,
+      "step": 16495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004917242094993054,
+      "loss": 3.4741,
+      "step": 16500
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 3.5877273082733154,
+      "eval_runtime": 150.0518,
+      "eval_samples_per_second": 12.269,
+      "eval_steps_per_second": 0.773,
+      "step": 16500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004916609289751683,
+      "loss": 3.5372,
+      "step": 16505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004915976340391116,
+      "loss": 3.5657,
+      "step": 16510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004915343246958947,
+      "loss": 3.5311,
+      "step": 16515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004914710009502782,
+      "loss": 3.5391,
+      "step": 16520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004914076628070238,
+      "loss": 3.6073,
+      "step": 16525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000491344310270894,
+      "loss": 3.4663,
+      "step": 16530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004912809433466527,
+      "loss": 3.5429,
+      "step": 16535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004912175620390648,
+      "loss": 3.5616,
+      "step": 16540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004911541663528961,
+      "loss": 3.5202,
+      "step": 16545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004910907562929139,
+      "loss": 3.6226,
+      "step": 16550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004910273318638861,
+      "loss": 3.5045,
+      "step": 16555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004909638930705819,
+      "loss": 3.4955,
+      "step": 16560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004909004399177717,
+      "loss": 3.5726,
+      "step": 16565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004908369724102267,
+      "loss": 3.4443,
+      "step": 16570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004907734905527193,
+      "loss": 3.5395,
+      "step": 16575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004907099943500233,
+      "loss": 3.4818,
+      "step": 16580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000490646483806913,
+      "loss": 3.6123,
+      "step": 16585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004905829589281642,
+      "loss": 3.5348,
+      "step": 16590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004905194197185536,
+      "loss": 3.5266,
+      "step": 16595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004904558661828589,
+      "loss": 3.5143,
+      "step": 16600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004903922983258593,
+      "loss": 3.6398,
+      "step": 16605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004903287161523346,
+      "loss": 3.5811,
+      "step": 16610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004902651196670658,
+      "loss": 3.4796,
+      "step": 16615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004902015088748352,
+      "loss": 3.4956,
+      "step": 16620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004901378837804258,
+      "loss": 3.4245,
+      "step": 16625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000490074244388622,
+      "loss": 3.6016,
+      "step": 16630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004900105907042092,
+      "loss": 3.5701,
+      "step": 16635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004899469227319737,
+      "loss": 3.5832,
+      "step": 16640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004898832404767031,
+      "loss": 3.5255,
+      "step": 16645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004898195439431858,
+      "loss": 3.5867,
+      "step": 16650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004897558331362119,
+      "loss": 3.4916,
+      "step": 16655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004896921080605716,
+      "loss": 3.5749,
+      "step": 16660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000489628368721057,
+      "loss": 3.5705,
+      "step": 16665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000489564615122461,
+      "loss": 3.5197,
+      "step": 16670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004895008472695772,
+      "loss": 3.669,
+      "step": 16675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004894370651672011,
+      "loss": 3.5425,
+      "step": 16680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004893732688201284,
+      "loss": 3.5529,
+      "step": 16685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004893094582331566,
+      "loss": 3.5687,
+      "step": 16690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004892456334110835,
+      "loss": 3.5815,
+      "step": 16695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004891817943587087,
+      "loss": 3.5539,
+      "step": 16700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004891179410808326,
+      "loss": 3.6557,
+      "step": 16705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004890540735822566,
+      "loss": 3.6539,
+      "step": 16710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000488990191867783,
+      "loss": 3.675,
+      "step": 16715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004889262959422155,
+      "loss": 3.5966,
+      "step": 16720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004888623858103589,
+      "loss": 3.4819,
+      "step": 16725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004887984614770188,
+      "loss": 3.4654,
+      "step": 16730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004887345229470019,
+      "loss": 3.5918,
+      "step": 16735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004886705702251161,
+      "loss": 3.5793,
+      "step": 16740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004886066033161704,
+      "loss": 3.5092,
+      "step": 16745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004885426222249747,
+      "loss": 3.6574,
+      "step": 16750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004884786269563401,
+      "loss": 3.5397,
+      "step": 16755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004884146175150786,
+      "loss": 3.4925,
+      "step": 16760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004883505939060036,
+      "loss": 3.5832,
+      "step": 16765
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004882865561339293,
+      "loss": 3.5626,
+      "step": 16770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048822250420367064,
+      "loss": 3.5953,
+      "step": 16775
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048815843812004446,
+      "loss": 3.6348,
+      "step": 16780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000488094357887868,
+      "loss": 3.5284,
+      "step": 16785
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048803026351195984,
+      "loss": 3.5496,
+      "step": 16790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048796615499713947,
+      "loss": 3.554,
+      "step": 16795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004879020323482276,
+      "loss": 3.5148,
+      "step": 16800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004878378955700458,
+      "loss": 3.5522,
+      "step": 16805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048777374466741697,
+      "loss": 3.5446,
+      "step": 16810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004877095796451648,
+      "loss": 3.5773,
+      "step": 16815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004876454005081144,
+      "loss": 3.6213,
+      "step": 16820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004875812072610913,
+      "loss": 3.621,
+      "step": 16825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004875169999089229,
+      "loss": 3.524,
+      "step": 16830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004874527784564371,
+      "loss": 3.5176,
+      "step": 16835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004873885429084631,
+      "loss": 3.4054,
+      "step": 16840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048732429326983097,
+      "loss": 3.5634,
+      "step": 16845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000487260029545372,
+      "loss": 3.6623,
+      "step": 16850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004871957517399184,
+      "loss": 3.5978,
+      "step": 16855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048713145985830386,
+      "loss": 3.5235,
+      "step": 16860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048706715390536235,
+      "loss": 3.5122,
+      "step": 16865
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048700283388592976,
+      "loss": 3.5124,
+      "step": 16870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048693849980484234,
+      "loss": 3.5231,
+      "step": 16875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048687415166693773,
+      "loss": 3.6416,
+      "step": 16880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004868097894770547,
+      "loss": 3.5731,
+      "step": 16885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048674541324003287,
+      "loss": 3.5577,
+      "step": 16890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048668102296071296,
+      "loss": 3.3237,
+      "step": 16895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048661661864393696,
+      "loss": 3.5944,
+      "step": 16900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048655220029454756,
+      "loss": 3.5177,
+      "step": 16905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004864877679173889,
+      "loss": 3.511,
+      "step": 16910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004864233215173056,
+      "loss": 3.5407,
+      "step": 16915
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004863588610991441,
+      "loss": 3.5129,
+      "step": 16920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004862943866677512,
+      "loss": 3.5516,
+      "step": 16925
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004862298982279752,
+      "loss": 3.6088,
+      "step": 16930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004861653957846653,
+      "loss": 3.4369,
+      "step": 16935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004861008793426717,
+      "loss": 3.4892,
+      "step": 16940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004860363489068457,
+      "loss": 3.5787,
+      "step": 16945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048597180448203977,
+      "loss": 3.5381,
+      "step": 16950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048590724607310714,
+      "loss": 3.4371,
+      "step": 16955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004858426736849024,
+      "loss": 3.4722,
+      "step": 16960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048577808732228105,
+      "loss": 3.5033,
+      "step": 16965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048571348699009964,
+      "loss": 3.5394,
+      "step": 16970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004856488726932158,
+      "loss": 3.5476,
+      "step": 16975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048558424443648804,
+      "loss": 3.5903,
+      "step": 16980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048551960222477635,
+      "loss": 3.4827,
+      "step": 16985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004854549460629412,
+      "loss": 3.4285,
+      "step": 16990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048539027595584464,
+      "loss": 3.5585,
+      "step": 16995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004853255919083494,
+      "loss": 3.4662,
+      "step": 17000
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 3.576127052307129,
+      "eval_runtime": 150.0801,
+      "eval_samples_per_second": 12.267,
+      "eval_steps_per_second": 0.773,
+      "step": 17000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048526089392531945,
+      "loss": 3.4641,
+      "step": 17005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004851961820116197,
+      "loss": 3.5871,
+      "step": 17010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004851314561721162,
+      "loss": 3.6005,
+      "step": 17015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000485066716411676,
+      "loss": 3.5406,
+      "step": 17020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004850019627351672,
+      "loss": 3.5104,
+      "step": 17025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004849371951474588,
+      "loss": 3.5767,
+      "step": 17030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048487241365342116,
+      "loss": 3.5849,
+      "step": 17035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00048480761825792556,
+      "loss": 3.5134,
+      "step": 17040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004847428089658442,
+      "loss": 3.5021,
+      "step": 17045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004846779857820503,
+      "loss": 3.5575,
+      "step": 17050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048461314871141835,
+      "loss": 3.5833,
+      "step": 17055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048454829775882374,
+      "loss": 3.4563,
+      "step": 17060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048448343292914296,
+      "loss": 3.5297,
+      "step": 17065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004844185542272534,
+      "loss": 3.576,
+      "step": 17070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048435366165803384,
+      "loss": 3.5006,
+      "step": 17075
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004842887552263637,
+      "loss": 3.579,
+      "step": 17080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004842238349371235,
+      "loss": 3.565,
+      "step": 17085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048415890079519516,
+      "loss": 3.4344,
+      "step": 17090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004840939528054612,
+      "loss": 3.4845,
+      "step": 17095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048402899097280556,
+      "loss": 3.5118,
+      "step": 17100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004839640153021129,
+      "loss": 3.4574,
+      "step": 17105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000483899025798269,
+      "loss": 3.5173,
+      "step": 17110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048383402246616087,
+      "loss": 3.5195,
+      "step": 17115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004837690053106765,
+      "loss": 3.4535,
+      "step": 17120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004837039743367045,
+      "loss": 3.4877,
+      "step": 17125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004836389295491353,
+      "loss": 3.5545,
+      "step": 17130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048357387095285964,
+      "loss": 3.4785,
+      "step": 17135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004835087985527697,
+      "loss": 3.5197,
+      "step": 17140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004834437123537586,
+      "loss": 3.5171,
+      "step": 17145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004833786123607205,
+      "loss": 3.5258,
+      "step": 17150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004833134985785505,
+      "loss": 3.6473,
+      "step": 17155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004832483710121448,
+      "loss": 3.5945,
+      "step": 17160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004831832296664009,
+      "loss": 3.577,
+      "step": 17165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048311807454621685,
+      "loss": 3.5094,
+      "step": 17170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048305290565649195,
+      "loss": 3.489,
+      "step": 17175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004829877230021268,
+      "loss": 3.524,
+      "step": 17180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048292252658802264,
+      "loss": 3.5145,
+      "step": 17185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048285731641908196,
+      "loss": 3.5293,
+      "step": 17190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004827920925002082,
+      "loss": 3.4996,
+      "step": 17195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048272685483630585,
+      "loss": 3.6237,
+      "step": 17200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048266160343228066,
+      "loss": 3.4632,
+      "step": 17205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004825963382930388,
+      "loss": 3.5058,
+      "step": 17210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004825310594234882,
+      "loss": 3.521,
+      "step": 17215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048246576682853736,
+      "loss": 3.4941,
+      "step": 17220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048240046051309593,
+      "loss": 3.6331,
+      "step": 17225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004823351404820747,
+      "loss": 3.5374,
+      "step": 17230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048226980674038534,
+      "loss": 3.5544,
+      "step": 17235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048220445929294067,
+      "loss": 3.4389,
+      "step": 17240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048213909814465443,
+      "loss": 3.5835,
+      "step": 17245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048207372330044137,
+      "loss": 3.4002,
+      "step": 17250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004820083347652176,
+      "loss": 3.5722,
+      "step": 17255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004819429325438997,
+      "loss": 3.6024,
+      "step": 17260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004818775166414057,
+      "loss": 3.6439,
+      "step": 17265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004818120870626547,
+      "loss": 3.4983,
+      "step": 17270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048174664381256643,
+      "loss": 3.4879,
+      "step": 17275
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048168118689606197,
+      "loss": 3.5338,
+      "step": 17280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048161571631806337,
+      "loss": 3.4761,
+      "step": 17285
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004815502320834936,
+      "loss": 3.5412,
+      "step": 17290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048148473419727694,
+      "loss": 3.5941,
+      "step": 17295
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004814192226643383,
+      "loss": 3.5358,
+      "step": 17300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048135369748960395,
+      "loss": 3.459,
+      "step": 17305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004812881586780008,
+      "loss": 3.4669,
+      "step": 17310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004812226062344573,
+      "loss": 3.5167,
+      "step": 17315
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048115704016390265,
+      "loss": 3.6433,
+      "step": 17320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004810914604712669,
+      "loss": 3.4723,
+      "step": 17325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048102586716148145,
+      "loss": 3.6129,
+      "step": 17330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048096026023947846,
+      "loss": 3.5888,
+      "step": 17335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048089463971019133,
+      "loss": 3.5604,
+      "step": 17340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004808290055785544,
+      "loss": 3.5614,
+      "step": 17345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004807633578495029,
+      "loss": 3.5267,
+      "step": 17350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004806976965279734,
+      "loss": 3.5437,
+      "step": 17355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004806320216189031,
+      "loss": 3.577,
+      "step": 17360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004805663331272306,
+      "loss": 3.5452,
+      "step": 17365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048050063105789515,
+      "loss": 3.5119,
+      "step": 17370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004804349154158373,
+      "loss": 3.5662,
+      "step": 17375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004803691862059985,
+      "loss": 3.5817,
+      "step": 17380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004803034434333214,
+      "loss": 3.4176,
+      "step": 17385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004802376871027493,
+      "loss": 3.5071,
+      "step": 17390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048017191721922683,
+      "loss": 3.4767,
+      "step": 17395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048010613378769945,
+      "loss": 3.4762,
+      "step": 17400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00048004033681311407,
+      "loss": 3.5571,
+      "step": 17405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047997452630041783,
+      "loss": 3.4705,
+      "step": 17410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047990870225455976,
+      "loss": 3.4789,
+      "step": 17415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047984286468048923,
+      "loss": 3.5571,
+      "step": 17420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047977701358315687,
+      "loss": 3.5161,
+      "step": 17425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047971114896751455,
+      "loss": 3.6078,
+      "step": 17430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047964527083851476,
+      "loss": 3.4624,
+      "step": 17435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047957937920111126,
+      "loss": 3.4843,
+      "step": 17440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047951347406025886,
+      "loss": 3.5239,
+      "step": 17445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047944755542091307,
+      "loss": 3.5355,
+      "step": 17450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047938162328803086,
+      "loss": 3.427,
+      "step": 17455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004793156776665699,
+      "loss": 3.5715,
+      "step": 17460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047924971856148896,
+      "loss": 3.6401,
+      "step": 17465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004791837459777478,
+      "loss": 3.5878,
+      "step": 17470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004791177599203072,
+      "loss": 3.4139,
+      "step": 17475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047905176039412904,
+      "loss": 3.5894,
+      "step": 17480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004789857474041762,
+      "loss": 3.5054,
+      "step": 17485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004789197209554124,
+      "loss": 3.4073,
+      "step": 17490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004788536810528025,
+      "loss": 3.546,
+      "step": 17495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004787876277013124,
+      "loss": 3.4948,
+      "step": 17500
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 3.5631024837493896,
+      "eval_runtime": 149.9514,
+      "eval_samples_per_second": 12.277,
+      "eval_steps_per_second": 0.774,
+      "step": 17500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000478721560905909,
+      "loss": 3.5123,
+      "step": 17505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004786554806715601,
+      "loss": 3.4436,
+      "step": 17510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004785893870032347,
+      "loss": 3.551,
+      "step": 17515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004785232799059026,
+      "loss": 3.5769,
+      "step": 17520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004784571593845349,
+      "loss": 3.5486,
+      "step": 17525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047839102544410325,
+      "loss": 3.6552,
+      "step": 17530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047832487808958083,
+      "loss": 3.5692,
+      "step": 17535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004782587173259414,
+      "loss": 3.4071,
+      "step": 17540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047819254315816006,
+      "loss": 3.5011,
+      "step": 17545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047812635559121256,
+      "loss": 3.5682,
+      "step": 17550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047806015463007615,
+      "loss": 3.4182,
+      "step": 17555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047799394027972866,
+      "loss": 3.5072,
+      "step": 17560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004779277125451489,
+      "loss": 3.4896,
+      "step": 17565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047786147143131716,
+      "loss": 3.3961,
+      "step": 17570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047779521694321425,
+      "loss": 3.5795,
+      "step": 17575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004777289490858222,
+      "loss": 3.4299,
+      "step": 17580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047766266786412415,
+      "loss": 3.5548,
+      "step": 17585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047759637328310386,
+      "loss": 3.4617,
+      "step": 17590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047753006534774646,
+      "loss": 3.4899,
+      "step": 17595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000477463744063038,
+      "loss": 3.5963,
+      "step": 17600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047739740943396553,
+      "loss": 3.4351,
+      "step": 17605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004773310614655169,
+      "loss": 3.5722,
+      "step": 17610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047726470016268134,
+      "loss": 3.5019,
+      "step": 17615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00047719832553044876,
+      "loss": 3.4795,
+      "step": 17620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004771319375738103,
+      "loss": 3.4939,
+      "step": 17625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004770655362977578,
+      "loss": 3.459,
+      "step": 17630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004769991217072844,
+      "loss": 3.4321,
+      "step": 17635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004769326938073843,
+      "loss": 3.588,
+      "step": 17640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004768662526030523,
+      "loss": 3.6168,
+      "step": 17645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004767997980992846,
+      "loss": 3.4792,
+      "step": 17650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004767333303010781,
+      "loss": 3.4928,
+      "step": 17655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004766668492134309,
+      "loss": 3.6081,
+      "step": 17660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004766003548413421,
+      "loss": 3.5809,
+      "step": 17665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047653384718981164,
+      "loss": 3.5654,
+      "step": 17670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047646732626384066,
+      "loss": 3.4998,
+      "step": 17675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000476400792068431,
+      "loss": 3.3947,
+      "step": 17680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047633424460858585,
+      "loss": 3.6234,
+      "step": 17685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004762676838893093,
+      "loss": 3.4851,
+      "step": 17690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047620110991560623,
+      "loss": 3.4785,
+      "step": 17695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047613452269248274,
+      "loss": 3.5703,
+      "step": 17700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047606792222494566,
+      "loss": 3.5448,
+      "step": 17705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004760013085180033,
+      "loss": 3.4537,
+      "step": 17710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004759346815766646,
+      "loss": 3.515,
+      "step": 17715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004758680414059393,
+      "loss": 3.5411,
+      "step": 17720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004758013880108387,
+      "loss": 3.5154,
+      "step": 17725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004757347213963747,
+      "loss": 3.4914,
+      "step": 17730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004756680415675603,
+      "loss": 3.4521,
+      "step": 17735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004756013485294094,
+      "loss": 3.5936,
+      "step": 17740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004755346422869371,
+      "loss": 3.4493,
+      "step": 17745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004754679228451592,
+      "loss": 3.5317,
+      "step": 17750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047540119020909286,
+      "loss": 3.5307,
+      "step": 17755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047533444438375587,
+      "loss": 3.5228,
+      "step": 17760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004752676853741673,
+      "loss": 3.4958,
+      "step": 17765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004752009131853469,
+      "loss": 3.4806,
+      "step": 17770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004751341278223159,
+      "loss": 3.4653,
+      "step": 17775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004750673292900959,
+      "loss": 3.5797,
+      "step": 17780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047500051759371,
+      "loss": 3.5707,
+      "step": 17785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000474933692738182,
+      "loss": 3.5677,
+      "step": 17790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047486685472853687,
+      "loss": 3.5105,
+      "step": 17795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004748000035698003,
+      "loss": 3.5625,
+      "step": 17800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047473313926699954,
+      "loss": 3.5567,
+      "step": 17805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047466626182516203,
+      "loss": 3.5687,
+      "step": 17810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047459937124931685,
+      "loss": 3.5199,
+      "step": 17815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004745324675444938,
+      "loss": 3.5174,
+      "step": 17820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047446555071572354,
+      "loss": 3.4744,
+      "step": 17825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004743986207680381,
+      "loss": 3.3464,
+      "step": 17830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047433167770647,
+      "loss": 3.55,
+      "step": 17835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004742647215360533,
+      "loss": 3.4027,
+      "step": 17840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047419775226182264,
+      "loss": 3.5334,
+      "step": 17845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004741307698888137,
+      "loss": 3.5083,
+      "step": 17850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047406377442206325,
+      "loss": 3.628,
+      "step": 17855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047399676586660914,
+      "loss": 3.6128,
+      "step": 17860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004739297442274899,
+      "loss": 3.5811,
+      "step": 17865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047386270950974525,
+      "loss": 3.5103,
+      "step": 17870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004737956617184159,
+      "loss": 3.5447,
+      "step": 17875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047372860085854345,
+      "loss": 3.6269,
+      "step": 17880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004736615269351706,
+      "loss": 3.5275,
+      "step": 17885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047359443995334083,
+      "loss": 3.4888,
+      "step": 17890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047352733991809894,
+      "loss": 3.5246,
+      "step": 17895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004734602268344903,
+      "loss": 3.4484,
+      "step": 17900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004733931007075617,
+      "loss": 3.5838,
+      "step": 17905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047332596154236046,
+      "loss": 3.5475,
+      "step": 17910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047325880934393524,
+      "loss": 3.5593,
+      "step": 17915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047319164411733545,
+      "loss": 3.6149,
+      "step": 17920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047312446586761164,
+      "loss": 3.4426,
+      "step": 17925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004730572745998153,
+      "loss": 3.524,
+      "step": 17930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004729900703189988,
+      "loss": 3.5848,
+      "step": 17935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047292285303021544,
+      "loss": 3.4916,
+      "step": 17940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004728556227385199,
+      "loss": 3.5414,
+      "step": 17945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004727883794489673,
+      "loss": 3.5673,
+      "step": 17950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004727211231666141,
+      "loss": 3.5469,
+      "step": 17955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004726538538965177,
+      "loss": 3.4563,
+      "step": 17960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047258657164373624,
+      "loss": 3.512,
+      "step": 17965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047251927641332915,
+      "loss": 3.481,
+      "step": 17970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004724519682103566,
+      "loss": 3.5488,
+      "step": 17975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004723846470398798,
+      "loss": 3.3891,
+      "step": 17980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000472317312906961,
+      "loss": 3.533,
+      "step": 17985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004722499658166634,
+      "loss": 3.6264,
+      "step": 17990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047218260577405123,
+      "loss": 3.5065,
+      "step": 17995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004721152327841894,
+      "loss": 3.5378,
+      "step": 18000
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 3.560037136077881,
+      "eval_runtime": 149.8718,
+      "eval_samples_per_second": 12.284,
+      "eval_steps_per_second": 0.774,
+      "step": 18000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047204784685214425,
+      "loss": 3.637,
+      "step": 18005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047198044798298273,
+      "loss": 3.5406,
+      "step": 18010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004719130361817728,
+      "loss": 3.5347,
+      "step": 18015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047184561145358376,
+      "loss": 3.5563,
+      "step": 18020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004717781738034853,
+      "loss": 3.6075,
+      "step": 18025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047171072323654847,
+      "loss": 3.411,
+      "step": 18030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004716432597578454,
+      "loss": 3.505,
+      "step": 18035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047157578337244873,
+      "loss": 3.6218,
+      "step": 18040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004715082940854326,
+      "loss": 3.4998,
+      "step": 18045
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004714407919018716,
+      "loss": 3.5515,
+      "step": 18050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004713732768268417,
+      "loss": 3.6125,
+      "step": 18055
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004713057488654197,
+      "loss": 3.5094,
+      "step": 18060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004712382080226833,
+      "loss": 3.5357,
+      "step": 18065
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004711706543037112,
+      "loss": 3.5171,
+      "step": 18070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004711030877135831,
+      "loss": 3.5248,
+      "step": 18075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004710355082573798,
+      "loss": 3.3577,
+      "step": 18080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047096791594018273,
+      "loss": 3.4937,
+      "step": 18085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004709003107670747,
+      "loss": 3.5847,
+      "step": 18090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047083269274313897,
+      "loss": 3.5041,
+      "step": 18095
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004707650618734604,
+      "loss": 3.518,
+      "step": 18100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004706974181631243,
+      "loss": 3.3611,
+      "step": 18105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004706297616172172,
+      "loss": 3.5533,
+      "step": 18110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00047056209224082636,
+      "loss": 3.4176,
+      "step": 18115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004704944100390404,
+      "loss": 3.5398,
+      "step": 18120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004704267150169485,
+      "loss": 3.4949,
+      "step": 18125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004703590071796411,
+      "loss": 3.5204,
+      "step": 18130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004702912865322095,
+      "loss": 3.4492,
+      "step": 18135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004702235530797458,
+      "loss": 3.6095,
+      "step": 18140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004701558068273433,
+      "loss": 3.4533,
+      "step": 18145
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004700880477800962,
+      "loss": 3.4747,
+      "step": 18150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004700202759430995,
+      "loss": 3.6068,
+      "step": 18155
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004699524913214495,
+      "loss": 3.5339,
+      "step": 18160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004698846939202431,
+      "loss": 3.5689,
+      "step": 18165
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00046981688374457835,
+      "loss": 3.4747,
+      "step": 18170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004697490607995543,
+      "loss": 3.5512,
+      "step": 18175
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00046968122509027075,
+      "loss": 3.5927,
+      "step": 18180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00046961337662182874,
+      "loss": 3.4706,
+      "step": 18185
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00046954551539933007,
+      "loss": 3.556,
+      "step": 18190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0004694776414278775,
+      "loss": 3.4009,
+      "step": 18195
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00046940975471257483,
+      "loss": 3.4902,
+      "step": 18200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046934185525852683,
+      "loss": 3.4386,
+      "step": 18205
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004692739430708392,
+      "loss": 3.4121,
+      "step": 18210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004692060181546186,
+      "loss": 3.5268,
+      "step": 18215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046913808051497253,
+      "loss": 3.5333,
+      "step": 18220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046907013015700977,
+      "loss": 3.5135,
+      "step": 18225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004690021670858395,
+      "loss": 3.4469,
+      "step": 18230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004689341913065725,
+      "loss": 3.4252,
+      "step": 18235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004688662028243201,
+      "loss": 3.5202,
+      "step": 18240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046879820164419454,
+      "loss": 3.596,
+      "step": 18245
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046873018777130945,
+      "loss": 3.5158,
+      "step": 18250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004686621612107789,
+      "loss": 3.4576,
+      "step": 18255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046859412196771817,
+      "loss": 3.4635,
+      "step": 18260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046852607004724354,
+      "loss": 3.5205,
+      "step": 18265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004684580054544721,
+      "loss": 3.5507,
+      "step": 18270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004683899281945221,
+      "loss": 3.4769,
+      "step": 18275
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046832183827251244,
+      "loss": 3.4863,
+      "step": 18280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046825373569356305,
+      "loss": 3.5763,
+      "step": 18285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004681856204627951,
+      "loss": 3.4605,
+      "step": 18290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004681174925853305,
+      "loss": 3.4173,
+      "step": 18295
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046804935206629187,
+      "loss": 3.4477,
+      "step": 18300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046798119891080343,
+      "loss": 3.4775,
+      "step": 18305
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004679130331239895,
+      "loss": 3.5202,
+      "step": 18310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004678448547109761,
+      "loss": 3.4684,
+      "step": 18315
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046777666367688996,
+      "loss": 3.4995,
+      "step": 18320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004677084600268584,
+      "loss": 3.4431,
+      "step": 18325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004676402437660102,
+      "loss": 3.5776,
+      "step": 18330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004675720148994747,
+      "loss": 3.5402,
+      "step": 18335
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046750377343238256,
+      "loss": 3.5223,
+      "step": 18340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046743551936986513,
+      "loss": 3.332,
+      "step": 18345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004673672527170547,
+      "loss": 3.5664,
+      "step": 18350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046729897347908467,
+      "loss": 3.5212,
+      "step": 18355
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004672306816610892,
+      "loss": 3.4357,
+      "step": 18360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004671623772682034,
+      "loss": 3.4875,
+      "step": 18365
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046709406030556364,
+      "loss": 3.6394,
+      "step": 18370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046702573077830677,
+      "loss": 3.5538,
+      "step": 18375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004669573886915711,
+      "loss": 3.4598,
+      "step": 18380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004668890340504954,
+      "loss": 3.6014,
+      "step": 18385
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004668206668602196,
+      "loss": 3.5697,
+      "step": 18390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004667522871258846,
+      "loss": 3.3325,
+      "step": 18395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004666838948526321,
+      "loss": 3.5063,
+      "step": 18400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046661549004560515,
+      "loss": 3.5601,
+      "step": 18405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004665470727099471,
+      "loss": 3.5294,
+      "step": 18410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046647864285080274,
+      "loss": 3.4877,
+      "step": 18415
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046641020047331764,
+      "loss": 3.5173,
+      "step": 18420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004663417455826383,
+      "loss": 3.5529,
+      "step": 18425
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004662732781839122,
+      "loss": 3.4485,
+      "step": 18430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046620479828228767,
+      "loss": 3.6021,
+      "step": 18435
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046613630588291413,
+      "loss": 3.5351,
+      "step": 18440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004660678009909419,
+      "loss": 3.5565,
+      "step": 18445
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000465999283611522,
+      "loss": 3.5563,
+      "step": 18450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004659307537498068,
+      "loss": 3.4886,
+      "step": 18455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046586221141094923,
+      "loss": 3.5287,
+      "step": 18460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004657936566001034,
+      "loss": 3.3976,
+      "step": 18465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046572508932242424,
+      "loss": 3.5678,
+      "step": 18470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046565650958306775,
+      "loss": 3.5914,
+      "step": 18475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004655879173871908,
+      "loss": 3.4357,
+      "step": 18480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046551931273995087,
+      "loss": 3.5462,
+      "step": 18485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000465450695646507,
+      "loss": 3.6432,
+      "step": 18490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004653820661120188,
+      "loss": 3.509,
+      "step": 18495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004653134241416467,
+      "loss": 3.3627,
+      "step": 18500
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 3.539888858795166,
+      "eval_runtime": 149.9774,
+      "eval_samples_per_second": 12.275,
+      "eval_steps_per_second": 0.773,
+      "step": 18500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004652447697405524,
+      "loss": 3.4926,
+      "step": 18505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046517610291389825,
+      "loss": 3.4943,
+      "step": 18510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004651074236668477,
+      "loss": 3.5992,
+      "step": 18515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000465038732004565,
+      "loss": 3.4454,
+      "step": 18520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046497002793221545,
+      "loss": 3.4608,
+      "step": 18525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004649013114549653,
+      "loss": 3.426,
+      "step": 18530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004648325825779817,
+      "loss": 3.5191,
+      "step": 18535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046476384130643254,
+      "loss": 3.6492,
+      "step": 18540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000464695087645487,
+      "loss": 3.4617,
+      "step": 18545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046462632160031484,
+      "loss": 3.506,
+      "step": 18550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000464557543176087,
+      "loss": 3.5133,
+      "step": 18555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004644887523779753,
+      "loss": 3.5006,
+      "step": 18560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004644199492111524,
+      "loss": 3.2805,
+      "step": 18565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046435113368079197,
+      "loss": 3.5573,
+      "step": 18570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046428230579206853,
+      "loss": 3.5132,
+      "step": 18575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004642134655501576,
+      "loss": 3.661,
+      "step": 18580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004641446129602358,
+      "loss": 3.3854,
+      "step": 18585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046407574802748017,
+      "loss": 3.5102,
+      "step": 18590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046400687075706925,
+      "loss": 3.4589,
+      "step": 18595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004639379811541821,
+      "loss": 3.4866,
+      "step": 18600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046386907922399894,
+      "loss": 3.5285,
+      "step": 18605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046380016497170096,
+      "loss": 3.5076,
+      "step": 18610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004637312384024699,
+      "loss": 3.4932,
+      "step": 18615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004636622995214889,
+      "loss": 3.4528,
+      "step": 18620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004635933483339417,
+      "loss": 3.586,
+      "step": 18625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004635243848450131,
+      "loss": 3.4847,
+      "step": 18630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004634554090598888,
+      "loss": 3.525,
+      "step": 18635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046338642098375544,
+      "loss": 3.4765,
+      "step": 18640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004633174206218006,
+      "loss": 3.577,
+      "step": 18645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004632484079792127,
+      "loss": 3.4143,
+      "step": 18650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046317938306118117,
+      "loss": 3.5152,
+      "step": 18655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004631103458728963,
+      "loss": 3.4873,
+      "step": 18660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004630412964195493,
+      "loss": 3.4814,
+      "step": 18665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046297223470633247,
+      "loss": 3.5824,
+      "step": 18670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004629031607384388,
+      "loss": 3.4629,
+      "step": 18675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046283407452106225,
+      "loss": 3.5677,
+      "step": 18680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046276497605939786,
+      "loss": 3.4402,
+      "step": 18685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004626958653586414,
+      "loss": 3.4675,
+      "step": 18690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004626267424239896,
+      "loss": 3.4872,
+      "step": 18695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004625576072606403,
+      "loss": 3.4781,
+      "step": 18700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046248845987379204,
+      "loss": 3.5359,
+      "step": 18705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046241930026864425,
+      "loss": 3.5084,
+      "step": 18710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004623501284503976,
+      "loss": 3.4936,
+      "step": 18715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004622809444242533,
+      "loss": 3.5277,
+      "step": 18720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004622117481954135,
+      "loss": 3.5329,
+      "step": 18725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046214253976908167,
+      "loss": 3.459,
+      "step": 18730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046207331915046183,
+      "loss": 3.5331,
+      "step": 18735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046200408634475894,
+      "loss": 3.4192,
+      "step": 18740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046193484135717913,
+      "loss": 3.4791,
+      "step": 18745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004618655841929291,
+      "loss": 3.4814,
+      "step": 18750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004617963148572167,
+      "loss": 3.4538,
+      "step": 18755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00046172703335525047,
+      "loss": 3.4978,
+      "step": 18760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004616577396922403,
+      "loss": 3.5329,
+      "step": 18765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004615884338733966,
+      "loss": 3.5177,
+      "step": 18770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0004615191159039308,
+      "loss": 3.5178,
+      "step": 18775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004614497857890552,
+      "loss": 3.5395,
+      "step": 18780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004613804435339832,
+      "loss": 3.5156,
+      "step": 18785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00046131108914392884,
+      "loss": 3.4949,
+      "step": 18790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004612417226241073,
+      "loss": 3.4067,
+      "step": 18795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004611723439797346,
+      "loss": 3.4388,
+      "step": 18800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004611029532160276,
+      "loss": 3.5248,
+      "step": 18805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004610335503382041,
+      "loss": 3.5521,
+      "step": 18810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004609641353514829,
+      "loss": 3.5943,
+      "step": 18815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004608947082610838,
+      "loss": 3.5184,
+      "step": 18820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004608252690722271,
+      "loss": 3.4862,
+      "step": 18825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004607558177901344,
+      "loss": 3.6086,
+      "step": 18830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000460686354420028,
+      "loss": 3.4269,
+      "step": 18835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004606168789671314,
+      "loss": 3.583,
+      "step": 18840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00046054739143666857,
+      "loss": 3.4697,
+      "step": 18845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004604778918338647,
+      "loss": 3.5615,
+      "step": 18850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004604083801639457,
+      "loss": 3.5213,
+      "step": 18855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00046033885643213866,
+      "loss": 3.443,
+      "step": 18860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004602693206436714,
+      "loss": 3.4458,
+      "step": 18865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004601997728037726,
+      "loss": 3.4955,
+      "step": 18870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00046013021291767183,
+      "loss": 3.5146,
+      "step": 18875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004600606409905997,
+      "loss": 3.5533,
+      "step": 18880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045999105702778775,
+      "loss": 3.415,
+      "step": 18885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004599214610344682,
+      "loss": 3.4769,
+      "step": 18890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045985185301587435,
+      "loss": 3.4555,
+      "step": 18895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004597822329772403,
+      "loss": 3.4481,
+      "step": 18900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045971260092380137,
+      "loss": 3.391,
+      "step": 18905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045964295686079325,
+      "loss": 3.3999,
+      "step": 18910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000459573300793453,
+      "loss": 3.5864,
+      "step": 18915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045950363272701824,
+      "loss": 3.5394,
+      "step": 18920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045943395266672787,
+      "loss": 3.5128,
+      "step": 18925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045936426061782135,
+      "loss": 3.5267,
+      "step": 18930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045929455658553895,
+      "loss": 3.4345,
+      "step": 18935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045922484057512247,
+      "loss": 3.6213,
+      "step": 18940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045915511259181385,
+      "loss": 3.5013,
+      "step": 18945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045908537264085657,
+      "loss": 3.561,
+      "step": 18950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004590156207274945,
+      "loss": 3.5136,
+      "step": 18955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045894585685697263,
+      "loss": 3.4877,
+      "step": 18960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000458876081034537,
+      "loss": 3.4736,
+      "step": 18965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004588062932654343,
+      "loss": 3.5516,
+      "step": 18970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045873649355491217,
+      "loss": 3.4,
+      "step": 18975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004586666819082192,
+      "loss": 3.5071,
+      "step": 18980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004585968583306049,
+      "loss": 3.5604,
+      "step": 18985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045852702282731976,
+      "loss": 3.5457,
+      "step": 18990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004584571754036149,
+      "loss": 3.4559,
+      "step": 18995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004583873160647425,
+      "loss": 3.6102,
+      "step": 19000
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 3.5276401042938232,
+      "eval_runtime": 149.8717,
+      "eval_samples_per_second": 12.284,
+      "eval_steps_per_second": 0.774,
+      "step": 19000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004583174448159557,
+      "loss": 3.4795,
+      "step": 19005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004582475616625084,
+      "loss": 3.4671,
+      "step": 19010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045817766660965543,
+      "loss": 3.5596,
+      "step": 19015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004581077596626526,
+      "loss": 3.6208,
+      "step": 19020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004580378408267565,
+      "loss": 3.5096,
+      "step": 19025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045796791010722487,
+      "loss": 3.4556,
+      "step": 19030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045789796750931574,
+      "loss": 3.4603,
+      "step": 19035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045782801303828873,
+      "loss": 3.4485,
+      "step": 19040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004577580466994041,
+      "loss": 3.4624,
+      "step": 19045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045768806849792276,
+      "loss": 3.4523,
+      "step": 19050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045761807843910685,
+      "loss": 3.5471,
+      "step": 19055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004575480765282192,
+      "loss": 3.3223,
+      "step": 19060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045747806277052363,
+      "loss": 3.4825,
+      "step": 19065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045740803717128484,
+      "loss": 3.3537,
+      "step": 19070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004573379997357682,
+      "loss": 3.3989,
+      "step": 19075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004572679504692405,
+      "loss": 3.5855,
+      "step": 19080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045719788937696883,
+      "loss": 3.5294,
+      "step": 19085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004571278164642215,
+      "loss": 3.4447,
+      "step": 19090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004570577317362677,
+      "loss": 3.6258,
+      "step": 19095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045698763519837725,
+      "loss": 3.4541,
+      "step": 19100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045691752685582133,
+      "loss": 3.5693,
+      "step": 19105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004568474067138716,
+      "loss": 3.4704,
+      "step": 19110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045677727477780056,
+      "loss": 3.5682,
+      "step": 19115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000456707131052882,
+      "loss": 3.4419,
+      "step": 19120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045663697554439034,
+      "loss": 3.4335,
+      "step": 19125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004565668082576008,
+      "loss": 3.5477,
+      "step": 19130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045649662919778976,
+      "loss": 3.5055,
+      "step": 19135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045642643837023406,
+      "loss": 3.6236,
+      "step": 19140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004563562357802121,
+      "loss": 3.5487,
+      "step": 19145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045628602143300243,
+      "loss": 3.4841,
+      "step": 19150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004562157953338848,
+      "loss": 3.4994,
+      "step": 19155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045614555748814005,
+      "loss": 3.4607,
+      "step": 19160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045607530790104956,
+      "loss": 3.4797,
+      "step": 19165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045600504657789594,
+      "loss": 3.5536,
+      "step": 19170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045593477352396213,
+      "loss": 3.5152,
+      "step": 19175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004558644887445326,
+      "loss": 3.5431,
+      "step": 19180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004557941922448923,
+      "loss": 3.4514,
+      "step": 19185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045572388403032724,
+      "loss": 3.3985,
+      "step": 19190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004556535641061241,
+      "loss": 3.4395,
+      "step": 19195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045558323247757063,
+      "loss": 3.4796,
+      "step": 19200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045551288914995536,
+      "loss": 3.4244,
+      "step": 19205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045544253412856806,
+      "loss": 3.4544,
+      "step": 19210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045537216741869865,
+      "loss": 3.5251,
+      "step": 19215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045530178902563847,
+      "loss": 3.5224,
+      "step": 19220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004552313989546798,
+      "loss": 3.2957,
+      "step": 19225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004551609972111154,
+      "loss": 3.4405,
+      "step": 19230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004550905838002392,
+      "loss": 3.4844,
+      "step": 19235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000455020158727346,
+      "loss": 3.549,
+      "step": 19240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004549497219977312,
+      "loss": 3.5311,
+      "step": 19245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004548792736166915,
+      "loss": 3.4819,
+      "step": 19250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045480881358952416,
+      "loss": 3.492,
+      "step": 19255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045473834192152744,
+      "loss": 3.4925,
+      "step": 19260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045466785861800034,
+      "loss": 3.4899,
+      "step": 19265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045459736368424304,
+      "loss": 3.5294,
+      "step": 19270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004545268571255563,
+      "loss": 3.4631,
+      "step": 19275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004544563389472417,
+      "loss": 3.436,
+      "step": 19280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004543858091546021,
+      "loss": 3.4474,
+      "step": 19285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045431526775294086,
+      "loss": 3.4655,
+      "step": 19290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045424471474756227,
+      "loss": 3.4941,
+      "step": 19295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045417415014377166,
+      "loss": 3.5932,
+      "step": 19300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004541035739468751,
+      "loss": 3.3979,
+      "step": 19305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045403298616217963,
+      "loss": 3.4772,
+      "step": 19310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045396238679499293,
+      "loss": 3.5518,
+      "step": 19315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004538917758506237,
+      "loss": 3.4183,
+      "step": 19320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004538211533343817,
+      "loss": 3.5486,
+      "step": 19325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004537505192515773,
+      "loss": 3.3749,
+      "step": 19330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045367987360752174,
+      "loss": 3.4757,
+      "step": 19335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045360921640752737,
+      "loss": 3.4974,
+      "step": 19340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0004535385476569071,
+      "loss": 3.4206,
+      "step": 19345
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000453467867360975,
+      "loss": 3.5497,
+      "step": 19350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00045339717552504564,
+      "loss": 3.4373,
+      "step": 19355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004533264721544349,
+      "loss": 3.4262,
+      "step": 19360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004532557572544593,
+      "loss": 3.5594,
+      "step": 19365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004531850308304361,
+      "loss": 3.4997,
+      "step": 19370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045311429288768374,
+      "loss": 3.464,
+      "step": 19375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045304354343152117,
+      "loss": 3.4962,
+      "step": 19380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045297278246726844,
+      "loss": 3.5346,
+      "step": 19385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045290201000024654,
+      "loss": 3.5029,
+      "step": 19390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000452831226035777,
+      "loss": 3.5546,
+      "step": 19395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045276043057918263,
+      "loss": 3.4901,
+      "step": 19400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004526896236357866,
+      "loss": 3.4695,
+      "step": 19405
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004526188052109135,
+      "loss": 3.5047,
+      "step": 19410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045254797530988844,
+      "loss": 3.451,
+      "step": 19415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004524771339380374,
+      "loss": 3.436,
+      "step": 19420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045240628110068725,
+      "loss": 3.5381,
+      "step": 19425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045233541680316593,
+      "loss": 3.4288,
+      "step": 19430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004522645410508019,
+      "loss": 3.5495,
+      "step": 19435
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004521936538489248,
+      "loss": 3.4837,
+      "step": 19440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045212275520286477,
+      "loss": 3.4782,
+      "step": 19445
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045205184511795335,
+      "loss": 3.4985,
+      "step": 19450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004519809235995223,
+      "loss": 3.535,
+      "step": 19455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004519099906529047,
+      "loss": 3.5755,
+      "step": 19460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045183904628343444,
+      "loss": 3.5931,
+      "step": 19465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000451768090496446,
+      "loss": 3.5447,
+      "step": 19470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004516971232972749,
+      "loss": 3.435,
+      "step": 19475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004516261446912576,
+      "loss": 3.43,
+      "step": 19480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004515551546837313,
+      "loss": 3.4134,
+      "step": 19485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004514841532800341,
+      "loss": 3.393,
+      "step": 19490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045141314048550486,
+      "loss": 3.4855,
+      "step": 19495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004513421163054834,
+      "loss": 3.4721,
+      "step": 19500
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 3.5094563961029053,
+      "eval_runtime": 149.8708,
+      "eval_samples_per_second": 12.284,
+      "eval_steps_per_second": 0.774,
+      "step": 19500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045127108074531045,
+      "loss": 3.4603,
+      "step": 19505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045120003381032755,
+      "loss": 3.3118,
+      "step": 19510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045112897550587705,
+      "loss": 3.4607,
+      "step": 19515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004510579058373019,
+      "loss": 3.4662,
+      "step": 19520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004509868248099465,
+      "loss": 3.5243,
+      "step": 19525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004509157324291557,
+      "loss": 3.4312,
+      "step": 19530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004508446287002752,
+      "loss": 3.3722,
+      "step": 19535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004507735136286517,
+      "loss": 3.3688,
+      "step": 19540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045070238721963264,
+      "loss": 3.3627,
+      "step": 19545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004506312494785665,
+      "loss": 3.4587,
+      "step": 19550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045056010041080236,
+      "loss": 3.3304,
+      "step": 19555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004504889400216902,
+      "loss": 3.4772,
+      "step": 19560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045041776831658103,
+      "loss": 3.4572,
+      "step": 19565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004503465853008266,
+      "loss": 3.504,
+      "step": 19570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004502753909797794,
+      "loss": 3.4871,
+      "step": 19575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00045020418535879293,
+      "loss": 3.5595,
+      "step": 19580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004501329684432215,
+      "loss": 3.5133,
+      "step": 19585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004500617402384204,
+      "loss": 3.4042,
+      "step": 19590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004499905007497454,
+      "loss": 3.6127,
+      "step": 19595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044991924998255344,
+      "loss": 3.344,
+      "step": 19600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004498479879422022,
+      "loss": 3.4879,
+      "step": 19605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004497767146340502,
+      "loss": 3.5619,
+      "step": 19610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044970543006345686,
+      "loss": 3.4435,
+      "step": 19615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044963413423578245,
+      "loss": 3.5145,
+      "step": 19620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044956282715638796,
+      "loss": 3.5095,
+      "step": 19625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044949150883063545,
+      "loss": 3.5551,
+      "step": 19630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044942017926388755,
+      "loss": 3.5651,
+      "step": 19635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044934883846150793,
+      "loss": 3.3975,
+      "step": 19640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004492774864288611,
+      "loss": 3.5149,
+      "step": 19645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044920612317131224,
+      "loss": 3.4926,
+      "step": 19650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004491347486942277,
+      "loss": 3.3984,
+      "step": 19655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044906336300297423,
+      "loss": 3.3962,
+      "step": 19660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004489919661029198,
+      "loss": 3.4718,
+      "step": 19665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044892055799943323,
+      "loss": 3.4969,
+      "step": 19670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044884913869788377,
+      "loss": 3.3791,
+      "step": 19675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004487777082036421,
+      "loss": 3.4044,
+      "step": 19680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044870626652207903,
+      "loss": 3.4759,
+      "step": 19685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004486348136585669,
+      "loss": 3.4839,
+      "step": 19690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044856334961847867,
+      "loss": 3.5427,
+      "step": 19695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044849187440718774,
+      "loss": 3.5028,
+      "step": 19700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004484203880300689,
+      "loss": 3.5073,
+      "step": 19705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044834889049249755,
+      "loss": 3.4769,
+      "step": 19710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004482773817998499,
+      "loss": 3.5275,
+      "step": 19715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004482058619575031,
+      "loss": 3.5522,
+      "step": 19720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044813433097083493,
+      "loss": 3.5725,
+      "step": 19725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004480627888452243,
+      "loss": 3.3296,
+      "step": 19730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004479912355860508,
+      "loss": 3.575,
+      "step": 19735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044791967119869473,
+      "loss": 3.485,
+      "step": 19740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044784809568853754,
+      "loss": 3.5094,
+      "step": 19745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004477765090609612,
+      "loss": 3.4432,
+      "step": 19750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004477049113213488,
+      "loss": 3.4815,
+      "step": 19755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000447633302475084,
+      "loss": 3.5307,
+      "step": 19760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004475616825275514,
+      "loss": 3.479,
+      "step": 19765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004474900514841366,
+      "loss": 3.4396,
+      "step": 19770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004474184093502258,
+      "loss": 3.5377,
+      "step": 19775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004473467561312061,
+      "loss": 3.5034,
+      "step": 19780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044727509183246553,
+      "loss": 3.4338,
+      "step": 19785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004472034164593927,
+      "loss": 3.492,
+      "step": 19790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004471317300173776,
+      "loss": 3.478,
+      "step": 19795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004470600325118102,
+      "loss": 3.5671,
+      "step": 19800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044698832394808206,
+      "loss": 3.3984,
+      "step": 19805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004469166043315853,
+      "loss": 3.5496,
+      "step": 19810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004468448736677129,
+      "loss": 3.4328,
+      "step": 19815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004467731319618585,
+      "loss": 3.4768,
+      "step": 19820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004467013792194168,
+      "loss": 3.4958,
+      "step": 19825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044662961544578313,
+      "loss": 3.4901,
+      "step": 19830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044655784064635406,
+      "loss": 3.4205,
+      "step": 19835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044648605482652625,
+      "loss": 3.5096,
+      "step": 19840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004464142579916979,
+      "loss": 3.5586,
+      "step": 19845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044634245014726773,
+      "loss": 3.5259,
+      "step": 19850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044627063129863535,
+      "loss": 3.396,
+      "step": 19855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004461988014512011,
+      "loss": 3.4005,
+      "step": 19860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044612696061036613,
+      "loss": 3.4798,
+      "step": 19865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044605510878153277,
+      "loss": 3.5662,
+      "step": 19870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044598324597010366,
+      "loss": 3.4116,
+      "step": 19875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004459113721814826,
+      "loss": 3.4119,
+      "step": 19880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044583948742107416,
+      "loss": 3.4519,
+      "step": 19885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044576759169428365,
+      "loss": 3.4095,
+      "step": 19890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004456956850065173,
+      "loss": 3.5294,
+      "step": 19895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004456237673631822,
+      "loss": 3.3993,
+      "step": 19900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044555183876968595,
+      "loss": 3.4356,
+      "step": 19905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004454798992314375,
+      "loss": 3.4959,
+      "step": 19910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0004454079487538462,
+      "loss": 3.5206,
+      "step": 19915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044533598734232226,
+      "loss": 3.3217,
+      "step": 19920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000445264015002277,
+      "loss": 3.486,
+      "step": 19925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044519203173912214,
+      "loss": 3.5203,
+      "step": 19930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00044512003755827075,
+      "loss": 3.4889,
+      "step": 19935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004450480324651362,
+      "loss": 3.4719,
+      "step": 19940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044497601646513294,
+      "loss": 3.4491,
+      "step": 19945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044490398956367635,
+      "loss": 3.4391,
+      "step": 19950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004448319517661823,
+      "loss": 3.3457,
+      "step": 19955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044475990307806765,
+      "loss": 3.3828,
+      "step": 19960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044468784350475024,
+      "loss": 3.4094,
+      "step": 19965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004446157730516485,
+      "loss": 3.3578,
+      "step": 19970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044454369172418183,
+      "loss": 3.4901,
+      "step": 19975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004444715995277702,
+      "loss": 3.4578,
+      "step": 19980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004443994964678347,
+      "loss": 3.5705,
+      "step": 19985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004443273825497972,
+      "loss": 3.356,
+      "step": 19990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044425525777908015,
+      "loss": 3.5245,
+      "step": 19995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044418312216110703,
+      "loss": 3.506,
+      "step": 20000
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 3.5002212524414062,
+      "eval_runtime": 149.672,
+      "eval_samples_per_second": 12.3,
+      "eval_steps_per_second": 0.775,
+      "step": 20000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044411097570130204,
+      "loss": 3.4232,
+      "step": 20005
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004440388184050902,
+      "loss": 3.4982,
+      "step": 20010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044396665027789746,
+      "loss": 3.4513,
+      "step": 20015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044389447132515037,
+      "loss": 3.3997,
+      "step": 20020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004438222815522765,
+      "loss": 3.5083,
+      "step": 20025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004437500809647042,
+      "loss": 3.4299,
+      "step": 20030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004436778695678625,
+      "loss": 3.4987,
+      "step": 20035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004436056473671813,
+      "loss": 3.5377,
+      "step": 20040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004435334143680914,
+      "loss": 3.5088,
+      "step": 20045
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044346117057602425,
+      "loss": 3.4873,
+      "step": 20050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004433889159964125,
+      "loss": 3.474,
+      "step": 20055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044331665063468894,
+      "loss": 3.5821,
+      "step": 20060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044324437449628785,
+      "loss": 3.5429,
+      "step": 20065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044317208758664386,
+      "loss": 3.2625,
+      "step": 20070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044309978991119263,
+      "loss": 3.3802,
+      "step": 20075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004430274814753706,
+      "loss": 3.5245,
+      "step": 20080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000442955162284615,
+      "loss": 3.5747,
+      "step": 20085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004428828323443638,
+      "loss": 3.4744,
+      "step": 20090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044281049166005585,
+      "loss": 3.5422,
+      "step": 20095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004427381402371308,
+      "loss": 3.3606,
+      "step": 20100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044266577808102915,
+      "loss": 3.4586,
+      "step": 20105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004425934051971921,
+      "loss": 3.4529,
+      "step": 20110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044252102159106187,
+      "loss": 3.4736,
+      "step": 20115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044244862726808114,
+      "loss": 3.415,
+      "step": 20120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004423762222336936,
+      "loss": 3.4178,
+      "step": 20125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044230380649334396,
+      "loss": 3.4815,
+      "step": 20130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004422313800524773,
+      "loss": 3.4722,
+      "step": 20135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004421589429165399,
+      "loss": 3.5723,
+      "step": 20140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044208649509097837,
+      "loss": 3.4964,
+      "step": 20145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044201403658124055,
+      "loss": 3.3457,
+      "step": 20150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044194156739277524,
+      "loss": 3.5067,
+      "step": 20155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044186908753103137,
+      "loss": 3.4389,
+      "step": 20160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004417965970014591,
+      "loss": 3.3161,
+      "step": 20165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044172409580950947,
+      "loss": 3.5093,
+      "step": 20170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004416515839606343,
+      "loss": 3.3348,
+      "step": 20175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044157906146028586,
+      "loss": 3.4568,
+      "step": 20180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044150652831391755,
+      "loss": 3.4137,
+      "step": 20185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004414339845269835,
+      "loss": 3.3819,
+      "step": 20190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044136143010493884,
+      "loss": 3.5321,
+      "step": 20195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044128886505323897,
+      "loss": 3.458,
+      "step": 20200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044121628937734053,
+      "loss": 3.4455,
+      "step": 20205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004411437030827008,
+      "loss": 3.4663,
+      "step": 20210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004410711061747781,
+      "loss": 3.5031,
+      "step": 20215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044099849865903107,
+      "loss": 3.526,
+      "step": 20220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004409258805409196,
+      "loss": 3.4007,
+      "step": 20225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004408532518259041,
+      "loss": 3.4619,
+      "step": 20230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000440780612519446,
+      "loss": 3.4583,
+      "step": 20235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004407079626270072,
+      "loss": 3.4842,
+      "step": 20240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004406353021540509,
+      "loss": 3.5053,
+      "step": 20245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044056263110604045,
+      "loss": 3.5203,
+      "step": 20250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044048994948844063,
+      "loss": 3.5216,
+      "step": 20255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004404172573067166,
+      "loss": 3.4012,
+      "step": 20260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004403445545663344,
+      "loss": 3.4549,
+      "step": 20265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000440271841272761,
+      "loss": 3.3967,
+      "step": 20270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044019911743146397,
+      "loss": 3.4056,
+      "step": 20275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004401263830479118,
+      "loss": 3.5852,
+      "step": 20280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00044005363812757377,
+      "loss": 3.5015,
+      "step": 20285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043998088267591996,
+      "loss": 3.5004,
+      "step": 20290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043990811669842114,
+      "loss": 3.4759,
+      "step": 20295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043983534020054907,
+      "loss": 3.4747,
+      "step": 20300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043976255318777587,
+      "loss": 3.4997,
+      "step": 20305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004396897556655751,
+      "loss": 3.4021,
+      "step": 20310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043961694763942053,
+      "loss": 3.5991,
+      "step": 20315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043954412911478705,
+      "loss": 3.4963,
+      "step": 20320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004394713000971501,
+      "loss": 3.3867,
+      "step": 20325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004393984605919863,
+      "loss": 3.493,
+      "step": 20330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004393256106047726,
+      "loss": 3.4838,
+      "step": 20335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004392527501409871,
+      "loss": 3.4146,
+      "step": 20340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004391798792061084,
+      "loss": 3.4319,
+      "step": 20345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000439106997805616,
+      "loss": 3.3523,
+      "step": 20350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004390341059449903,
+      "loss": 3.4308,
+      "step": 20355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043896120362971246,
+      "loss": 3.3949,
+      "step": 20360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004388882908652642,
+      "loss": 3.5024,
+      "step": 20365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004388153676571283,
+      "loss": 3.4641,
+      "step": 20370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043874243401078813,
+      "loss": 3.4261,
+      "step": 20375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043866948993172793,
+      "loss": 3.4187,
+      "step": 20380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043859653542543286,
+      "loss": 3.4059,
+      "step": 20385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043852357049738855,
+      "loss": 3.381,
+      "step": 20390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004384505951530817,
+      "loss": 3.5046,
+      "step": 20395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004383776093979996,
+      "loss": 3.4859,
+      "step": 20400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043830461323763047,
+      "loss": 3.5565,
+      "step": 20405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004382316066774633,
+      "loss": 3.4298,
+      "step": 20410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043815858972298763,
+      "loss": 3.4938,
+      "step": 20415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043808556237969403,
+      "loss": 3.5072,
+      "step": 20420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043801252465307396,
+      "loss": 3.5391,
+      "step": 20425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004379394765486192,
+      "loss": 3.4818,
+      "step": 20430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004378664180718228,
+      "loss": 3.4382,
+      "step": 20435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004377933492281783,
+      "loss": 3.5275,
+      "step": 20440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004377202700231801,
+      "loss": 3.4213,
+      "step": 20445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043764718046232346,
+      "loss": 3.359,
+      "step": 20450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004375740805511043,
+      "loss": 3.5426,
+      "step": 20455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043750097029501925,
+      "loss": 3.4999,
+      "step": 20460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004374278496995659,
+      "loss": 3.4504,
+      "step": 20465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004373547187702426,
+      "loss": 3.4479,
+      "step": 20470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043728157751254834,
+      "loss": 3.4309,
+      "step": 20475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004372084259319831,
+      "loss": 3.468,
+      "step": 20480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043713526403404746,
+      "loss": 3.4853,
+      "step": 20485
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004370620918242426,
+      "loss": 3.4808,
+      "step": 20490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000436988909308071,
+      "loss": 3.4663,
+      "step": 20495
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0004369157164910354,
+      "loss": 3.3601,
+      "step": 20500
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 3.4890780448913574,
+      "eval_runtime": 149.7679,
+      "eval_samples_per_second": 12.292,
+      "eval_steps_per_second": 0.775,
+      "step": 20500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043684251337863957,
+      "loss": 3.503,
+      "step": 20505
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00043676929997638807,
+      "loss": 3.4588,
+      "step": 20510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043669607628978616,
+      "loss": 3.5404,
+      "step": 20515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043662284232433995,
+      "loss": 3.4366,
+      "step": 20520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004365495980855561,
+      "loss": 3.3995,
+      "step": 20525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043647634357894225,
+      "loss": 3.3957,
+      "step": 20530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004364030788100069,
+      "loss": 3.5176,
+      "step": 20535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043632980378425894,
+      "loss": 3.5703,
+      "step": 20540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043625651850720846,
+      "loss": 3.4949,
+      "step": 20545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043618322298436614,
+      "loss": 3.4727,
+      "step": 20550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004361099172212433,
+      "loss": 3.4264,
+      "step": 20555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004360366012233523,
+      "loss": 3.4806,
+      "step": 20560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043596327499620613,
+      "loss": 3.4705,
+      "step": 20565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043588993854531835,
+      "loss": 3.4814,
+      "step": 20570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004358165918762037,
+      "loss": 3.4198,
+      "step": 20575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043574323499437736,
+      "loss": 3.3648,
+      "step": 20580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004356698679053554,
+      "loss": 3.5215,
+      "step": 20585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004355964906146547,
+      "loss": 3.4305,
+      "step": 20590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004355231031277928,
+      "loss": 3.3691,
+      "step": 20595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004354497054502882,
+      "loss": 3.4175,
+      "step": 20600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043537629758765977,
+      "loss": 3.4657,
+      "step": 20605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004353028795454275,
+      "loss": 3.5085,
+      "step": 20610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043522945132911235,
+      "loss": 3.4169,
+      "step": 20615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004351560129442354,
+      "loss": 3.4685,
+      "step": 20620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043508256439631884,
+      "loss": 3.5605,
+      "step": 20625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043500910569088577,
+      "loss": 3.4576,
+      "step": 20630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004349356368334599,
+      "loss": 3.5073,
+      "step": 20635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004348621578295657,
+      "loss": 3.456,
+      "step": 20640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043478866868472833,
+      "loss": 3.3731,
+      "step": 20645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004347151694044738,
+      "loss": 3.4255,
+      "step": 20650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043464165999432903,
+      "loss": 3.4228,
+      "step": 20655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004345681404598215,
+      "loss": 3.4786,
+      "step": 20660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004344946108064794,
+      "loss": 3.4763,
+      "step": 20665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004344210710398318,
+      "loss": 3.4635,
+      "step": 20670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043434752116540854,
+      "loss": 3.4831,
+      "step": 20675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004342739611887403,
+      "loss": 3.4503,
+      "step": 20680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004342003911153583,
+      "loss": 3.4472,
+      "step": 20685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043412681095079465,
+      "loss": 3.346,
+      "step": 20690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043405322070058216,
+      "loss": 3.3185,
+      "step": 20695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004339796203702545,
+      "loss": 3.5082,
+      "step": 20700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043390600996534606,
+      "loss": 3.5576,
+      "step": 20705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004338323894913919,
+      "loss": 3.4343,
+      "step": 20710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043375875895392793,
+      "loss": 3.5085,
+      "step": 20715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004336851183584908,
+      "loss": 3.3682,
+      "step": 20720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004336114677106179,
+      "loss": 3.419,
+      "step": 20725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004335378070158474,
+      "loss": 3.5034,
+      "step": 20730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043346413627971803,
+      "loss": 3.4361,
+      "step": 20735
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043339045550776976,
+      "loss": 3.402,
+      "step": 20740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043331676470554286,
+      "loss": 3.4727,
+      "step": 20745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004332430638785784,
+      "loss": 3.5097,
+      "step": 20750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004331693530324184,
+      "loss": 3.4251,
+      "step": 20755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043309563217260564,
+      "loss": 3.4258,
+      "step": 20760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004330219013046834,
+      "loss": 3.3033,
+      "step": 20765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004329481604341959,
+      "loss": 3.4339,
+      "step": 20770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000432874409566688,
+      "loss": 3.4677,
+      "step": 20775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004328006487077056,
+      "loss": 3.3985,
+      "step": 20780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043272687786279497,
+      "loss": 3.4579,
+      "step": 20785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043265309703750327,
+      "loss": 3.3846,
+      "step": 20790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004325793062373786,
+      "loss": 3.5574,
+      "step": 20795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043250550546796953,
+      "loss": 3.4578,
+      "step": 20800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004324316947348255,
+      "loss": 3.4623,
+      "step": 20805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004323578740434967,
+      "loss": 3.5179,
+      "step": 20810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004322840433995342,
+      "loss": 3.459,
+      "step": 20815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004322102028084896,
+      "loss": 3.478,
+      "step": 20820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043213635227591516,
+      "loss": 3.3421,
+      "step": 20825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004320624918073643,
+      "loss": 3.3589,
+      "step": 20830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004319886214083908,
+      "loss": 3.4584,
+      "step": 20835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004319147410845495,
+      "loss": 3.3751,
+      "step": 20840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004318408508413957,
+      "loss": 3.6159,
+      "step": 20845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043176695068448555,
+      "loss": 3.3952,
+      "step": 20850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000431693040619376,
+      "loss": 3.5016,
+      "step": 20855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004316191206516248,
+      "loss": 3.412,
+      "step": 20860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004315451907867902,
+      "loss": 3.5053,
+      "step": 20865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043147125103043147,
+      "loss": 3.4214,
+      "step": 20870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043139730138810835,
+      "loss": 3.4585,
+      "step": 20875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004313233418653816,
+      "loss": 3.3709,
+      "step": 20880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043124937246781254,
+      "loss": 3.419,
+      "step": 20885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043117539320096333,
+      "loss": 3.4385,
+      "step": 20890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043110140407039686,
+      "loss": 3.4396,
+      "step": 20895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004310274050816767,
+      "loss": 3.3851,
+      "step": 20900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043095339624036705,
+      "loss": 3.4022,
+      "step": 20905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004308793775520333,
+      "loss": 3.4697,
+      "step": 20910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043080534902224094,
+      "loss": 3.3228,
+      "step": 20915
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043073131065655685,
+      "loss": 3.4832,
+      "step": 20920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004306572624605481,
+      "loss": 3.4429,
+      "step": 20925
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004305832044397828,
+      "loss": 3.3656,
+      "step": 20930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043050913659982984,
+      "loss": 3.4722,
+      "step": 20935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043043505894625864,
+      "loss": 3.4771,
+      "step": 20940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004303609714846394,
+      "loss": 3.4003,
+      "step": 20945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00043028687422054317,
+      "loss": 3.5443,
+      "step": 20950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004302127671595418,
+      "loss": 3.4675,
+      "step": 20955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004301386503072077,
+      "loss": 3.4031,
+      "step": 20960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000430064523669114,
+      "loss": 3.3901,
+      "step": 20965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004299903872508346,
+      "loss": 3.4691,
+      "step": 20970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00042991624105794435,
+      "loss": 3.4782,
+      "step": 20975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004298420850960186,
+      "loss": 3.481,
+      "step": 20980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004297679193706335,
+      "loss": 3.4205,
+      "step": 20985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00042969374388736586,
+      "loss": 3.4272,
+      "step": 20990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004296195586517933,
+      "loss": 3.4391,
+      "step": 20995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00042954536366949435,
+      "loss": 3.5532,
+      "step": 21000
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 3.4810240268707275,
+      "eval_runtime": 149.6804,
+      "eval_samples_per_second": 12.3,
+      "eval_steps_per_second": 0.775,
+      "step": 21000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004294711589460479,
+      "loss": 3.5336,
+      "step": 21005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00042939694448703385,
+      "loss": 3.5589,
+      "step": 21010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004293227202980327,
+      "loss": 3.3975,
+      "step": 21015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004292484863846257,
+      "loss": 3.4718,
+      "step": 21020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000429174242752395,
+      "loss": 3.4397,
+      "step": 21025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004290999894069232,
+      "loss": 3.5044,
+      "step": 21030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004290257263537938,
+      "loss": 3.4743,
+      "step": 21035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004289514535985911,
+      "loss": 3.4612,
+      "step": 21040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004288771711468999,
+      "loss": 3.5065,
+      "step": 21045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004288028790043059,
+      "loss": 3.3358,
+      "step": 21050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00042872857717639556,
+      "loss": 3.4931,
+      "step": 21055
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004286542656687559,
+      "loss": 3.4695,
+      "step": 21060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00042857994448697485,
+      "loss": 3.3413,
+      "step": 21065
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004285056136366409,
+      "loss": 3.497,
+      "step": 21070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00042843127312334345,
+      "loss": 3.4309,
+      "step": 21075
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00042835692295267237,
+      "loss": 3.5423,
+      "step": 21080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004282825631302185,
+      "loss": 3.4508,
+      "step": 21085
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0004282081936615734,
+      "loss": 3.4325,
+      "step": 21090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004281338145523291,
+      "loss": 3.4188,
+      "step": 21095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042805942580807863,
+      "loss": 3.3961,
+      "step": 21100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042798502743441577,
+      "loss": 3.5294,
+      "step": 21105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004279106194369346,
+      "loss": 3.5052,
+      "step": 21110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004278362018212304,
+      "loss": 3.4474,
+      "step": 21115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004277617745928991,
+      "loss": 3.4268,
+      "step": 21120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042768733775753705,
+      "loss": 3.4271,
+      "step": 21125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004276128913207415,
+      "loss": 3.319,
+      "step": 21130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042753843528811055,
+      "loss": 3.4287,
+      "step": 21135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000427463969665243,
+      "loss": 3.502,
+      "step": 21140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004273894944577381,
+      "loss": 3.3916,
+      "step": 21145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042731500967119613,
+      "loss": 3.4751,
+      "step": 21150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004272405153112179,
+      "loss": 3.3869,
+      "step": 21155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000427166011383405,
+      "loss": 3.3328,
+      "step": 21160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004270914978933598,
+      "loss": 3.5021,
+      "step": 21165
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042701697484668515,
+      "loss": 3.5497,
+      "step": 21170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004269424422489851,
+      "loss": 3.4791,
+      "step": 21175
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042686790010586386,
+      "loss": 3.4404,
+      "step": 21180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004267933484229267,
+      "loss": 3.3707,
+      "step": 21185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004267187872057797,
+      "loss": 3.4202,
+      "step": 21190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042664421646002916,
+      "loss": 3.4687,
+      "step": 21195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042656963619128267,
+      "loss": 3.4317,
+      "step": 21200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042649504640514816,
+      "loss": 3.3562,
+      "step": 21205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004264204471072345,
+      "loss": 3.5205,
+      "step": 21210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042634583830315103,
+      "loss": 3.3985,
+      "step": 21215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042627121999850806,
+      "loss": 3.4669,
+      "step": 21220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004261965921989165,
+      "loss": 3.3633,
+      "step": 21225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042612195490998795,
+      "loss": 3.4944,
+      "step": 21230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004260473081373347,
+      "loss": 3.4313,
+      "step": 21235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042597265188656995,
+      "loss": 3.3019,
+      "step": 21240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004258979861633073,
+      "loss": 3.5405,
+      "step": 21245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004258233109731613,
+      "loss": 3.4114,
+      "step": 21250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004257486263217473,
+      "loss": 3.5809,
+      "step": 21255
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042567393221468094,
+      "loss": 3.4726,
+      "step": 21260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042559922865757896,
+      "loss": 3.4404,
+      "step": 21265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004255245156560586,
+      "loss": 3.4897,
+      "step": 21270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042544979321573794,
+      "loss": 3.4024,
+      "step": 21275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042537506134223595,
+      "loss": 3.4146,
+      "step": 21280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004253003200411717,
+      "loss": 3.2768,
+      "step": 21285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004252255693181657,
+      "loss": 3.3862,
+      "step": 21290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004251508091788385,
+      "loss": 3.4472,
+      "step": 21295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042507603962881193,
+      "loss": 3.3695,
+      "step": 21300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042500126067370814,
+      "loss": 3.5154,
+      "step": 21305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042492647231915016,
+      "loss": 3.466,
+      "step": 21310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004248516745707618,
+      "loss": 3.5312,
+      "step": 21315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004247768674341672,
+      "loss": 3.4309,
+      "step": 21320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004247020509149918,
+      "loss": 3.3506,
+      "step": 21325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042462722501886114,
+      "loss": 3.3954,
+      "step": 21330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042455238975140186,
+      "loss": 3.4317,
+      "step": 21335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004244775451182413,
+      "loss": 3.4081,
+      "step": 21340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004244026911250072,
+      "loss": 3.4154,
+      "step": 21345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004243278277773283,
+      "loss": 3.378,
+      "step": 21350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000424252955080834,
+      "loss": 3.4763,
+      "step": 21355
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004241780730411541,
+      "loss": 3.3921,
+      "step": 21360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004241031816639197,
+      "loss": 3.4287,
+      "step": 21365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004240282809547619,
+      "loss": 3.4556,
+      "step": 21370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042395337091931314,
+      "loss": 3.5046,
+      "step": 21375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004238784515632061,
+      "loss": 3.3785,
+      "step": 21380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004238035228920743,
+      "loss": 3.4726,
+      "step": 21385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042372858491155203,
+      "loss": 3.5175,
+      "step": 21390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004236536376272744,
+      "loss": 3.4565,
+      "step": 21395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042357868104487683,
+      "loss": 3.4254,
+      "step": 21400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042350371516999586,
+      "loss": 3.4177,
+      "step": 21405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004234287400082684,
+      "loss": 3.4484,
+      "step": 21410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004233537555653322,
+      "loss": 3.4054,
+      "step": 21415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042327876184682594,
+      "loss": 3.3857,
+      "step": 21420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004232037588583885,
+      "loss": 3.451,
+      "step": 21425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004231287466056597,
+      "loss": 3.4559,
+      "step": 21430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004230537250942803,
+      "loss": 3.4196,
+      "step": 21435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004229786943298914,
+      "loss": 3.454,
+      "step": 21440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004229036543181349,
+      "loss": 3.4613,
+      "step": 21445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004228286050646535,
+      "loss": 3.4041,
+      "step": 21450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042275354657509036,
+      "loss": 3.4881,
+      "step": 21455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042267847885508976,
+      "loss": 3.5192,
+      "step": 21460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042260340191029624,
+      "loss": 3.4635,
+      "step": 21465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004225283157463553,
+      "loss": 3.4071,
+      "step": 21470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004224532203689128,
+      "loss": 3.4137,
+      "step": 21475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004223781157836158,
+      "loss": 3.3856,
+      "step": 21480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004223030019961116,
+      "loss": 3.4598,
+      "step": 21485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042222787901204854,
+      "loss": 3.4487,
+      "step": 21490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004221527468370754,
+      "loss": 3.4663,
+      "step": 21495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042207760547684155,
+      "loss": 3.4861,
+      "step": 21500
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 3.4763381481170654,
+      "eval_runtime": 149.6778,
+      "eval_samples_per_second": 12.3,
+      "eval_steps_per_second": 0.775,
+      "step": 21500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004220024549369975,
+      "loss": 3.5112,
+      "step": 21505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004219272952231942,
+      "loss": 3.4445,
+      "step": 21510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000421852126341083,
+      "loss": 3.4347,
+      "step": 21515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042177694829631653,
+      "loss": 3.4521,
+      "step": 21520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042170176109454754,
+      "loss": 3.3605,
+      "step": 21525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004216265647414298,
+      "loss": 3.3925,
+      "step": 21530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042155135924261785,
+      "loss": 3.4833,
+      "step": 21535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042147614460376646,
+      "loss": 3.435,
+      "step": 21540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004214009208305317,
+      "loss": 3.4267,
+      "step": 21545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004213256879285698,
+      "loss": 3.4272,
+      "step": 21550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042125044590353796,
+      "loss": 3.3872,
+      "step": 21555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042117519476109395,
+      "loss": 3.406,
+      "step": 21560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004210999345068963,
+      "loss": 3.3582,
+      "step": 21565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004210246651466043,
+      "loss": 3.4148,
+      "step": 21570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042094938668587765,
+      "loss": 3.4063,
+      "step": 21575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000420874099130377,
+      "loss": 3.472,
+      "step": 21580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004207988024857636,
+      "loss": 3.5276,
+      "step": 21585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042072349675769923,
+      "loss": 3.5215,
+      "step": 21590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042064818195184663,
+      "loss": 3.5187,
+      "step": 21595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004205728580738691,
+      "loss": 3.4223,
+      "step": 21600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004204975251294306,
+      "loss": 3.3929,
+      "step": 21605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042042218312419576,
+      "loss": 3.5108,
+      "step": 21610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042034683206382983,
+      "loss": 3.571,
+      "step": 21615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004202714719539989,
+      "loss": 3.5817,
+      "step": 21620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004201961028003697,
+      "loss": 3.4792,
+      "step": 21625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00042012072460860956,
+      "loss": 3.4894,
+      "step": 21630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004200453373843865,
+      "loss": 3.4205,
+      "step": 21635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004199699411333693,
+      "loss": 3.5227,
+      "step": 21640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00041989453586122744,
+      "loss": 3.3261,
+      "step": 21645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00041981912157363087,
+      "loss": 3.4374,
+      "step": 21650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0004197436982762504,
+      "loss": 3.4541,
+      "step": 21655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00041966826597475757,
+      "loss": 3.4462,
+      "step": 21660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00041959282467482435,
+      "loss": 3.4077,
+      "step": 21665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041951737438212367,
+      "loss": 3.3667,
+      "step": 21670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004194419151023289,
+      "loss": 3.4438,
+      "step": 21675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004193664468411143,
+      "loss": 3.4317,
+      "step": 21680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041929096960415457,
+      "loss": 3.4413,
+      "step": 21685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004192154833971254,
+      "loss": 3.507,
+      "step": 21690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041913998822570267,
+      "loss": 3.5172,
+      "step": 21695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004190644840955636,
+      "loss": 3.4648,
+      "step": 21700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004189889710123854,
+      "loss": 3.4228,
+      "step": 21705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041891344898184647,
+      "loss": 3.383,
+      "step": 21710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041883791800962554,
+      "loss": 3.4611,
+      "step": 21715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004187623781014021,
+      "loss": 3.4332,
+      "step": 21720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041868682926285677,
+      "loss": 3.3851,
+      "step": 21725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004186112714996699,
+      "loss": 3.4926,
+      "step": 21730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041853570481752335,
+      "loss": 3.3628,
+      "step": 21735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004184601292220993,
+      "loss": 3.4518,
+      "step": 21740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041838454471908066,
+      "loss": 3.2995,
+      "step": 21745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000418308951314151,
+      "loss": 3.4446,
+      "step": 21750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041823334901299435,
+      "loss": 3.5188,
+      "step": 21755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041815773782129587,
+      "loss": 3.4001,
+      "step": 21760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004180821177447411,
+      "loss": 3.4094,
+      "step": 21765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041800648878901627,
+      "loss": 3.3152,
+      "step": 21770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041793085095980816,
+      "loss": 3.3421,
+      "step": 21775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004178552042628045,
+      "loss": 3.4248,
+      "step": 21780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041777954870369344,
+      "loss": 3.401,
+      "step": 21785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041770388428816397,
+      "loss": 3.5127,
+      "step": 21790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041762821102190554,
+      "loss": 3.4763,
+      "step": 21795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004175525289106085,
+      "loss": 3.475,
+      "step": 21800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041747683795996385,
+      "loss": 3.3599,
+      "step": 21805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000417401138175663,
+      "loss": 3.4129,
+      "step": 21810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041732542956339816,
+      "loss": 3.4717,
+      "step": 21815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004172497121288623,
+      "loss": 3.4032,
+      "step": 21820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041717398587774913,
+      "loss": 3.442,
+      "step": 21825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041709825081575263,
+      "loss": 3.4066,
+      "step": 21830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004170225069485677,
+      "loss": 3.4771,
+      "step": 21835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041694675428189006,
+      "loss": 3.4156,
+      "step": 21840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004168709928214159,
+      "loss": 3.3453,
+      "step": 21845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041679522257284196,
+      "loss": 3.5143,
+      "step": 21850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041671944354186585,
+      "loss": 3.5789,
+      "step": 21855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041664365573418577,
+      "loss": 3.406,
+      "step": 21860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041656785915550055,
+      "loss": 3.4958,
+      "step": 21865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004164920538115098,
+      "loss": 3.4977,
+      "step": 21870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004164162397079135,
+      "loss": 3.4383,
+      "step": 21875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041634041685041266,
+      "loss": 3.5104,
+      "step": 21880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004162645852447087,
+      "loss": 3.4549,
+      "step": 21885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004161887448965039,
+      "loss": 3.4399,
+      "step": 21890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004161128958115008,
+      "loss": 3.3542,
+      "step": 21895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004160370379954029,
+      "loss": 3.4418,
+      "step": 21900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041596117145391466,
+      "loss": 3.3958,
+      "step": 21905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004158852961927405,
+      "loss": 3.3919,
+      "step": 21910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004158094122175859,
+      "loss": 3.3159,
+      "step": 21915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004157335195341571,
+      "loss": 3.5038,
+      "step": 21920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004156576181481608,
+      "loss": 3.3401,
+      "step": 21925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041558170806530435,
+      "loss": 3.4432,
+      "step": 21930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004155057892912957,
+      "loss": 3.3199,
+      "step": 21935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041542986183184367,
+      "loss": 3.5088,
+      "step": 21940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041535392569265765,
+      "loss": 3.5437,
+      "step": 21945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004152779808794476,
+      "loss": 3.3675,
+      "step": 21950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004152020273979242,
+      "loss": 3.4963,
+      "step": 21955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041512606525379866,
+      "loss": 3.422,
+      "step": 21960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004150500944527831,
+      "loss": 3.3885,
+      "step": 21965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041497411500059,
+      "loss": 3.4297,
+      "step": 21970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041489812690293274,
+      "loss": 3.447,
+      "step": 21975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041482213016552513,
+      "loss": 3.4361,
+      "step": 21980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004147461247940818,
+      "loss": 3.423,
+      "step": 21985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004146701107943179,
+      "loss": 3.4564,
+      "step": 21990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041459408817194946,
+      "loss": 3.3803,
+      "step": 21995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041451805693269283,
+      "loss": 3.4608,
+      "step": 22000
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 3.4630024433135986,
+      "eval_runtime": 149.7666,
+      "eval_samples_per_second": 12.292,
+      "eval_steps_per_second": 0.775,
+      "step": 22000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004144420170822652,
+      "loss": 3.4355,
+      "step": 22005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004143659686263844,
+      "loss": 3.4343,
+      "step": 22010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004142899115707688,
+      "loss": 3.3277,
+      "step": 22015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004142138459211377,
+      "loss": 3.4995,
+      "step": 22020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004141377716832107,
+      "loss": 3.4184,
+      "step": 22025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041406168886270816,
+      "loss": 3.3363,
+      "step": 22030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041398559746535113,
+      "loss": 3.3641,
+      "step": 22035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004139094974968614,
+      "loss": 3.4262,
+      "step": 22040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004138333889629613,
+      "loss": 3.4048,
+      "step": 22045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004137572718693737,
+      "loss": 3.4102,
+      "step": 22050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041368114622182214,
+      "loss": 3.4041,
+      "step": 22055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004136050120260311,
+      "loss": 3.4719,
+      "step": 22060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004135288692877254,
+      "loss": 3.4398,
+      "step": 22065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004134527180126305,
+      "loss": 3.3921,
+      "step": 22070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004133765582064726,
+      "loss": 3.4075,
+      "step": 22075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041330038987497856,
+      "loss": 3.4711,
+      "step": 22080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000413224213023876,
+      "loss": 3.4251,
+      "step": 22085
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041314802765889276,
+      "loss": 3.3675,
+      "step": 22090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004130718337857578,
+      "loss": 3.362,
+      "step": 22095
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004129956314102003,
+      "loss": 3.4152,
+      "step": 22100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004129194205379506,
+      "loss": 3.4164,
+      "step": 22105
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041284320117473906,
+      "loss": 3.476,
+      "step": 22110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041276697332629707,
+      "loss": 3.3858,
+      "step": 22115
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004126907369983566,
+      "loss": 3.3862,
+      "step": 22120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004126144921966503,
+      "loss": 3.3574,
+      "step": 22125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041253823892691135,
+      "loss": 3.358,
+      "step": 22130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004124619771948735,
+      "loss": 3.4601,
+      "step": 22135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041238570700627143,
+      "loss": 3.566,
+      "step": 22140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041230942836684,
+      "loss": 3.4623,
+      "step": 22145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041223314128231535,
+      "loss": 3.4965,
+      "step": 22150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041215684575843345,
+      "loss": 3.3462,
+      "step": 22155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041208054180093175,
+      "loss": 3.3736,
+      "step": 22160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041200422941554753,
+      "loss": 3.503,
+      "step": 22165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004119279086080193,
+      "loss": 3.5064,
+      "step": 22170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041185157938408603,
+      "loss": 3.4897,
+      "step": 22175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041177524174948715,
+      "loss": 3.4477,
+      "step": 22180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000411698895709963,
+      "loss": 3.3709,
+      "step": 22185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004116225412712544,
+      "loss": 3.4257,
+      "step": 22190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004115461784391026,
+      "loss": 3.3655,
+      "step": 22195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004114698072192499,
+      "loss": 3.4449,
+      "step": 22200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000411393427617439,
+      "loss": 3.4398,
+      "step": 22205
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004113170396394133,
+      "loss": 3.1629,
+      "step": 22210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004112406432909165,
+      "loss": 3.3966,
+      "step": 22215
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004111642385776935,
+      "loss": 3.3941,
+      "step": 22220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004110878255054896,
+      "loss": 3.3217,
+      "step": 22225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0004110114040800504,
+      "loss": 3.3977,
+      "step": 22230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041093497430712267,
+      "loss": 3.4394,
+      "step": 22235
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041085853619245323,
+      "loss": 3.4824,
+      "step": 22240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00041078208974179017,
+      "loss": 3.4449,
+      "step": 22245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00041070563496088165,
+      "loss": 3.4106,
+      "step": 22250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00041062917185547666,
+      "loss": 3.4365,
+      "step": 22255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00041055270043132493,
+      "loss": 3.4706,
+      "step": 22260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004104762206941768,
+      "loss": 3.3919,
+      "step": 22265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000410399732649783,
+      "loss": 3.4725,
+      "step": 22270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00041032323630389516,
+      "loss": 3.5326,
+      "step": 22275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00041024673166226524,
+      "loss": 3.2729,
+      "step": 22280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00041017021873064613,
+      "loss": 3.4428,
+      "step": 22285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004100936975147913,
+      "loss": 3.4666,
+      "step": 22290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004100171680204545,
+      "loss": 3.4196,
+      "step": 22295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004099406302533905,
+      "loss": 3.4659,
+      "step": 22300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040986408421935454,
+      "loss": 3.4275,
+      "step": 22305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040978752992410264,
+      "loss": 3.4052,
+      "step": 22310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000409710967373391,
+      "loss": 3.5221,
+      "step": 22315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004096343965729769,
+      "loss": 3.5078,
+      "step": 22320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040955781752861805,
+      "loss": 3.4321,
+      "step": 22325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040948123024607286,
+      "loss": 3.4621,
+      "step": 22330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040940463473110013,
+      "loss": 3.3809,
+      "step": 22335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004093280309894596,
+      "loss": 3.4517,
+      "step": 22340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040925141902691144,
+      "loss": 3.4528,
+      "step": 22345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040917479884921654,
+      "loss": 3.4547,
+      "step": 22350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004090981704621362,
+      "loss": 3.4308,
+      "step": 22355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004090215338714325,
+      "loss": 3.2692,
+      "step": 22360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004089448890828683,
+      "loss": 3.4193,
+      "step": 22365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040886823610220674,
+      "loss": 3.4402,
+      "step": 22370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004087915749352117,
+      "loss": 3.4521,
+      "step": 22375
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040871490558764786,
+      "loss": 3.3807,
+      "step": 22380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004086382280652802,
+      "loss": 3.4643,
+      "step": 22385
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004085615423738747,
+      "loss": 3.4981,
+      "step": 22390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004084848485191974,
+      "loss": 3.5087,
+      "step": 22395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040840814650701553,
+      "loss": 3.453,
+      "step": 22400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004083314363430967,
+      "loss": 3.3977,
+      "step": 22405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040825471803320894,
+      "loss": 3.357,
+      "step": 22410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004081779915831213,
+      "loss": 3.4247,
+      "step": 22415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040810125699860297,
+      "loss": 3.3161,
+      "step": 22420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040802451428542414,
+      "loss": 3.4269,
+      "step": 22425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004079477634493556,
+      "loss": 3.4299,
+      "step": 22430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040787100449616833,
+      "loss": 3.4994,
+      "step": 22435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004077942374316344,
+      "loss": 3.3577,
+      "step": 22440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004077174622615262,
+      "loss": 3.353,
+      "step": 22445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004076406789916169,
+      "loss": 3.5383,
+      "step": 22450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040756388762768023,
+      "loss": 3.4003,
+      "step": 22455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004074870881754904,
+      "loss": 3.4799,
+      "step": 22460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004074102806408225,
+      "loss": 3.3984,
+      "step": 22465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004073334650294519,
+      "loss": 3.4481,
+      "step": 22470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040725664134715485,
+      "loss": 3.4943,
+      "step": 22475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040717980959970794,
+      "loss": 3.4733,
+      "step": 22480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004071029697928887,
+      "loss": 3.4076,
+      "step": 22485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004070261219324751,
+      "loss": 3.4849,
+      "step": 22490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040694926602424553,
+      "loss": 3.4302,
+      "step": 22495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004068724020739793,
+      "loss": 3.4945,
+      "step": 22500
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 3.4543726444244385,
+      "eval_runtime": 149.6841,
+      "eval_samples_per_second": 12.299,
+      "eval_steps_per_second": 0.775,
+      "step": 22500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040679553008745613,
+      "loss": 3.5376,
+      "step": 22505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040671865007045654,
+      "loss": 3.2163,
+      "step": 22510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040664176202876123,
+      "loss": 3.4521,
+      "step": 22515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040656486596815194,
+      "loss": 3.3826,
+      "step": 22520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040648796189441094,
+      "loss": 3.3821,
+      "step": 22525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040641104981332097,
+      "loss": 3.4799,
+      "step": 22530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040633412973066534,
+      "loss": 3.3088,
+      "step": 22535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004062572016522281,
+      "loss": 3.4578,
+      "step": 22540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040618026558379396,
+      "loss": 3.4492,
+      "step": 22545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040610332153114795,
+      "loss": 3.4271,
+      "step": 22550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004060263695000759,
+      "loss": 3.3281,
+      "step": 22555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040594940949636424,
+      "loss": 3.5284,
+      "step": 22560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040587244152579996,
+      "loss": 3.4063,
+      "step": 22565
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040579546559417083,
+      "loss": 3.3343,
+      "step": 22570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040571848170726463,
+      "loss": 3.4706,
+      "step": 22575
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004056414898708706,
+      "loss": 3.4401,
+      "step": 22580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040556449009077777,
+      "loss": 3.4597,
+      "step": 22585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040548748237277644,
+      "loss": 3.4925,
+      "step": 22590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000405410466722657,
+      "loss": 3.2829,
+      "step": 22595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004053334431462106,
+      "loss": 3.4492,
+      "step": 22600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004052564116492292,
+      "loss": 3.315,
+      "step": 22605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004051793722375051,
+      "loss": 3.4348,
+      "step": 22610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004051023249168312,
+      "loss": 3.4744,
+      "step": 22615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004050252696930011,
+      "loss": 3.4817,
+      "step": 22620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004049482065718089,
+      "loss": 3.4592,
+      "step": 22625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004048711355590495,
+      "loss": 3.3313,
+      "step": 22630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040479405666051817,
+      "loss": 3.4485,
+      "step": 22635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040471696988201076,
+      "loss": 3.3691,
+      "step": 22640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040463987522932395,
+      "loss": 3.4079,
+      "step": 22645
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004045627727082548,
+      "loss": 3.4411,
+      "step": 22650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000404485662324601,
+      "loss": 3.4073,
+      "step": 22655
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004044085440841609,
+      "loss": 3.3304,
+      "step": 22660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004043314179927332,
+      "loss": 3.3583,
+      "step": 22665
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004042542840561178,
+      "loss": 3.4128,
+      "step": 22670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004041771422801144,
+      "loss": 3.4874,
+      "step": 22675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040409999267052385,
+      "loss": 3.4342,
+      "step": 22680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004040228352331474,
+      "loss": 3.4489,
+      "step": 22685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040394566997378674,
+      "loss": 3.4437,
+      "step": 22690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040386849689824444,
+      "loss": 3.3225,
+      "step": 22695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004037913160123235,
+      "loss": 3.4923,
+      "step": 22700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004037141273218275,
+      "loss": 3.5339,
+      "step": 22705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004036369308325607,
+      "loss": 3.4442,
+      "step": 22710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004035597265503278,
+      "loss": 3.2786,
+      "step": 22715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004034825144809342,
+      "loss": 3.3097,
+      "step": 22720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004034052946301858,
+      "loss": 3.3156,
+      "step": 22725
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004033280670038893,
+      "loss": 3.4123,
+      "step": 22730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004032508316078517,
+      "loss": 3.3769,
+      "step": 22735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040317358844788064,
+      "loss": 3.4664,
+      "step": 22740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004030963375297845,
+      "loss": 3.4246,
+      "step": 22745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040301907885937217,
+      "loss": 3.4092,
+      "step": 22750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004029418124424531,
+      "loss": 3.4981,
+      "step": 22755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040286453828483723,
+      "loss": 3.3302,
+      "step": 22760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040278725639233525,
+      "loss": 3.3779,
+      "step": 22765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004027099667707584,
+      "loss": 3.4032,
+      "step": 22770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004026326694259185,
+      "loss": 3.4799,
+      "step": 22775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040255536436362767,
+      "loss": 3.4285,
+      "step": 22780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040247805158969915,
+      "loss": 3.4642,
+      "step": 22785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040240073110994637,
+      "loss": 3.4589,
+      "step": 22790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004023234029301833,
+      "loss": 3.4435,
+      "step": 22795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004022460670562248,
+      "loss": 3.3843,
+      "step": 22800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0004021687234938861,
+      "loss": 3.5109,
+      "step": 22805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000402091372248983,
+      "loss": 3.3412,
+      "step": 22810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040201401332733177,
+      "loss": 3.2672,
+      "step": 22815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00040193664673474967,
+      "loss": 3.3403,
+      "step": 22820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004018592724770541,
+      "loss": 3.4237,
+      "step": 22825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00040178189056006334,
+      "loss": 3.2901,
+      "step": 22830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004017045009895959,
+      "loss": 3.5062,
+      "step": 22835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00040162710377147134,
+      "loss": 3.45,
+      "step": 22840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004015496989115093,
+      "loss": 3.435,
+      "step": 22845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004014722864155304,
+      "loss": 3.4186,
+      "step": 22850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004013948662893557,
+      "loss": 3.3804,
+      "step": 22855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004013174385388065,
+      "loss": 3.4171,
+      "step": 22860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004012400031697053,
+      "loss": 3.3592,
+      "step": 22865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00040116256018787465,
+      "loss": 3.4479,
+      "step": 22870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00040108510959913795,
+      "loss": 3.283,
+      "step": 22875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00040100765140931905,
+      "loss": 3.5205,
+      "step": 22880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00040093018562424235,
+      "loss": 3.3975,
+      "step": 22885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00040085271224973303,
+      "loss": 3.4526,
+      "step": 22890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004007752312916167,
+      "loss": 3.4228,
+      "step": 22895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00040069774275571933,
+      "loss": 3.4961,
+      "step": 22900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00040062024664786773,
+      "loss": 3.2809,
+      "step": 22905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004005427429738893,
+      "loss": 3.3984,
+      "step": 22910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00040046523173961195,
+      "loss": 3.3236,
+      "step": 22915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00040038771295086397,
+      "loss": 3.4506,
+      "step": 22920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004003101866134744,
+      "loss": 3.4803,
+      "step": 22925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00040023265273327296,
+      "loss": 3.4333,
+      "step": 22930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004001551113160897,
+      "loss": 3.4086,
+      "step": 22935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00040007756236775543,
+      "loss": 3.329,
+      "step": 22940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0004000000058941012,
+      "loss": 3.4441,
+      "step": 22945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003999224419009591,
+      "loss": 3.4387,
+      "step": 22950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039984487039416153,
+      "loss": 3.3819,
+      "step": 22955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003997672913795413,
+      "loss": 3.3857,
+      "step": 22960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039968970486293213,
+      "loss": 3.4039,
+      "step": 22965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000399612110850168,
+      "loss": 3.4457,
+      "step": 22970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039953450934708366,
+      "loss": 3.4657,
+      "step": 22975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039945690035951434,
+      "loss": 3.3894,
+      "step": 22980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039937928389329575,
+      "loss": 3.5035,
+      "step": 22985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003993016599542644,
+      "loss": 3.3818,
+      "step": 22990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003992240285482571,
+      "loss": 3.3472,
+      "step": 22995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003991463896811114,
+      "loss": 3.3529,
+      "step": 23000
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 3.442578077316284,
+      "eval_runtime": 149.8744,
+      "eval_samples_per_second": 12.284,
+      "eval_steps_per_second": 0.774,
+      "step": 23000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039906874335866535,
+      "loss": 3.3731,
+      "step": 23005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003989910895867574,
+      "loss": 3.4402,
+      "step": 23010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000398913428371227,
+      "loss": 3.4643,
+      "step": 23015
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003988357597179135,
+      "loss": 3.501,
+      "step": 23020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003987580836326575,
+      "loss": 3.4019,
+      "step": 23025
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003986804001212998,
+      "loss": 3.351,
+      "step": 23030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003986027091896817,
+      "loss": 3.313,
+      "step": 23035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003985250108436452,
+      "loss": 3.4688,
+      "step": 23040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003984473050890327,
+      "loss": 3.3642,
+      "step": 23045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003983695919316875,
+      "loss": 3.3882,
+      "step": 23050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039829187137745316,
+      "loss": 3.3845,
+      "step": 23055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003982141434321738,
+      "loss": 3.3129,
+      "step": 23060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039813640810169415,
+      "loss": 3.2984,
+      "step": 23065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003980586653918596,
+      "loss": 3.4537,
+      "step": 23070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039798091530851586,
+      "loss": 3.4673,
+      "step": 23075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003979031578575096,
+      "loss": 3.4474,
+      "step": 23080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003978253930446875,
+      "loss": 3.4899,
+      "step": 23085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039774762087589713,
+      "loss": 3.4094,
+      "step": 23090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003976698413569868,
+      "loss": 3.4574,
+      "step": 23095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039759205449380476,
+      "loss": 3.4646,
+      "step": 23100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003975142602922005,
+      "loss": 3.4299,
+      "step": 23105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003974364587580236,
+      "loss": 3.4025,
+      "step": 23110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003973586498971243,
+      "loss": 3.4642,
+      "step": 23115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039728083371535355,
+      "loss": 3.3419,
+      "step": 23120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003972030102185626,
+      "loss": 3.4246,
+      "step": 23125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039712517941260344,
+      "loss": 3.4522,
+      "step": 23130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003970473413033286,
+      "loss": 3.331,
+      "step": 23135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000396969495896591,
+      "loss": 3.3998,
+      "step": 23140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039689164319824436,
+      "loss": 3.3278,
+      "step": 23145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003968137832141426,
+      "loss": 3.4126,
+      "step": 23150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003967359159501406,
+      "loss": 3.4785,
+      "step": 23155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039665804141209346,
+      "loss": 3.3449,
+      "step": 23160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003965801596058569,
+      "loss": 3.2645,
+      "step": 23165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039650227053728736,
+      "loss": 3.4129,
+      "step": 23170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003964243742122417,
+      "loss": 3.4324,
+      "step": 23175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039634647063657724,
+      "loss": 3.3419,
+      "step": 23180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039626855981615184,
+      "loss": 3.4273,
+      "step": 23185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039619064175682416,
+      "loss": 3.3218,
+      "step": 23190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003961127164644533,
+      "loss": 3.5256,
+      "step": 23195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003960347839448986,
+      "loss": 3.4073,
+      "step": 23200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003959568442040203,
+      "loss": 3.4519,
+      "step": 23205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003958788972476792,
+      "loss": 3.4716,
+      "step": 23210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003958009430817364,
+      "loss": 3.3856,
+      "step": 23215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003957229817120536,
+      "loss": 3.3233,
+      "step": 23220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039564501314449317,
+      "loss": 3.4366,
+      "step": 23225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039556703738491785,
+      "loss": 3.3467,
+      "step": 23230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039548905443919117,
+      "loss": 3.324,
+      "step": 23235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000395411064313177,
+      "loss": 3.3125,
+      "step": 23240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003953330670127397,
+      "loss": 3.3889,
+      "step": 23245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003952550625437443,
+      "loss": 3.4227,
+      "step": 23250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003951770509120565,
+      "loss": 3.454,
+      "step": 23255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039509903212354215,
+      "loss": 3.4451,
+      "step": 23260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039502100618406797,
+      "loss": 3.3362,
+      "step": 23265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003949429730995011,
+      "loss": 3.4359,
+      "step": 23270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003948649328757093,
+      "loss": 3.3884,
+      "step": 23275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039478688551856055,
+      "loss": 3.4077,
+      "step": 23280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039470883103392393,
+      "loss": 3.3771,
+      "step": 23285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039463076942766853,
+      "loss": 3.4804,
+      "step": 23290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003945527007056644,
+      "loss": 3.3656,
+      "step": 23295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003944746248737816,
+      "loss": 3.4062,
+      "step": 23300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003943965419378911,
+      "loss": 3.3631,
+      "step": 23305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039431845190386454,
+      "loss": 3.4399,
+      "step": 23310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003942403547775738,
+      "loss": 3.4739,
+      "step": 23315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003941622505648913,
+      "loss": 3.3574,
+      "step": 23320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039408413927169016,
+      "loss": 3.463,
+      "step": 23325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039400602090384384,
+      "loss": 3.3802,
+      "step": 23330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003939278954672267,
+      "loss": 3.4331,
+      "step": 23335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039384976296771303,
+      "loss": 3.5204,
+      "step": 23340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039377162341117816,
+      "loss": 3.4733,
+      "step": 23345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039369347680349786,
+      "loss": 3.3425,
+      "step": 23350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039361532315054825,
+      "loss": 3.4375,
+      "step": 23355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039353716245820613,
+      "loss": 3.4392,
+      "step": 23360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003934589947323487,
+      "loss": 3.3395,
+      "step": 23365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039338081997885384,
+      "loss": 3.3579,
+      "step": 23370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003933026382036,
+      "loss": 3.4948,
+      "step": 23375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003932244494124658,
+      "loss": 3.4524,
+      "step": 23380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039314625361133086,
+      "loss": 3.4354,
+      "step": 23385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039306805080607495,
+      "loss": 3.3287,
+      "step": 23390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003929898410025786,
+      "loss": 3.3441,
+      "step": 23395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039291162420672284,
+      "loss": 3.3456,
+      "step": 23400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039283340042438905,
+      "loss": 3.3949,
+      "step": 23405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003927551696614593,
+      "loss": 3.3505,
+      "step": 23410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003926769319238162,
+      "loss": 3.4162,
+      "step": 23415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039259868721734274,
+      "loss": 3.3707,
+      "step": 23420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003925204355479226,
+      "loss": 3.3314,
+      "step": 23425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039244217692143975,
+      "loss": 3.4314,
+      "step": 23430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003923639113437791,
+      "loss": 3.4921,
+      "step": 23435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039228563882082564,
+      "loss": 3.3425,
+      "step": 23440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000392207359358465,
+      "loss": 3.3939,
+      "step": 23445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039212907296258363,
+      "loss": 3.391,
+      "step": 23450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039205077963906806,
+      "loss": 3.3738,
+      "step": 23455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003919724793938056,
+      "loss": 3.5164,
+      "step": 23460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039189417223268406,
+      "loss": 3.3694,
+      "step": 23465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039181585816159176,
+      "loss": 3.354,
+      "step": 23470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003917375371864174,
+      "loss": 3.4643,
+      "step": 23475
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003916592093130505,
+      "loss": 3.3646,
+      "step": 23480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039158087454738063,
+      "loss": 3.38,
+      "step": 23485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003915025328952984,
+      "loss": 3.5276,
+      "step": 23490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039142418436269473,
+      "loss": 3.5326,
+      "step": 23495
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039134582895546094,
+      "loss": 3.4432,
+      "step": 23500
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 3.4460275173187256,
+      "eval_runtime": 149.7826,
+      "eval_samples_per_second": 12.291,
+      "eval_steps_per_second": 0.774,
+      "step": 23500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003912674666794889,
+      "loss": 3.5093,
+      "step": 23505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039118909754067097,
+      "loss": 3.4232,
+      "step": 23510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003911107215449005,
+      "loss": 3.3742,
+      "step": 23515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039103233869807055,
+      "loss": 3.5677,
+      "step": 23520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039095394900607537,
+      "loss": 3.412,
+      "step": 23525
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003908755524748092,
+      "loss": 3.4859,
+      "step": 23530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039079714911016727,
+      "loss": 3.3516,
+      "step": 23535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003907187389180451,
+      "loss": 3.4358,
+      "step": 23540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003906403219043386,
+      "loss": 3.4573,
+      "step": 23545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003905618980749445,
+      "loss": 3.4623,
+      "step": 23550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039048346743575976,
+      "loss": 3.5096,
+      "step": 23555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039040502999268203,
+      "loss": 3.3192,
+      "step": 23560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003903265857516094,
+      "loss": 3.4335,
+      "step": 23565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039024813471844034,
+      "loss": 3.4312,
+      "step": 23570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039016967689907406,
+      "loss": 3.3351,
+      "step": 23575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039009121229941027,
+      "loss": 3.4376,
+      "step": 23580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003900127409253489,
+      "loss": 3.378,
+      "step": 23585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003899342627827909,
+      "loss": 3.4918,
+      "step": 23590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038985577787763713,
+      "loss": 3.5062,
+      "step": 23595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003897772862157893,
+      "loss": 3.3182,
+      "step": 23600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038969878780314976,
+      "loss": 3.4342,
+      "step": 23605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000389620282645621,
+      "loss": 3.3611,
+      "step": 23610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003895417707491064,
+      "loss": 3.2832,
+      "step": 23615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038946325211950945,
+      "loss": 3.4133,
+      "step": 23620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003893847267627344,
+      "loss": 3.3847,
+      "step": 23625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003893061946846861,
+      "loss": 3.2987,
+      "step": 23630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038922765589126954,
+      "loss": 3.3355,
+      "step": 23635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038914911038839063,
+      "loss": 3.2932,
+      "step": 23640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003890705581819554,
+      "loss": 3.4283,
+      "step": 23645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038899199927787067,
+      "loss": 3.3789,
+      "step": 23650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038891343368204374,
+      "loss": 3.4124,
+      "step": 23655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003888348614003823,
+      "loss": 3.388,
+      "step": 23660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003887562824387945,
+      "loss": 3.4347,
+      "step": 23665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038867769680318913,
+      "loss": 3.4287,
+      "step": 23670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003885991044994755,
+      "loss": 3.4319,
+      "step": 23675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003885205055335632,
+      "loss": 3.3868,
+      "step": 23680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003884418999113625,
+      "loss": 3.4081,
+      "step": 23685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038836328763878423,
+      "loss": 3.5162,
+      "step": 23690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003882846687217396,
+      "loss": 3.439,
+      "step": 23695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038820604316614036,
+      "loss": 3.4161,
+      "step": 23700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003881274109778987,
+      "loss": 3.408,
+      "step": 23705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003880487721629273,
+      "loss": 3.3882,
+      "step": 23710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003879701267271396,
+      "loss": 3.2077,
+      "step": 23715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038789147467644915,
+      "loss": 3.346,
+      "step": 23720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038781281601677023,
+      "loss": 3.4345,
+      "step": 23725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003877341507540176,
+      "loss": 3.2781,
+      "step": 23730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038765547889410643,
+      "loss": 3.3699,
+      "step": 23735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003875768004429525,
+      "loss": 3.4467,
+      "step": 23740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038749811540647204,
+      "loss": 3.3993,
+      "step": 23745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038741942379058163,
+      "loss": 3.3744,
+      "step": 23750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038734072560119866,
+      "loss": 3.4018,
+      "step": 23755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038726202084424076,
+      "loss": 3.412,
+      "step": 23760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000387183309525626,
+      "loss": 3.4137,
+      "step": 23765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003871045916512732,
+      "loss": 3.4427,
+      "step": 23770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038702586722710153,
+      "loss": 3.4318,
+      "step": 23775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003869471362590306,
+      "loss": 3.4418,
+      "step": 23780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038686839875298073,
+      "loss": 3.3053,
+      "step": 23785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003867896547148724,
+      "loss": 3.4195,
+      "step": 23790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038671090415062683,
+      "loss": 3.343,
+      "step": 23795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038663214706616565,
+      "loss": 3.4823,
+      "step": 23800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000386553383467411,
+      "loss": 3.2957,
+      "step": 23805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003864746133602855,
+      "loss": 3.3719,
+      "step": 23810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003863958367507122,
+      "loss": 3.3919,
+      "step": 23815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038631705364461485,
+      "loss": 3.3362,
+      "step": 23820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038623826404791733,
+      "loss": 3.4481,
+      "step": 23825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003861594679665443,
+      "loss": 3.4172,
+      "step": 23830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038608066540642095,
+      "loss": 3.3761,
+      "step": 23835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003860018563734727,
+      "loss": 3.4139,
+      "step": 23840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003859230408736256,
+      "loss": 3.4002,
+      "step": 23845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038584421891280606,
+      "loss": 3.3431,
+      "step": 23850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003857653904969413,
+      "loss": 3.4767,
+      "step": 23855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038568655563195873,
+      "loss": 3.3485,
+      "step": 23860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003856077143237863,
+      "loss": 3.4446,
+      "step": 23865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003855288665783525,
+      "loss": 3.4077,
+      "step": 23870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003854500124015861,
+      "loss": 3.4456,
+      "step": 23875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038537115179941686,
+      "loss": 3.4435,
+      "step": 23880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003852922847777745,
+      "loss": 3.3529,
+      "step": 23885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038521341134258947,
+      "loss": 3.3197,
+      "step": 23890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003851345314997926,
+      "loss": 3.3096,
+      "step": 23895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003850556452553152,
+      "loss": 3.4025,
+      "step": 23900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003849767526150892,
+      "loss": 3.3826,
+      "step": 23905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000384897853585047,
+      "loss": 3.3757,
+      "step": 23910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038481894817112114,
+      "loss": 3.4822,
+      "step": 23915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038474003637924516,
+      "loss": 3.3612,
+      "step": 23920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003846611182153528,
+      "loss": 3.4059,
+      "step": 23925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038458219368537813,
+      "loss": 3.2958,
+      "step": 23930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000384503262795256,
+      "loss": 3.3191,
+      "step": 23935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038442432555092163,
+      "loss": 3.4621,
+      "step": 23940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038434538195831054,
+      "loss": 3.3922,
+      "step": 23945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038426643202335906,
+      "loss": 3.3505,
+      "step": 23950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038418747575200374,
+      "loss": 3.4525,
+      "step": 23955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003841085131501818,
+      "loss": 3.4467,
+      "step": 23960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003840295442238306,
+      "loss": 3.4365,
+      "step": 23965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038395056897888824,
+      "loss": 3.471,
+      "step": 23970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00038387158742129345,
+      "loss": 3.3748,
+      "step": 23975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038379259955698516,
+      "loss": 3.4801,
+      "step": 23980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003837136053919028,
+      "loss": 3.3963,
+      "step": 23985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003836346049319862,
+      "loss": 3.4244,
+      "step": 23990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000383555598183176,
+      "loss": 3.469,
+      "step": 23995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003834765851514131,
+      "loss": 3.4334,
+      "step": 24000
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 3.4270379543304443,
+      "eval_runtime": 150.6758,
+      "eval_samples_per_second": 12.218,
+      "eval_steps_per_second": 0.77,
+      "step": 24000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038339756584263875,
+      "loss": 3.3244,
+      "step": 24005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038331854026279485,
+      "loss": 3.3813,
+      "step": 24010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038323950841782375,
+      "loss": 3.3714,
+      "step": 24015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003831604703136683,
+      "loss": 3.3798,
+      "step": 24020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003830814259562715,
+      "loss": 3.3194,
+      "step": 24025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003830023753515773,
+      "loss": 3.4367,
+      "step": 24030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038292331850552986,
+      "loss": 3.4029,
+      "step": 24035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038284425542407395,
+      "loss": 3.4108,
+      "step": 24040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003827651861131545,
+      "loss": 3.3577,
+      "step": 24045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038268611057871727,
+      "loss": 3.4693,
+      "step": 24050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003826070288267082,
+      "loss": 3.3681,
+      "step": 24055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038252794086307405,
+      "loss": 3.4132,
+      "step": 24060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003824488466937616,
+      "loss": 3.4845,
+      "step": 24065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003823697463247184,
+      "loss": 3.4444,
+      "step": 24070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038229063976189247,
+      "loss": 3.4076,
+      "step": 24075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038221152701123216,
+      "loss": 3.3896,
+      "step": 24080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038213240807868633,
+      "loss": 3.3713,
+      "step": 24085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003820532829702043,
+      "loss": 3.3014,
+      "step": 24090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003819741516917359,
+      "loss": 3.4203,
+      "step": 24095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038189501424923156,
+      "loss": 3.3145,
+      "step": 24100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003818158706486417,
+      "loss": 3.36,
+      "step": 24105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003817367208959177,
+      "loss": 3.3946,
+      "step": 24110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003816575649970112,
+      "loss": 3.3836,
+      "step": 24115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003815784029578743,
+      "loss": 3.4324,
+      "step": 24120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003814992347844595,
+      "loss": 3.4453,
+      "step": 24125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003814200604827201,
+      "loss": 3.4197,
+      "step": 24130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038134088005860924,
+      "loss": 3.4852,
+      "step": 24135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038126169351808127,
+      "loss": 3.408,
+      "step": 24140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003811825008670902,
+      "loss": 3.4374,
+      "step": 24145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003811033021115912,
+      "loss": 3.4174,
+      "step": 24150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038102409725753947,
+      "loss": 3.4054,
+      "step": 24155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000380944886310891,
+      "loss": 3.3431,
+      "step": 24160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038086566927760185,
+      "loss": 3.5198,
+      "step": 24165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038078644616362876,
+      "loss": 3.3664,
+      "step": 24170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038070721697492895,
+      "loss": 3.3576,
+      "step": 24175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003806279817174601,
+      "loss": 3.4815,
+      "step": 24180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003805487403971802,
+      "loss": 3.4107,
+      "step": 24185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038046949302004783,
+      "loss": 3.3978,
+      "step": 24190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038039023959202193,
+      "loss": 3.3128,
+      "step": 24195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003803109801190621,
+      "loss": 3.4749,
+      "step": 24200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003802317146071281,
+      "loss": 3.3715,
+      "step": 24205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038015244306218034,
+      "loss": 3.3911,
+      "step": 24210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00038007316549017964,
+      "loss": 3.364,
+      "step": 24215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037999388189708733,
+      "loss": 3.4206,
+      "step": 24220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037991459228886495,
+      "loss": 3.4327,
+      "step": 24225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003798352966714749,
+      "loss": 3.3105,
+      "step": 24230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003797559950508796,
+      "loss": 3.4319,
+      "step": 24235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037967668743304225,
+      "loss": 3.3862,
+      "step": 24240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037959737382392637,
+      "loss": 3.3361,
+      "step": 24245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003795180542294958,
+      "loss": 3.4328,
+      "step": 24250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037943872865571525,
+      "loss": 3.4733,
+      "step": 24255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037935939710854934,
+      "loss": 3.3578,
+      "step": 24260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003792800595939634,
+      "loss": 3.3603,
+      "step": 24265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037920071611792337,
+      "loss": 3.4714,
+      "step": 24270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003791213666863953,
+      "loss": 3.4574,
+      "step": 24275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037904201130534604,
+      "loss": 3.3614,
+      "step": 24280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003789626499807426,
+      "loss": 3.3123,
+      "step": 24285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003788832827185525,
+      "loss": 3.3467,
+      "step": 24290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037880390952474386,
+      "loss": 3.4375,
+      "step": 24295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037872453040528516,
+      "loss": 3.3211,
+      "step": 24300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003786451453661451,
+      "loss": 3.3171,
+      "step": 24305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037856575441329315,
+      "loss": 3.43,
+      "step": 24310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003784863575526992,
+      "loss": 3.36,
+      "step": 24315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037840695479033335,
+      "loss": 3.4438,
+      "step": 24320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003783275461321663,
+      "loss": 3.323,
+      "step": 24325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037824813158416937,
+      "loss": 3.4666,
+      "step": 24330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037816871115231376,
+      "loss": 3.4379,
+      "step": 24335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037808928484257175,
+      "loss": 3.501,
+      "step": 24340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003780098526609158,
+      "loss": 3.4066,
+      "step": 24345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037793041461331857,
+      "loss": 3.2325,
+      "step": 24350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003778509707057536,
+      "loss": 3.3971,
+      "step": 24355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037777152094419474,
+      "loss": 3.384,
+      "step": 24360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003776920653346159,
+      "loss": 3.3162,
+      "step": 24365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037761260388299205,
+      "loss": 3.4612,
+      "step": 24370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037753313659529805,
+      "loss": 3.4324,
+      "step": 24375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003774536634775097,
+      "loss": 3.4222,
+      "step": 24380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003773741845356026,
+      "loss": 3.3924,
+      "step": 24385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003772946997755534,
+      "loss": 3.4305,
+      "step": 24390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003772152092033389,
+      "loss": 3.402,
+      "step": 24395
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003771357128249365,
+      "loss": 3.5204,
+      "step": 24400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003770562106463237,
+      "loss": 3.3448,
+      "step": 24405
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003769767026734788,
+      "loss": 3.2582,
+      "step": 24410
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003768971889123803,
+      "loss": 3.4989,
+      "step": 24415
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003768176693690074,
+      "loss": 3.2957,
+      "step": 24420
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003767381440493394,
+      "loss": 3.4028,
+      "step": 24425
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003766586129593562,
+      "loss": 3.4257,
+      "step": 24430
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003765790761050383,
+      "loss": 3.4178,
+      "step": 24435
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003764995334923662,
+      "loss": 3.4097,
+      "step": 24440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003764199851273213,
+      "loss": 3.4026,
+      "step": 24445
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003763404310158851,
+      "loss": 3.4466,
+      "step": 24450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037626087116403973,
+      "loss": 3.3257,
+      "step": 24455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037618130557776774,
+      "loss": 3.4463,
+      "step": 24460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000376101734263052,
+      "loss": 3.2616,
+      "step": 24465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037602215722587587,
+      "loss": 3.3278,
+      "step": 24470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003759425744722231,
+      "loss": 3.3324,
+      "step": 24475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037586298600807795,
+      "loss": 3.3937,
+      "step": 24480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037578339183942503,
+      "loss": 3.3143,
+      "step": 24485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037570379197224935,
+      "loss": 3.3917,
+      "step": 24490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003756241864125365,
+      "loss": 3.3902,
+      "step": 24495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003755445751662725,
+      "loss": 3.4043,
+      "step": 24500
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 3.4218356609344482,
+      "eval_runtime": 149.9736,
+      "eval_samples_per_second": 12.275,
+      "eval_steps_per_second": 0.773,
+      "step": 24500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003754649582394435,
+      "loss": 3.3665,
+      "step": 24505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003753853356380365,
+      "loss": 3.4224,
+      "step": 24510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037530570736803853,
+      "loss": 3.2309,
+      "step": 24515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003752260734354374,
+      "loss": 3.32,
+      "step": 24520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000375146433846221,
+      "loss": 3.3945,
+      "step": 24525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003750667886063778,
+      "loss": 3.4484,
+      "step": 24530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037498713772189695,
+      "loss": 3.3952,
+      "step": 24535
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003749074811987676,
+      "loss": 3.4158,
+      "step": 24540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037482781904297954,
+      "loss": 3.45,
+      "step": 24545
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00037474815126052304,
+      "loss": 3.3781,
+      "step": 24550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003746684778573885,
+      "loss": 3.285,
+      "step": 24555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003745887988395672,
+      "loss": 3.3847,
+      "step": 24560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037450911421305056,
+      "loss": 3.4632,
+      "step": 24565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037442942398383027,
+      "loss": 3.364,
+      "step": 24570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003743497281578987,
+      "loss": 3.3277,
+      "step": 24575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003742700267412487,
+      "loss": 3.4421,
+      "step": 24580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037419031973987335,
+      "loss": 3.3185,
+      "step": 24585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037411060715976613,
+      "loss": 3.45,
+      "step": 24590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000374030889006921,
+      "loss": 3.4009,
+      "step": 24595
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003739511652873325,
+      "loss": 3.4462,
+      "step": 24600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003738714360069954,
+      "loss": 3.2398,
+      "step": 24605
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037379170117190477,
+      "loss": 3.4085,
+      "step": 24610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003737119607880565,
+      "loss": 3.4324,
+      "step": 24615
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037363221486144644,
+      "loss": 3.4352,
+      "step": 24620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003735524633980713,
+      "loss": 3.451,
+      "step": 24625
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037347270640392783,
+      "loss": 3.3561,
+      "step": 24630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003733929438850133,
+      "loss": 3.3655,
+      "step": 24635
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003733131758473256,
+      "loss": 3.4145,
+      "step": 24640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003732334022968628,
+      "loss": 3.4254,
+      "step": 24645
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003731536232396235,
+      "loss": 3.341,
+      "step": 24650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003730738386816065,
+      "loss": 3.2968,
+      "step": 24655
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037299404862881145,
+      "loss": 3.3911,
+      "step": 24660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000372914253087238,
+      "loss": 3.3917,
+      "step": 24665
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037283445206288637,
+      "loss": 3.3416,
+      "step": 24670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037275464556175725,
+      "loss": 3.3614,
+      "step": 24675
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003726748335898516,
+      "loss": 3.3741,
+      "step": 24680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037259501615317096,
+      "loss": 3.343,
+      "step": 24685
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003725151932577172,
+      "loss": 3.3623,
+      "step": 24690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003724353649094924,
+      "loss": 3.4274,
+      "step": 24695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003723555311144995,
+      "loss": 3.3472,
+      "step": 24700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037227569187874146,
+      "loss": 3.4346,
+      "step": 24705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003721958472082217,
+      "loss": 3.5295,
+      "step": 24710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003721159971089443,
+      "loss": 3.5022,
+      "step": 24715
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037203614158691354,
+      "loss": 3.3116,
+      "step": 24720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003719562806481341,
+      "loss": 3.2844,
+      "step": 24725
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003718764142986111,
+      "loss": 3.4344,
+      "step": 24730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003717965425443501,
+      "loss": 3.4131,
+      "step": 24735
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037171666539135705,
+      "loss": 3.3643,
+      "step": 24740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037163678284563836,
+      "loss": 3.2979,
+      "step": 24745
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037155689491320066,
+      "loss": 3.358,
+      "step": 24750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037147700160005125,
+      "loss": 3.4448,
+      "step": 24755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003713971029121975,
+      "loss": 3.3963,
+      "step": 24760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003713171988556478,
+      "loss": 3.2131,
+      "step": 24765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003712372894364099,
+      "loss": 3.392,
+      "step": 24770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037115737466049306,
+      "loss": 3.3659,
+      "step": 24775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003710774545339064,
+      "loss": 3.3049,
+      "step": 24780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037099752906265943,
+      "loss": 3.3554,
+      "step": 24785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037091759825276207,
+      "loss": 3.318,
+      "step": 24790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037083766211022467,
+      "loss": 3.2913,
+      "step": 24795
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037075772064105815,
+      "loss": 3.4315,
+      "step": 24800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037067777385127383,
+      "loss": 3.3689,
+      "step": 24805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037059782174688296,
+      "loss": 3.4208,
+      "step": 24810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037051786433389775,
+      "loss": 3.2799,
+      "step": 24815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003704379016183306,
+      "loss": 3.5046,
+      "step": 24820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037035793360619423,
+      "loss": 3.4735,
+      "step": 24825
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00037027796030350183,
+      "loss": 3.3752,
+      "step": 24830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003701979817162669,
+      "loss": 3.4892,
+      "step": 24835
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003701179978505035,
+      "loss": 3.2982,
+      "step": 24840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003700380087122262,
+      "loss": 3.5165,
+      "step": 24845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003699580143074494,
+      "loss": 3.3722,
+      "step": 24850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036987801464218853,
+      "loss": 3.456,
+      "step": 24855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000369798009722459,
+      "loss": 3.3965,
+      "step": 24860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036971799955427695,
+      "loss": 3.4197,
+      "step": 24865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003696379841436586,
+      "loss": 3.3727,
+      "step": 24870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036955796349662066,
+      "loss": 3.3547,
+      "step": 24875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036947793761918043,
+      "loss": 3.3268,
+      "step": 24880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036939790651735536,
+      "loss": 3.4494,
+      "step": 24885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003693178701971633,
+      "loss": 3.3767,
+      "step": 24890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003692378286646227,
+      "loss": 3.4077,
+      "step": 24895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036915778192575223,
+      "loss": 3.4866,
+      "step": 24900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036907772998657107,
+      "loss": 3.2929,
+      "step": 24905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036899767285309846,
+      "loss": 3.5017,
+      "step": 24910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003689176105313545,
+      "loss": 3.3838,
+      "step": 24915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003688375430273595,
+      "loss": 3.2769,
+      "step": 24920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000368757470347134,
+      "loss": 3.3815,
+      "step": 24925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003686773924966992,
+      "loss": 3.381,
+      "step": 24930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036859730948207627,
+      "loss": 3.3977,
+      "step": 24935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003685172213092873,
+      "loss": 3.5147,
+      "step": 24940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003684371279843546,
+      "loss": 3.3731,
+      "step": 24945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003683570295133004,
+      "loss": 3.4501,
+      "step": 24950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003682769259021479,
+      "loss": 3.3245,
+      "step": 24955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036819681715692063,
+      "loss": 3.4224,
+      "step": 24960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036811670328364214,
+      "loss": 3.3934,
+      "step": 24965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003680365842883367,
+      "loss": 3.3077,
+      "step": 24970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003679564601770287,
+      "loss": 3.3332,
+      "step": 24975
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003678763309557432,
+      "loss": 3.3636,
+      "step": 24980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036779619663050554,
+      "loss": 3.3194,
+      "step": 24985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036771605720734127,
+      "loss": 3.487,
+      "step": 24990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036763591269227663,
+      "loss": 3.3416,
+      "step": 24995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003675557630913379,
+      "loss": 3.3194,
+      "step": 25000
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 3.4098989963531494,
+      "eval_runtime": 149.8725,
+      "eval_samples_per_second": 12.284,
+      "eval_steps_per_second": 0.774,
+      "step": 25000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036747560841055206,
+      "loss": 3.4664,
+      "step": 25005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036739544865594623,
+      "loss": 3.3433,
+      "step": 25010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036731528383354803,
+      "loss": 3.3198,
+      "step": 25015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003672351139493855,
+      "loss": 3.3991,
+      "step": 25020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036715493900948704,
+      "loss": 3.319,
+      "step": 25025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036707475901988124,
+      "loss": 3.3713,
+      "step": 25030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003669945739865974,
+      "loss": 3.34,
+      "step": 25035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003669143839156648,
+      "loss": 3.4226,
+      "step": 25040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036683418881311355,
+      "loss": 3.3767,
+      "step": 25045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036675398868497394,
+      "loss": 3.4798,
+      "step": 25050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003666737835372763,
+      "loss": 3.3018,
+      "step": 25055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003665935733760519,
+      "loss": 3.4508,
+      "step": 25060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036651335820733213,
+      "loss": 3.351,
+      "step": 25065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003664331380371486,
+      "loss": 3.4597,
+      "step": 25070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036635291287153365,
+      "loss": 3.3536,
+      "step": 25075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036627268271651955,
+      "loss": 3.4136,
+      "step": 25080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036619244757813953,
+      "loss": 3.24,
+      "step": 25085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003661122074624266,
+      "loss": 3.3397,
+      "step": 25090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036603196237541445,
+      "loss": 3.3929,
+      "step": 25095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036595171232313715,
+      "loss": 3.3473,
+      "step": 25100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003658714573116292,
+      "loss": 3.3839,
+      "step": 25105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003657911973469252,
+      "loss": 3.3912,
+      "step": 25110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003657109324350602,
+      "loss": 3.2797,
+      "step": 25115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003656306625820699,
+      "loss": 3.3835,
+      "step": 25120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003655503877939902,
+      "loss": 3.3576,
+      "step": 25125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00036547010807685726,
+      "loss": 3.4066,
+      "step": 25130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003653898234367078,
+      "loss": 3.4692,
+      "step": 25135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003653095338795786,
+      "loss": 3.3662,
+      "step": 25140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003652292394115072,
+      "loss": 3.2443,
+      "step": 25145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036514894003853126,
+      "loss": 3.4306,
+      "step": 25150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036506863576668894,
+      "loss": 3.3949,
+      "step": 25155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003649883266020187,
+      "loss": 3.3633,
+      "step": 25160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003649080125505593,
+      "loss": 3.4211,
+      "step": 25165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036482769361835,
+      "loss": 3.3037,
+      "step": 25170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003647473698114305,
+      "loss": 3.2987,
+      "step": 25175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003646670411358405,
+      "loss": 3.32,
+      "step": 25180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036458670759762044,
+      "loss": 3.3066,
+      "step": 25185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000364506369202811,
+      "loss": 3.446,
+      "step": 25190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036442602595745314,
+      "loss": 3.4656,
+      "step": 25195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036434567786758836,
+      "loss": 3.3859,
+      "step": 25200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003642653249392584,
+      "loss": 3.363,
+      "step": 25205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003641849671785053,
+      "loss": 3.3929,
+      "step": 25210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003641046045913716,
+      "loss": 3.3327,
+      "step": 25215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003640242371839002,
+      "loss": 3.4705,
+      "step": 25220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036394386496213423,
+      "loss": 3.4551,
+      "step": 25225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036386348793211737,
+      "loss": 3.4754,
+      "step": 25230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003637831060998935,
+      "loss": 3.341,
+      "step": 25235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036370271947150693,
+      "loss": 3.4458,
+      "step": 25240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003636223280530024,
+      "loss": 3.3916,
+      "step": 25245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003635419318504248,
+      "loss": 3.3774,
+      "step": 25250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036346153086981955,
+      "loss": 3.3418,
+      "step": 25255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036338112511723235,
+      "loss": 3.2995,
+      "step": 25260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036330071459870946,
+      "loss": 3.318,
+      "step": 25265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036322029932029726,
+      "loss": 3.3669,
+      "step": 25270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036313987928804256,
+      "loss": 3.4883,
+      "step": 25275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036305945450799243,
+      "loss": 3.3465,
+      "step": 25280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003629790249861946,
+      "loss": 3.3789,
+      "step": 25285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036289859072869687,
+      "loss": 3.408,
+      "step": 25290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003628181517415473,
+      "loss": 3.3993,
+      "step": 25295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003627377080307947,
+      "loss": 3.323,
+      "step": 25300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036265725960248806,
+      "loss": 3.3462,
+      "step": 25305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003625768064626766,
+      "loss": 3.3265,
+      "step": 25310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003624963486174101,
+      "loss": 3.2759,
+      "step": 25315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036241588607273823,
+      "loss": 3.4092,
+      "step": 25320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003623354188347118,
+      "loss": 3.2942,
+      "step": 25325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036225494690938126,
+      "loss": 3.3437,
+      "step": 25330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036217447030279773,
+      "loss": 3.3779,
+      "step": 25335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003620939890210128,
+      "loss": 3.3958,
+      "step": 25340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036201350307007795,
+      "loss": 3.438,
+      "step": 25345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003619330124560456,
+      "loss": 3.4422,
+      "step": 25350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003618525171849681,
+      "loss": 3.3307,
+      "step": 25355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003617720172628982,
+      "loss": 3.4132,
+      "step": 25360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003616915126958893,
+      "loss": 3.3374,
+      "step": 25365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003616110034899948,
+      "loss": 3.3578,
+      "step": 25370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003615304896512685,
+      "loss": 3.4068,
+      "step": 25375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003614499711857648,
+      "loss": 3.2033,
+      "step": 25380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036136944809953816,
+      "loss": 3.3313,
+      "step": 25385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003612889203986436,
+      "loss": 3.2963,
+      "step": 25390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036120838808913625,
+      "loss": 3.3687,
+      "step": 25395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003611278511770718,
+      "loss": 3.3163,
+      "step": 25400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036104730966850625,
+      "loss": 3.4108,
+      "step": 25405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003609667635694959,
+      "loss": 3.4129,
+      "step": 25410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003608862128860973,
+      "loss": 3.3316,
+      "step": 25415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036080565762436757,
+      "loss": 3.2811,
+      "step": 25420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036072509779036397,
+      "loss": 3.4021,
+      "step": 25425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036064453339014436,
+      "loss": 3.4404,
+      "step": 25430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036056396442976657,
+      "loss": 3.4088,
+      "step": 25435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036048339091528895,
+      "loss": 3.3295,
+      "step": 25440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003604028128527704,
+      "loss": 3.4301,
+      "step": 25445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036032223024826983,
+      "loss": 3.3048,
+      "step": 25450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003602416431078468,
+      "loss": 3.2633,
+      "step": 25455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003601610514375608,
+      "loss": 3.3771,
+      "step": 25460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00036008045524347213,
+      "loss": 3.3616,
+      "step": 25465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035999985453164115,
+      "loss": 3.3544,
+      "step": 25470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035991924930812855,
+      "loss": 3.3366,
+      "step": 25475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003598386395789955,
+      "loss": 3.405,
+      "step": 25480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003597580253503034,
+      "loss": 3.3767,
+      "step": 25485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003596774066281141,
+      "loss": 3.2993,
+      "step": 25490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035959678341848964,
+      "loss": 3.4348,
+      "step": 25495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035951615572749247,
+      "loss": 3.2641,
+      "step": 25500
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 3.403398036956787,
+      "eval_runtime": 149.6768,
+      "eval_samples_per_second": 12.3,
+      "eval_steps_per_second": 0.775,
+      "step": 25500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035943552356118547,
+      "loss": 3.3219,
+      "step": 25505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035935488692563167,
+      "loss": 3.3383,
+      "step": 25510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003592742458268945,
+      "loss": 3.3004,
+      "step": 25515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035919360027103785,
+      "loss": 3.2707,
+      "step": 25520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003591129502641259,
+      "loss": 3.3976,
+      "step": 25525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035903229581222293,
+      "loss": 3.3484,
+      "step": 25530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000358951636921394,
+      "loss": 3.3598,
+      "step": 25535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035887097359770393,
+      "loss": 3.258,
+      "step": 25540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003587903058472184,
+      "loss": 3.3335,
+      "step": 25545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035870963367600324,
+      "loss": 3.3303,
+      "step": 25550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035862895709012443,
+      "loss": 3.4024,
+      "step": 25555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003585482760956486,
+      "loss": 3.3626,
+      "step": 25560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003584675906986423,
+      "loss": 3.3502,
+      "step": 25565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035838690090517303,
+      "loss": 3.4236,
+      "step": 25570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003583062067213079,
+      "loss": 3.3696,
+      "step": 25575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035822550815311485,
+      "loss": 3.3772,
+      "step": 25580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000358144805206662,
+      "loss": 3.445,
+      "step": 25585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035806409788801785,
+      "loss": 3.4372,
+      "step": 25590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003579833862032511,
+      "loss": 3.4523,
+      "step": 25595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035790267015843076,
+      "loss": 3.3395,
+      "step": 25600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003578219497596264,
+      "loss": 3.3512,
+      "step": 25605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003577412250129079,
+      "loss": 3.3397,
+      "step": 25610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003576604959243451,
+      "loss": 3.3981,
+      "step": 25615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035757976250000846,
+      "loss": 3.3503,
+      "step": 25620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003574990247459688,
+      "loss": 3.3889,
+      "step": 25625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003574182826682972,
+      "loss": 3.3094,
+      "step": 25630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035733753627306496,
+      "loss": 3.3504,
+      "step": 25635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003572567855663439,
+      "loss": 3.3083,
+      "step": 25640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003571760305542059,
+      "loss": 3.4901,
+      "step": 25645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003570952712427235,
+      "loss": 3.3134,
+      "step": 25650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003570145076379693,
+      "loss": 3.4654,
+      "step": 25655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003569337397460164,
+      "loss": 3.347,
+      "step": 25660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003568529675729379,
+      "loss": 3.4081,
+      "step": 25665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035677219112480765,
+      "loss": 3.381,
+      "step": 25670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003566914104076996,
+      "loss": 3.2434,
+      "step": 25675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035661062542768804,
+      "loss": 3.4147,
+      "step": 25680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035652983619084754,
+      "loss": 3.3619,
+      "step": 25685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035644904270325315,
+      "loss": 3.3556,
+      "step": 25690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035636824497098,
+      "loss": 3.362,
+      "step": 25695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035628744300010367,
+      "loss": 3.3565,
+      "step": 25700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00035620663679670014,
+      "loss": 3.3171,
+      "step": 25705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003561258263668456,
+      "loss": 3.3843,
+      "step": 25710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003560450117166165,
+      "loss": 3.4695,
+      "step": 25715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003559641928520897,
+      "loss": 3.432,
+      "step": 25720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003558833697793425,
+      "loss": 3.4059,
+      "step": 25725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035580254250445234,
+      "loss": 3.3132,
+      "step": 25730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003557217110334969,
+      "loss": 3.4028,
+      "step": 25735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003556408753725543,
+      "loss": 3.3004,
+      "step": 25740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000355560035527703,
+      "loss": 3.3349,
+      "step": 25745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000355479191505022,
+      "loss": 3.3751,
+      "step": 25750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035539834331058996,
+      "loss": 3.3602,
+      "step": 25755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035531749095048633,
+      "loss": 3.3683,
+      "step": 25760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035523663443079104,
+      "loss": 3.4222,
+      "step": 25765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035515577375758384,
+      "loss": 3.3746,
+      "step": 25770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035507490893694517,
+      "loss": 3.3384,
+      "step": 25775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035499403997495555,
+      "loss": 3.3536,
+      "step": 25780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035491316687769596,
+      "loss": 3.3968,
+      "step": 25785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003548322896512477,
+      "loss": 3.3728,
+      "step": 25790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035475140830169225,
+      "loss": 3.2715,
+      "step": 25795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003546705228351115,
+      "loss": 3.4149,
+      "step": 25800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003545896332575876,
+      "loss": 3.4619,
+      "step": 25805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000354508739575203,
+      "loss": 3.3945,
+      "step": 25810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003544278417940406,
+      "loss": 3.204,
+      "step": 25815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003543469399201834,
+      "loss": 3.3092,
+      "step": 25820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003542660339597149,
+      "loss": 3.3261,
+      "step": 25825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003541851239187188,
+      "loss": 3.2748,
+      "step": 25830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003541042098032789,
+      "loss": 3.2542,
+      "step": 25835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003540232916194798,
+      "loss": 3.3005,
+      "step": 25840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035394236937340604,
+      "loss": 3.2868,
+      "step": 25845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035386144307114257,
+      "loss": 3.3108,
+      "step": 25850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003537805127187746,
+      "loss": 3.3181,
+      "step": 25855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003536995783223876,
+      "loss": 3.4812,
+      "step": 25860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035361863988806765,
+      "loss": 3.369,
+      "step": 25865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035353769742190073,
+      "loss": 3.3039,
+      "step": 25870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003534567509299733,
+      "loss": 3.3852,
+      "step": 25875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003533758004183722,
+      "loss": 3.4023,
+      "step": 25880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003532948458931844,
+      "loss": 3.4037,
+      "step": 25885
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003532138873604974,
+      "loss": 3.416,
+      "step": 25890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035313292482639876,
+      "loss": 3.3859,
+      "step": 25895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003530519582969765,
+      "loss": 3.2529,
+      "step": 25900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035297098777831886,
+      "loss": 3.3192,
+      "step": 25905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035289001327651445,
+      "loss": 3.3498,
+      "step": 25910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035280903479765205,
+      "loss": 3.4666,
+      "step": 25915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035272805234782075,
+      "loss": 3.3699,
+      "step": 25920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035264706593311025,
+      "loss": 3.2185,
+      "step": 25925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035256607555961023,
+      "loss": 3.4041,
+      "step": 25930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035248508123341066,
+      "loss": 3.2361,
+      "step": 25935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035240408296060205,
+      "loss": 3.289,
+      "step": 25940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003523230807472749,
+      "loss": 3.4608,
+      "step": 25945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035224207459952025,
+      "loss": 3.3191,
+      "step": 25950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003521610645234294,
+      "loss": 3.4509,
+      "step": 25955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035208005052509366,
+      "loss": 3.3884,
+      "step": 25960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035199903261060516,
+      "loss": 3.2898,
+      "step": 25965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035191801078605586,
+      "loss": 3.3371,
+      "step": 25970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035183698505753815,
+      "loss": 3.3688,
+      "step": 25975
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000351755955431145,
+      "loss": 3.3285,
+      "step": 25980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035167492191296904,
+      "loss": 3.3297,
+      "step": 25985
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003515938845091039,
+      "loss": 3.3815,
+      "step": 25990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035151284322564307,
+      "loss": 3.3746,
+      "step": 25995
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003514317980686803,
+      "loss": 3.3711,
+      "step": 26000
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 3.395720958709717,
+      "eval_runtime": 149.6727,
+      "eval_samples_per_second": 12.3,
+      "eval_steps_per_second": 0.775,
+      "step": 26000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003513507490443101,
+      "loss": 3.3657,
+      "step": 26005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003512696961586266,
+      "loss": 3.3346,
+      "step": 26010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035118863941772477,
+      "loss": 3.4844,
+      "step": 26015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035110757882769955,
+      "loss": 3.3899,
+      "step": 26020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003510265143946463,
+      "loss": 3.399,
+      "step": 26025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035094544612466073,
+      "loss": 3.2881,
+      "step": 26030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003508643740238387,
+      "loss": 3.4151,
+      "step": 26035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035078329809827633,
+      "loss": 3.2765,
+      "step": 26040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003507022183540703,
+      "loss": 3.4003,
+      "step": 26045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003506211347973173,
+      "loss": 3.4405,
+      "step": 26050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035054004743411435,
+      "loss": 3.3197,
+      "step": 26055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003504589562705589,
+      "loss": 3.27,
+      "step": 26060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035037786131274836,
+      "loss": 3.335,
+      "step": 26065
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035029676256678104,
+      "loss": 3.2878,
+      "step": 26070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003502156600387549,
+      "loss": 3.3474,
+      "step": 26075
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003501345537347684,
+      "loss": 3.4218,
+      "step": 26080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00035005344366092047,
+      "loss": 3.4272,
+      "step": 26085
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003499723298233101,
+      "loss": 3.3154,
+      "step": 26090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034989121222803665,
+      "loss": 3.3007,
+      "step": 26095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003498100908811996,
+      "loss": 3.5048,
+      "step": 26100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000349728965788899,
+      "loss": 3.411,
+      "step": 26105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034964783695723515,
+      "loss": 3.3317,
+      "step": 26110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034956670439230836,
+      "loss": 3.424,
+      "step": 26115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003494855681002194,
+      "loss": 3.3243,
+      "step": 26120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034940442808706926,
+      "loss": 3.4425,
+      "step": 26125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034932328435895946,
+      "loss": 3.413,
+      "step": 26130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003492421369219914,
+      "loss": 3.4129,
+      "step": 26135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034916098578226696,
+      "loss": 3.4568,
+      "step": 26140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034907983094588846,
+      "loss": 3.2578,
+      "step": 26145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003489986724189581,
+      "loss": 3.4228,
+      "step": 26150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003489175102075787,
+      "loss": 3.2936,
+      "step": 26155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034883634431785325,
+      "loss": 3.4025,
+      "step": 26160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003487551747558849,
+      "loss": 3.308,
+      "step": 26165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003486740015277774,
+      "loss": 3.351,
+      "step": 26170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003485928246396344,
+      "loss": 3.2869,
+      "step": 26175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034851164409756,
+      "loss": 3.3198,
+      "step": 26180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003484304599076586,
+      "loss": 3.3552,
+      "step": 26185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034834927207603474,
+      "loss": 3.4593,
+      "step": 26190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034826808060879356,
+      "loss": 3.3524,
+      "step": 26195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034818688551204,
+      "loss": 3.2367,
+      "step": 26200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003481056867918795,
+      "loss": 3.2815,
+      "step": 26205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003480244844544181,
+      "loss": 3.4033,
+      "step": 26210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003479432785057615,
+      "loss": 3.3144,
+      "step": 26215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000347862068952016,
+      "loss": 3.3572,
+      "step": 26220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034778085579928827,
+      "loss": 3.3663,
+      "step": 26225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034769963905368495,
+      "loss": 3.3651,
+      "step": 26230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003476184187213135,
+      "loss": 3.3908,
+      "step": 26235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034753719480828076,
+      "loss": 3.3412,
+      "step": 26240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034745596732069456,
+      "loss": 3.1838,
+      "step": 26245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034737473626466296,
+      "loss": 3.3703,
+      "step": 26250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034729350164629393,
+      "loss": 3.4167,
+      "step": 26255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000347212263471696,
+      "loss": 3.445,
+      "step": 26260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034713102174697773,
+      "loss": 3.3665,
+      "step": 26265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003470497764782483,
+      "loss": 3.4791,
+      "step": 26270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003469685276716167,
+      "loss": 3.327,
+      "step": 26275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00034688727533319254,
+      "loss": 3.479,
+      "step": 26280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003468060194690856,
+      "loss": 3.2287,
+      "step": 26285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003467247600854059,
+      "loss": 3.3918,
+      "step": 26290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003466434971882636,
+      "loss": 3.511,
+      "step": 26295
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003465622307837695,
+      "loss": 3.3726,
+      "step": 26300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003464809608780341,
+      "loss": 3.2298,
+      "step": 26305
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003463996874771688,
+      "loss": 3.3942,
+      "step": 26310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034631841058728474,
+      "loss": 3.4149,
+      "step": 26315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003462371302144935,
+      "loss": 3.3502,
+      "step": 26320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034615584636490715,
+      "loss": 3.2583,
+      "step": 26325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034607455904463766,
+      "loss": 3.357,
+      "step": 26330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003459932682597974,
+      "loss": 3.3674,
+      "step": 26335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034591197401649914,
+      "loss": 3.4351,
+      "step": 26340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003458306763208557,
+      "loss": 3.3208,
+      "step": 26345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034574937517898034,
+      "loss": 3.3437,
+      "step": 26350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034566807059698647,
+      "loss": 3.4414,
+      "step": 26355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003455867625809876,
+      "loss": 3.3091,
+      "step": 26360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034550545113709786,
+      "loss": 3.3744,
+      "step": 26365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003454241362714314,
+      "loss": 3.354,
+      "step": 26370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003453428179901028,
+      "loss": 3.3764,
+      "step": 26375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003452614962992265,
+      "loss": 3.345,
+      "step": 26380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034518017120491765,
+      "loss": 3.3632,
+      "step": 26385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003450988427132915,
+      "loss": 3.3498,
+      "step": 26390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003450175108304636,
+      "loss": 3.3696,
+      "step": 26395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034493617556254953,
+      "loss": 3.3475,
+      "step": 26400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034485483691566537,
+      "loss": 3.3516,
+      "step": 26405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003447734948959273,
+      "loss": 3.3455,
+      "step": 26410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003446921495094519,
+      "loss": 3.3736,
+      "step": 26415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034461080076235597,
+      "loss": 3.4373,
+      "step": 26420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003445294486607564,
+      "loss": 3.3134,
+      "step": 26425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034444809321077045,
+      "loss": 3.3514,
+      "step": 26430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003443667344185157,
+      "loss": 3.4766,
+      "step": 26435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034428537229011,
+      "loss": 3.3528,
+      "step": 26440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003442040068316711,
+      "loss": 3.3033,
+      "step": 26445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003441226380493174,
+      "loss": 3.4208,
+      "step": 26450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003440412659491677,
+      "loss": 3.3896,
+      "step": 26455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034395989053734034,
+      "loss": 3.3191,
+      "step": 26460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003438785118199545,
+      "loss": 3.3925,
+      "step": 26465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003437971298031295,
+      "loss": 3.3533,
+      "step": 26470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034371574449298476,
+      "loss": 3.3886,
+      "step": 26475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003436343558956401,
+      "loss": 3.3418,
+      "step": 26480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034355296401721545,
+      "loss": 3.4094,
+      "step": 26485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003434715688638312,
+      "loss": 3.4429,
+      "step": 26490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034339017044160766,
+      "loss": 3.3538,
+      "step": 26495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003433087687566657,
+      "loss": 3.3811,
+      "step": 26500
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 3.386625289916992,
+      "eval_runtime": 149.9818,
+      "eval_samples_per_second": 12.275,
+      "eval_steps_per_second": 0.773,
+      "step": 26500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003432273638151263,
+      "loss": 3.4065,
+      "step": 26505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034314595562311064,
+      "loss": 3.3432,
+      "step": 26510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003430645441867404,
+      "loss": 3.4248,
+      "step": 26515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034298312951213696,
+      "loss": 3.3458,
+      "step": 26520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034290171160542243,
+      "loss": 3.4478,
+      "step": 26525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003428202904727192,
+      "loss": 3.3292,
+      "step": 26530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003427388661201495,
+      "loss": 3.4159,
+      "step": 26535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034265743855383616,
+      "loss": 3.3999,
+      "step": 26540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000342576007779902,
+      "loss": 3.4405,
+      "step": 26545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003424945738044702,
+      "loss": 3.2289,
+      "step": 26550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034241313663366436,
+      "loss": 3.3146,
+      "step": 26555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034233169627360797,
+      "loss": 3.2682,
+      "step": 26560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003422502527304249,
+      "loss": 3.401,
+      "step": 26565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003421688060102394,
+      "loss": 3.4786,
+      "step": 26570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003420873561191758,
+      "loss": 3.3598,
+      "step": 26575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034200590306335873,
+      "loss": 3.3288,
+      "step": 26580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000341924446848913,
+      "loss": 3.3177,
+      "step": 26585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003418429874819637,
+      "loss": 3.3151,
+      "step": 26590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003417615249686363,
+      "loss": 3.3541,
+      "step": 26595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003416800593150561,
+      "loss": 3.3449,
+      "step": 26600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034159859052734917,
+      "loss": 3.3314,
+      "step": 26605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034151711861164135,
+      "loss": 3.3612,
+      "step": 26610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000341435643574059,
+      "loss": 3.3928,
+      "step": 26615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034135416542072864,
+      "loss": 3.3273,
+      "step": 26620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034127268415777687,
+      "loss": 3.3225,
+      "step": 26625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034119119979133085,
+      "loss": 3.3973,
+      "step": 26630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034110971232751775,
+      "loss": 3.3591,
+      "step": 26635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034102822177246487,
+      "loss": 3.1721,
+      "step": 26640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003409467281323,
+      "loss": 3.1719,
+      "step": 26645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034086523141315105,
+      "loss": 3.3968,
+      "step": 26650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034078373162114615,
+      "loss": 3.3185,
+      "step": 26655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034070222876241365,
+      "loss": 3.3102,
+      "step": 26660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003406207228430821,
+      "loss": 3.4479,
+      "step": 26665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034053921386928036,
+      "loss": 3.4227,
+      "step": 26670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00034045770184713755,
+      "loss": 3.4028,
+      "step": 26675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003403761867827829,
+      "loss": 3.4202,
+      "step": 26680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003402946686823459,
+      "loss": 3.3627,
+      "step": 26685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003402131475519563,
+      "loss": 3.4673,
+      "step": 26690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003401316233977441,
+      "loss": 3.4651,
+      "step": 26695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003400500962258395,
+      "loss": 3.2451,
+      "step": 26700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003399685660423729,
+      "loss": 3.3571,
+      "step": 26705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000339887032853475,
+      "loss": 3.3063,
+      "step": 26710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003398054966652767,
+      "loss": 3.1844,
+      "step": 26715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00033972395748390905,
+      "loss": 3.3216,
+      "step": 26720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00033964241531550326,
+      "loss": 3.2759,
+      "step": 26725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00033956087016619104,
+      "loss": 3.2445,
+      "step": 26730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003394793220421042,
+      "loss": 3.2716,
+      "step": 26735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003393977709493746,
+      "loss": 3.4037,
+      "step": 26740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00033931621689413464,
+      "loss": 3.3809,
+      "step": 26745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003392346598825165,
+      "loss": 3.3758,
+      "step": 26750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00033915309992065316,
+      "loss": 3.3075,
+      "step": 26755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00033907153701467733,
+      "loss": 3.463,
+      "step": 26760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00033898997117072206,
+      "loss": 3.3957,
+      "step": 26765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00033890840239492084,
+      "loss": 3.3882,
+      "step": 26770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003388268306934072,
+      "loss": 3.3903,
+      "step": 26775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003387452560723148,
+      "loss": 3.3708,
+      "step": 26780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00033866367853777784,
+      "loss": 3.3458,
+      "step": 26785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00033858209809593023,
+      "loss": 3.3112,
+      "step": 26790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003385005147529067,
+      "loss": 3.2947,
+      "step": 26795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003384189285148417,
+      "loss": 3.3433,
+      "step": 26800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003383373393878702,
+      "loss": 3.3527,
+      "step": 26805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00033825574737812725,
+      "loss": 3.1773,
+      "step": 26810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00033817415249174824,
+      "loss": 3.3979,
+      "step": 26815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003380925547348684,
+      "loss": 3.4185,
+      "step": 26820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003380109541136238,
+      "loss": 3.3596,
+      "step": 26825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00033792935063415024,
+      "loss": 3.4408,
+      "step": 26830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00033784774430258393,
+      "loss": 3.2784,
+      "step": 26835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003377661351250612,
+      "loss": 3.3703,
+      "step": 26840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003376845231077186,
+      "loss": 3.3976,
+      "step": 26845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00033760290825669305,
+      "loss": 3.4119,
+      "step": 26850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003375212905781215,
+      "loss": 3.4389,
+      "step": 26855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00033743967007814123,
+      "loss": 3.4103,
+      "step": 26860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003373580467628895,
+      "loss": 3.3018,
+      "step": 26865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003372764206385042,
+      "loss": 3.3944,
+      "step": 26870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033719479171112317,
+      "loss": 3.3529,
+      "step": 26875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033711315998688436,
+      "loss": 3.3456,
+      "step": 26880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003370315254719261,
+      "loss": 3.3234,
+      "step": 26885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033694988817238694,
+      "loss": 3.2187,
+      "step": 26890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003368682480944056,
+      "loss": 3.4029,
+      "step": 26895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033678660524412093,
+      "loss": 3.3446,
+      "step": 26900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033670495962767193,
+      "loss": 3.3761,
+      "step": 26905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033662331125119824,
+      "loss": 3.2874,
+      "step": 26910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033654166012083916,
+      "loss": 3.2817,
+      "step": 26915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033646000624273453,
+      "loss": 3.4034,
+      "step": 26920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033637834962302435,
+      "loss": 3.4035,
+      "step": 26925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003362966902678486,
+      "loss": 3.3829,
+      "step": 26930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033621502818334786,
+      "loss": 3.3859,
+      "step": 26935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003361333633756626,
+      "loss": 3.2862,
+      "step": 26940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003360516958509335,
+      "loss": 3.3791,
+      "step": 26945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003359700256153017,
+      "loss": 3.3683,
+      "step": 26950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033588835267490827,
+      "loss": 3.3453,
+      "step": 26955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003358066770358947,
+      "loss": 3.3683,
+      "step": 26960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003357249987044025,
+      "loss": 3.3918,
+      "step": 26965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003356433176865735,
+      "loss": 3.3034,
+      "step": 26970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033556163398854966,
+      "loss": 3.3282,
+      "step": 26975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003354799476164732,
+      "loss": 3.4101,
+      "step": 26980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003353982585764865,
+      "loss": 3.3256,
+      "step": 26985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003353165668747322,
+      "loss": 3.3154,
+      "step": 26990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033523487251735303,
+      "loss": 3.2524,
+      "step": 26995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033515317551049213,
+      "loss": 3.3979,
+      "step": 27000
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 3.3817105293273926,
+      "eval_runtime": 149.7679,
+      "eval_samples_per_second": 12.292,
+      "eval_steps_per_second": 0.775,
+      "step": 27000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003350714758602924,
+      "loss": 3.3361,
+      "step": 27005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003349897735728975,
+      "loss": 3.3722,
+      "step": 27010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033490806865445097,
+      "loss": 3.4067,
+      "step": 27015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003348263611110966,
+      "loss": 3.3635,
+      "step": 27020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033474465094897836,
+      "loss": 3.4518,
+      "step": 27025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003346629381742404,
+      "loss": 3.3886,
+      "step": 27030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033458122279302706,
+      "loss": 3.3627,
+      "step": 27035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003344995048114832,
+      "loss": 3.2823,
+      "step": 27040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033441778423575315,
+      "loss": 3.3224,
+      "step": 27045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033433606107198223,
+      "loss": 3.4366,
+      "step": 27050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033425433532631546,
+      "loss": 3.2814,
+      "step": 27055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003341726070048983,
+      "loss": 3.4437,
+      "step": 27060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003340908761138762,
+      "loss": 3.3634,
+      "step": 27065
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033400914265939484,
+      "loss": 3.3136,
+      "step": 27070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033392740664760033,
+      "loss": 3.2769,
+      "step": 27075
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003338456680846387,
+      "loss": 3.3517,
+      "step": 27080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033376392697665624,
+      "loss": 3.299,
+      "step": 27085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033368218332979955,
+      "loss": 3.3863,
+      "step": 27090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003336004371502152,
+      "loss": 3.3336,
+      "step": 27095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003335186884440503,
+      "loss": 3.2838,
+      "step": 27100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033343693721745186,
+      "loss": 3.3958,
+      "step": 27105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003333551834765669,
+      "loss": 3.2276,
+      "step": 27110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033327342722754325,
+      "loss": 3.3027,
+      "step": 27115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003331916684765284,
+      "loss": 3.3461,
+      "step": 27120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033310990722967,
+      "loss": 3.4142,
+      "step": 27125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033302814349311645,
+      "loss": 3.4698,
+      "step": 27130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033294637727301565,
+      "loss": 3.3017,
+      "step": 27135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003328646085755163,
+      "loss": 3.2897,
+      "step": 27140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003327828374067667,
+      "loss": 3.4139,
+      "step": 27145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033270106377291563,
+      "loss": 3.4206,
+      "step": 27150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033261928768011235,
+      "loss": 3.3275,
+      "step": 27155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003325375091345057,
+      "loss": 3.304,
+      "step": 27160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003324557281422451,
+      "loss": 3.2816,
+      "step": 27165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003323739447094802,
+      "loss": 3.3583,
+      "step": 27170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003322921588423604,
+      "loss": 3.2299,
+      "step": 27175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003322103705470359,
+      "loss": 3.2416,
+      "step": 27180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033212857982965657,
+      "loss": 3.3558,
+      "step": 27185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003320467866963726,
+      "loss": 3.3987,
+      "step": 27190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003319649911533346,
+      "loss": 3.3418,
+      "step": 27195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003318831932066931,
+      "loss": 3.3419,
+      "step": 27200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033180139286259885,
+      "loss": 3.4227,
+      "step": 27205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033171959012720276,
+      "loss": 3.3857,
+      "step": 27210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033163778500665606,
+      "loss": 3.2885,
+      "step": 27215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033155597750711023,
+      "loss": 3.3045,
+      "step": 27220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033147416763471644,
+      "loss": 3.3942,
+      "step": 27225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003313923553956266,
+      "loss": 3.3544,
+      "step": 27230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003313105407959926,
+      "loss": 3.2217,
+      "step": 27235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033122872384196627,
+      "loss": 3.2347,
+      "step": 27240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003311469045397001,
+      "loss": 3.2675,
+      "step": 27245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033106508289534615,
+      "loss": 3.274,
+      "step": 27250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003309832589150573,
+      "loss": 3.3786,
+      "step": 27255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033090143260498616,
+      "loss": 3.3703,
+      "step": 27260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003308196039712856,
+      "loss": 3.2512,
+      "step": 27265
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033073777302010875,
+      "loss": 3.3575,
+      "step": 27270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003306559397576089,
+      "loss": 3.2458,
+      "step": 27275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003305741041899396,
+      "loss": 3.3806,
+      "step": 27280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003304922663232543,
+      "loss": 3.3828,
+      "step": 27285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033041042616370676,
+      "loss": 3.3612,
+      "step": 27290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003303285837174511,
+      "loss": 3.2727,
+      "step": 27295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033024673899064143,
+      "loss": 3.2924,
+      "step": 27300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033016489198943184,
+      "loss": 3.3078,
+      "step": 27305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003300830427199771,
+      "loss": 3.2827,
+      "step": 27310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00033000119118843164,
+      "loss": 3.328,
+      "step": 27315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003299193374009505,
+      "loss": 3.3192,
+      "step": 27320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00032983748136368836,
+      "loss": 3.4029,
+      "step": 27325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00032975562308280057,
+      "loss": 3.4059,
+      "step": 27330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00032967376256444247,
+      "loss": 3.3607,
+      "step": 27335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00032959189981476947,
+      "loss": 3.2867,
+      "step": 27340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00032951003483993734,
+      "loss": 3.3518,
+      "step": 27345
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00032942816764610164,
+      "loss": 3.2867,
+      "step": 27350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003293462982394187,
+      "loss": 3.3899,
+      "step": 27355
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003292644266260445,
+      "loss": 3.22,
+      "step": 27360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003291825528121354,
+      "loss": 3.4481,
+      "step": 27365
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00032910067680384786,
+      "loss": 3.2794,
+      "step": 27370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00032901879860733856,
+      "loss": 3.3851,
+      "step": 27375
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003289369182287644,
+      "loss": 3.3057,
+      "step": 27380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00032885503567428224,
+      "loss": 3.3024,
+      "step": 27385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003287731509500492,
+      "loss": 3.3614,
+      "step": 27390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003286912640622227,
+      "loss": 3.2706,
+      "step": 27395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003286093750169603,
+      "loss": 3.2274,
+      "step": 27400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003285274838204194,
+      "loss": 3.3175,
+      "step": 27405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000328445590478758,
+      "loss": 3.4169,
+      "step": 27410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00032836369499813383,
+      "loss": 3.3511,
+      "step": 27415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00032828179738470534,
+      "loss": 3.4634,
+      "step": 27420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00032819989764463055,
+      "loss": 3.3259,
+      "step": 27425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00032811799578406794,
+      "loss": 3.3551,
+      "step": 27430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003280360918091762,
+      "loss": 3.2537,
+      "step": 27435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00032795418572611405,
+      "loss": 3.2446,
+      "step": 27440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00032787227754104033,
+      "loss": 3.289,
+      "step": 27445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032779036726011425,
+      "loss": 3.3049,
+      "step": 27450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003277084548894949,
+      "loss": 3.3346,
+      "step": 27455
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032762654043534185,
+      "loss": 3.3805,
+      "step": 27460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003275446239038145,
+      "loss": 3.343,
+      "step": 27465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003274627053010725,
+      "loss": 3.2794,
+      "step": 27470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000327380784633276,
+      "loss": 3.34,
+      "step": 27475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032729886190658463,
+      "loss": 3.3971,
+      "step": 27480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003272169371271589,
+      "loss": 3.3285,
+      "step": 27485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003271350103011589,
+      "loss": 3.2486,
+      "step": 27490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003270530814347452,
+      "loss": 3.4702,
+      "step": 27495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032697115053407843,
+      "loss": 3.4144,
+      "step": 27500
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 3.368710994720459,
+      "eval_runtime": 149.7624,
+      "eval_samples_per_second": 12.293,
+      "eval_steps_per_second": 0.775,
+      "step": 27500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032688921760531936,
+      "loss": 3.3579,
+      "step": 27505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032680728265462885,
+      "loss": 3.2252,
+      "step": 27510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003267253456881681,
+      "loss": 3.3385,
+      "step": 27515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032664340671209846,
+      "loss": 3.4116,
+      "step": 27520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032656146573258114,
+      "loss": 3.3403,
+      "step": 27525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003264795227557776,
+      "loss": 3.1292,
+      "step": 27530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032639757778784965,
+      "loss": 3.3242,
+      "step": 27535
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003263156308349593,
+      "loss": 3.259,
+      "step": 27540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032623368190326827,
+      "loss": 3.4236,
+      "step": 27545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003261517309989389,
+      "loss": 3.3361,
+      "step": 27550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032606977812813323,
+      "loss": 3.4113,
+      "step": 27555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032598782329701397,
+      "loss": 3.2926,
+      "step": 27560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003259058665117436,
+      "loss": 3.3761,
+      "step": 27565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032582390777848464,
+      "loss": 3.371,
+      "step": 27570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032574194710340037,
+      "loss": 3.455,
+      "step": 27575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032565998449265353,
+      "loss": 3.3533,
+      "step": 27580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003255780199524073,
+      "loss": 3.4031,
+      "step": 27585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003254960534888252,
+      "loss": 3.3147,
+      "step": 27590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003254140851080704,
+      "loss": 3.2974,
+      "step": 27595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003253321148163067,
+      "loss": 3.3253,
+      "step": 27600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032525014261969786,
+      "loss": 3.4219,
+      "step": 27605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032516816852440756,
+      "loss": 3.3182,
+      "step": 27610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003250861925366001,
+      "loss": 3.4381,
+      "step": 27615
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032500421466243946,
+      "loss": 3.2992,
+      "step": 27620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032492223490809,
+      "loss": 3.3346,
+      "step": 27625
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003248402532797163,
+      "loss": 3.4011,
+      "step": 27630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032475826978348274,
+      "loss": 3.3957,
+      "step": 27635
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003246762844255542,
+      "loss": 3.4597,
+      "step": 27640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032459429721209564,
+      "loss": 3.418,
+      "step": 27645
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032451230814927183,
+      "loss": 3.2963,
+      "step": 27650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003244303172432482,
+      "loss": 3.2607,
+      "step": 27655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032434832450018987,
+      "loss": 3.3302,
+      "step": 27660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032426632992626234,
+      "loss": 3.3218,
+      "step": 27665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032418433352763106,
+      "loss": 3.2973,
+      "step": 27670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032410233531046185,
+      "loss": 3.2454,
+      "step": 27675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003240203352809206,
+      "loss": 3.4339,
+      "step": 27680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032393833344517334,
+      "loss": 3.3274,
+      "step": 27685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000323856329809386,
+      "loss": 3.2768,
+      "step": 27690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003237743243797249,
+      "loss": 3.2868,
+      "step": 27695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003236923171623564,
+      "loss": 3.3539,
+      "step": 27700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003236103081634472,
+      "loss": 3.2585,
+      "step": 27705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032352829738916365,
+      "loss": 3.279,
+      "step": 27710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003234462848456729,
+      "loss": 3.3725,
+      "step": 27715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003233642705391416,
+      "loss": 3.3193,
+      "step": 27720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003232822544757369,
+      "loss": 3.2873,
+      "step": 27725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032320023666162606,
+      "loss": 3.2741,
+      "step": 27730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003231182171029761,
+      "loss": 3.3445,
+      "step": 27735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003230361958059549,
+      "loss": 3.1675,
+      "step": 27740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032295417277672985,
+      "loss": 3.3555,
+      "step": 27745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032287214802146854,
+      "loss": 3.3363,
+      "step": 27750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000322790121546339,
+      "loss": 3.3608,
+      "step": 27755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032270809335750895,
+      "loss": 3.3135,
+      "step": 27760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032262606346114683,
+      "loss": 3.3241,
+      "step": 27765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032254403186342067,
+      "loss": 3.3791,
+      "step": 27770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032246199857049877,
+      "loss": 3.2657,
+      "step": 27775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003223799635885498,
+      "loss": 3.3029,
+      "step": 27780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003222979269237423,
+      "loss": 3.3314,
+      "step": 27785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003222158885822447,
+      "loss": 3.1889,
+      "step": 27790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032213384857022635,
+      "loss": 3.3532,
+      "step": 27795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000322051806893856,
+      "loss": 3.2229,
+      "step": 27800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003219697635593029,
+      "loss": 3.2576,
+      "step": 27805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000321887718572736,
+      "loss": 3.4196,
+      "step": 27810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032180567194032476,
+      "loss": 3.2642,
+      "step": 27815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003217236236682389,
+      "loss": 3.3044,
+      "step": 27820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003216415737626479,
+      "loss": 3.277,
+      "step": 27825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003215595222297214,
+      "loss": 3.3251,
+      "step": 27830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003214774690756293,
+      "loss": 3.291,
+      "step": 27835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003213954143065416,
+      "loss": 3.3613,
+      "step": 27840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032131335792862846,
+      "loss": 3.3399,
+      "step": 27845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032123129994805993,
+      "loss": 3.2345,
+      "step": 27850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003211492403710065,
+      "loss": 3.2515,
+      "step": 27855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032106717920363855,
+      "loss": 3.369,
+      "step": 27860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003209851164521268,
+      "loss": 3.2912,
+      "step": 27865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032090305212264175,
+      "loss": 3.355,
+      "step": 27870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032082098622135424,
+      "loss": 3.3553,
+      "step": 27875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003207389187544354,
+      "loss": 3.3148,
+      "step": 27880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003206568497280562,
+      "loss": 3.2627,
+      "step": 27885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003205747791483876,
+      "loss": 3.388,
+      "step": 27890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003204927070216011,
+      "loss": 3.2773,
+      "step": 27895
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003204106333538681,
+      "loss": 3.363,
+      "step": 27900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032032855815136,
+      "loss": 3.442,
+      "step": 27905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032024648142024866,
+      "loss": 3.2348,
+      "step": 27910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032016440316670554,
+      "loss": 3.2679,
+      "step": 27915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003200823233969027,
+      "loss": 3.2793,
+      "step": 27920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00032000024211701207,
+      "loss": 3.2952,
+      "step": 27925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00031991815933320574,
+      "loss": 3.1779,
+      "step": 27930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00031983607505165585,
+      "loss": 3.2989,
+      "step": 27935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003197539892785348,
+      "loss": 3.3453,
+      "step": 27940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003196719020200151,
+      "loss": 3.2617,
+      "step": 27945
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000319589813282269,
+      "loss": 3.4275,
+      "step": 27950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00031950772307146936,
+      "loss": 3.4124,
+      "step": 27955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00031942563139378906,
+      "loss": 3.4406,
+      "step": 27960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00031934353825540074,
+      "loss": 3.4792,
+      "step": 27965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003192614436624774,
+      "loss": 3.3811,
+      "step": 27970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00031917934762119236,
+      "loss": 3.3476,
+      "step": 27975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003190972501377186,
+      "loss": 3.366,
+      "step": 27980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003190151512182296,
+      "loss": 3.3788,
+      "step": 27985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00031893305086889853,
+      "loss": 3.3044,
+      "step": 27990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003188509490958991,
+      "loss": 3.3833,
+      "step": 27995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00031876884590540495,
+      "loss": 3.3097,
+      "step": 28000
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 3.3711462020874023,
+      "eval_runtime": 149.7727,
+      "eval_samples_per_second": 12.292,
+      "eval_steps_per_second": 0.775,
+      "step": 28000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003186867413035898,
+      "loss": 3.2273,
+      "step": 28005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003186046352966274,
+      "loss": 3.2312,
+      "step": 28010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00031852252789069184,
+      "loss": 3.3193,
+      "step": 28015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000318440419091957,
+      "loss": 3.3245,
+      "step": 28020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003183583089065973,
+      "loss": 3.3313,
+      "step": 28025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031827619734078676,
+      "loss": 3.3352,
+      "step": 28030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003181940844006999,
+      "loss": 3.3963,
+      "step": 28035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031811197009251103,
+      "loss": 3.3619,
+      "step": 28040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003180298544223949,
+      "loss": 3.3542,
+      "step": 28045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031794773739652623,
+      "loss": 3.3766,
+      "step": 28050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003178656190210795,
+      "loss": 3.4085,
+      "step": 28055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003177834993022299,
+      "loss": 3.3648,
+      "step": 28060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031770137824615227,
+      "loss": 3.4042,
+      "step": 28065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003176192558590217,
+      "loss": 3.3609,
+      "step": 28070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031753713214701334,
+      "loss": 3.4013,
+      "step": 28075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031745500711630257,
+      "loss": 3.3054,
+      "step": 28080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003173728807730647,
+      "loss": 3.373,
+      "step": 28085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003172907531234752,
+      "loss": 3.3296,
+      "step": 28090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031720862417370976,
+      "loss": 3.3439,
+      "step": 28095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003171264939299439,
+      "loss": 3.3268,
+      "step": 28100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031704436239835356,
+      "loss": 3.3792,
+      "step": 28105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031696222958511443,
+      "loss": 3.3698,
+      "step": 28110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031688009549640263,
+      "loss": 3.3603,
+      "step": 28115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003167979601383941,
+      "loss": 3.2905,
+      "step": 28120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031671582351726514,
+      "loss": 3.1407,
+      "step": 28125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031663368563919183,
+      "loss": 3.3483,
+      "step": 28130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003165515465103506,
+      "loss": 3.3586,
+      "step": 28135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000316469406136918,
+      "loss": 3.3597,
+      "step": 28140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003163872645250704,
+      "loss": 3.4144,
+      "step": 28145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003163051216809845,
+      "loss": 3.3252,
+      "step": 28150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031622297761083696,
+      "loss": 3.2158,
+      "step": 28155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003161408323208047,
+      "loss": 3.268,
+      "step": 28160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031605868581706457,
+      "loss": 3.2614,
+      "step": 28165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003159765381057935,
+      "loss": 3.32,
+      "step": 28170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031589438919316874,
+      "loss": 3.4281,
+      "step": 28175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003158122390853673,
+      "loss": 3.4212,
+      "step": 28180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003157300877885665,
+      "loss": 3.254,
+      "step": 28185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031564793530894386,
+      "loss": 3.3829,
+      "step": 28190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031556578165267647,
+      "loss": 3.2446,
+      "step": 28195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031548362682594215,
+      "loss": 3.3505,
+      "step": 28200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003154014708349186,
+      "loss": 3.3956,
+      "step": 28205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003153193136857832,
+      "loss": 3.3643,
+      "step": 28210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031523715538471394,
+      "loss": 3.2072,
+      "step": 28215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003151549959378887,
+      "loss": 3.3384,
+      "step": 28220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031507283535148553,
+      "loss": 3.3184,
+      "step": 28225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003149906736316823,
+      "loss": 3.3089,
+      "step": 28230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031490851078465717,
+      "loss": 3.2242,
+      "step": 28235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031482634681658856,
+      "loss": 3.3506,
+      "step": 28240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031474418173365465,
+      "loss": 3.2773,
+      "step": 28245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003146620155420337,
+      "loss": 3.3073,
+      "step": 28250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003145798482479044,
+      "loss": 3.3887,
+      "step": 28255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031449767985744525,
+      "loss": 3.2862,
+      "step": 28260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003144155103768349,
+      "loss": 3.3459,
+      "step": 28265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031433333981225204,
+      "loss": 3.3802,
+      "step": 28270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031425116816987536,
+      "loss": 3.3579,
+      "step": 28275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031416899545588404,
+      "loss": 3.4191,
+      "step": 28280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003140868216764568,
+      "loss": 3.3477,
+      "step": 28285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003140046468377728,
+      "loss": 3.3956,
+      "step": 28290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031392247094601104,
+      "loss": 3.4055,
+      "step": 28295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031384029400735087,
+      "loss": 3.3708,
+      "step": 28300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031375811602797155,
+      "loss": 3.4004,
+      "step": 28305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031367593701405236,
+      "loss": 3.307,
+      "step": 28310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031359375697177284,
+      "loss": 3.3867,
+      "step": 28315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031351157590731244,
+      "loss": 3.3628,
+      "step": 28320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003134293938268508,
+      "loss": 3.3526,
+      "step": 28325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031334721073656763,
+      "loss": 3.2619,
+      "step": 28330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003132650266426425,
+      "loss": 3.2303,
+      "step": 28335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031318284155125534,
+      "loss": 3.3438,
+      "step": 28340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031310065546858606,
+      "loss": 3.3175,
+      "step": 28345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031301846840081475,
+      "loss": 3.348,
+      "step": 28350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003129362803541213,
+      "loss": 3.4054,
+      "step": 28355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031285409133468574,
+      "loss": 3.2905,
+      "step": 28360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031277190134868855,
+      "loss": 3.3457,
+      "step": 28365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031268971040230973,
+      "loss": 3.2979,
+      "step": 28370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031260751850172974,
+      "loss": 3.3189,
+      "step": 28375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031252532565312894,
+      "loss": 3.2012,
+      "step": 28380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003124431318626879,
+      "loss": 3.2903,
+      "step": 28385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031236093713658705,
+      "loss": 3.394,
+      "step": 28390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003122787414810072,
+      "loss": 3.376,
+      "step": 28395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003121965449021288,
+      "loss": 3.424,
+      "step": 28400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003121143474061328,
+      "loss": 3.2631,
+      "step": 28405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003120321489991999,
+      "loss": 3.3431,
+      "step": 28410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003119499496875111,
+      "loss": 3.3405,
+      "step": 28415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003118677494772473,
+      "loss": 3.288,
+      "step": 28420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003117855483745896,
+      "loss": 3.3591,
+      "step": 28425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003117033463857191,
+      "loss": 3.3612,
+      "step": 28430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003116211435168168,
+      "loss": 3.2559,
+      "step": 28435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031153893977406405,
+      "loss": 3.3729,
+      "step": 28440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031145673516364225,
+      "loss": 3.3824,
+      "step": 28445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031137452969173275,
+      "loss": 3.3824,
+      "step": 28450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031129232336451676,
+      "loss": 3.3525,
+      "step": 28455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000311210116188176,
+      "loss": 3.2539,
+      "step": 28460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000311127908168892,
+      "loss": 3.2978,
+      "step": 28465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003110456993128464,
+      "loss": 3.2967,
+      "step": 28470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031096348962622063,
+      "loss": 3.352,
+      "step": 28475
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031088127911519666,
+      "loss": 3.3765,
+      "step": 28480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031079906778595627,
+      "loss": 3.4548,
+      "step": 28485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031071685564468146,
+      "loss": 3.2177,
+      "step": 28490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031063464269755393,
+      "loss": 3.2067,
+      "step": 28495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031055242895075574,
+      "loss": 3.2593,
+      "step": 28500
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 3.359267473220825,
+      "eval_runtime": 149.9726,
+      "eval_samples_per_second": 12.276,
+      "eval_steps_per_second": 0.773,
+      "step": 28500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031047021441046906,
+      "loss": 3.3636,
+      "step": 28505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003103879990828759,
+      "loss": 3.3139,
+      "step": 28510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031030578297415836,
+      "loss": 3.379,
+      "step": 28515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003102235660904988,
+      "loss": 3.3304,
+      "step": 28520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031014134843807944,
+      "loss": 3.3501,
+      "step": 28525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00031005913002308275,
+      "loss": 3.3744,
+      "step": 28530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000309976910851691,
+      "loss": 3.2913,
+      "step": 28535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003098946909300866,
+      "loss": 3.2803,
+      "step": 28540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00030981247026445224,
+      "loss": 3.3851,
+      "step": 28545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003097302488609704,
+      "loss": 3.2995,
+      "step": 28550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00030964802672582364,
+      "loss": 3.356,
+      "step": 28555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00030956580386519477,
+      "loss": 3.2917,
+      "step": 28560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00030948358028526644,
+      "loss": 3.3783,
+      "step": 28565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003094013559922215,
+      "loss": 3.2218,
+      "step": 28570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003093191309922428,
+      "loss": 3.163,
+      "step": 28575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00030923690529151315,
+      "loss": 3.3771,
+      "step": 28580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003091546788962156,
+      "loss": 3.3725,
+      "step": 28585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003090724518125331,
+      "loss": 3.3931,
+      "step": 28590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003089902240466487,
+      "loss": 3.3111,
+      "step": 28595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003089079956047456,
+      "loss": 3.3732,
+      "step": 28600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030882576649300675,
+      "loss": 3.442,
+      "step": 28605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003087435367176157,
+      "loss": 3.393,
+      "step": 28610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003086613062847553,
+      "loss": 3.3046,
+      "step": 28615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003085790752006091,
+      "loss": 3.3409,
+      "step": 28620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003084968434713605,
+      "loss": 3.3822,
+      "step": 28625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003084146111031927,
+      "loss": 3.2859,
+      "step": 28630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003083323781022894,
+      "loss": 3.3541,
+      "step": 28635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030825014447483383,
+      "loss": 3.286,
+      "step": 28640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003081679102270098,
+      "loss": 3.3848,
+      "step": 28645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030808567536500073,
+      "loss": 3.3427,
+      "step": 28650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003080034398949903,
+      "loss": 3.2654,
+      "step": 28655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003079212038231622,
+      "loss": 3.2767,
+      "step": 28660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030783896715570015,
+      "loss": 3.2923,
+      "step": 28665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030775672989878796,
+      "loss": 3.2185,
+      "step": 28670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003076744920586095,
+      "loss": 3.317,
+      "step": 28675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030759225364134846,
+      "loss": 3.2964,
+      "step": 28680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003075100146531889,
+      "loss": 3.3482,
+      "step": 28685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003074277751003148,
+      "loss": 3.456,
+      "step": 28690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030734553498891,
+      "loss": 3.3257,
+      "step": 28695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030726329432515865,
+      "loss": 3.4125,
+      "step": 28700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030718105311524475,
+      "loss": 3.3209,
+      "step": 28705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030709881136535246,
+      "loss": 3.2977,
+      "step": 28710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000307016569081666,
+      "loss": 3.4025,
+      "step": 28715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003069343262703694,
+      "loss": 3.2998,
+      "step": 28720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030685208293764713,
+      "loss": 3.3609,
+      "step": 28725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030676983908968324,
+      "loss": 3.3702,
+      "step": 28730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030668759473266214,
+      "loss": 3.3413,
+      "step": 28735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030660534987276823,
+      "loss": 3.3318,
+      "step": 28740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030652310451618586,
+      "loss": 3.3335,
+      "step": 28745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030644085866909953,
+      "loss": 3.1442,
+      "step": 28750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030635861233769353,
+      "loss": 3.4118,
+      "step": 28755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030627636552815247,
+      "loss": 3.3533,
+      "step": 28760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003061941182466609,
+      "loss": 3.3464,
+      "step": 28765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003061118704994035,
+      "loss": 3.3673,
+      "step": 28770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003060296222925646,
+      "loss": 3.2905,
+      "step": 28775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003059473736323291,
+      "loss": 3.3231,
+      "step": 28780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003058651245248815,
+      "loss": 3.3518,
+      "step": 28785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030578287497640675,
+      "loss": 3.2899,
+      "step": 28790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003057006249930893,
+      "loss": 3.4217,
+      "step": 28795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003056183745811141,
+      "loss": 3.3105,
+      "step": 28800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030553612374666593,
+      "loss": 3.3716,
+      "step": 28805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030545387249592964,
+      "loss": 3.3545,
+      "step": 28810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030537162083509007,
+      "loss": 3.2589,
+      "step": 28815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003052893687703321,
+      "loss": 3.3462,
+      "step": 28820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030520711630784074,
+      "loss": 3.3051,
+      "step": 28825
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030512486345380096,
+      "loss": 3.2761,
+      "step": 28830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003050426102143976,
+      "loss": 3.2527,
+      "step": 28835
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000304960356595816,
+      "loss": 3.2873,
+      "step": 28840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030487810260424077,
+      "loss": 3.3043,
+      "step": 28845
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003047958482458574,
+      "loss": 3.2753,
+      "step": 28850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003047135935268508,
+      "loss": 3.4139,
+      "step": 28855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000304631338453406,
+      "loss": 3.2713,
+      "step": 28860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030454908303170845,
+      "loss": 3.2992,
+      "step": 28865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003044668272679431,
+      "loss": 3.3822,
+      "step": 28870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030438457116829516,
+      "loss": 3.3267,
+      "step": 28875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030430231473895,
+      "loss": 3.2699,
+      "step": 28880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030422005798609277,
+      "loss": 3.2666,
+      "step": 28885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003041378009159089,
+      "loss": 3.3945,
+      "step": 28890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030405554353458357,
+      "loss": 3.2789,
+      "step": 28895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030397328584830214,
+      "loss": 3.3467,
+      "step": 28900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030389102786325,
+      "loss": 3.3268,
+      "step": 28905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003038087695856125,
+      "loss": 3.3759,
+      "step": 28910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000303726511021575,
+      "loss": 3.3726,
+      "step": 28915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000303644252177323,
+      "loss": 3.3361,
+      "step": 28920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030356199305904183,
+      "loss": 3.381,
+      "step": 28925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003034797336729173,
+      "loss": 3.35,
+      "step": 28930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003033974740251343,
+      "loss": 3.3854,
+      "step": 28935
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003033152141218788,
+      "loss": 3.359,
+      "step": 28940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003032329539693362,
+      "loss": 3.2957,
+      "step": 28945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003031506935736921,
+      "loss": 3.2719,
+      "step": 28950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000303068432941132,
+      "loss": 3.2425,
+      "step": 28955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030298617207784133,
+      "loss": 3.3014,
+      "step": 28960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030290391099000585,
+      "loss": 3.2075,
+      "step": 28965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030282164968381125,
+      "loss": 3.3241,
+      "step": 28970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030273938816544293,
+      "loss": 3.2996,
+      "step": 28975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003026571264410867,
+      "loss": 3.3114,
+      "step": 28980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003025748645169282,
+      "loss": 3.2326,
+      "step": 28985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003024926023991531,
+      "loss": 3.3887,
+      "step": 28990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030241034009394706,
+      "loss": 3.4096,
+      "step": 28995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003023280776074957,
+      "loss": 3.2992,
+      "step": 29000
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 3.3513615131378174,
+      "eval_runtime": 150.0716,
+      "eval_samples_per_second": 12.267,
+      "eval_steps_per_second": 0.773,
+      "step": 29000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030224581494598497,
+      "loss": 3.2705,
+      "step": 29005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030216355211560046,
+      "loss": 3.1582,
+      "step": 29010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003020812891225279,
+      "loss": 3.3338,
+      "step": 29015
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030199902597295307,
+      "loss": 3.355,
+      "step": 29020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030191676267306164,
+      "loss": 3.3007,
+      "step": 29025
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003018344992290396,
+      "loss": 3.3631,
+      "step": 29030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030175223564707257,
+      "loss": 3.3573,
+      "step": 29035
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003016699719333463,
+      "loss": 3.3362,
+      "step": 29040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030158770809404686,
+      "loss": 3.324,
+      "step": 29045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030150544413535987,
+      "loss": 3.3532,
+      "step": 29050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003014231800634711,
+      "loss": 3.3602,
+      "step": 29055
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030134091588456663,
+      "loss": 3.3985,
+      "step": 29060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000301258651604832,
+      "loss": 3.3109,
+      "step": 29065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030117638723045333,
+      "loss": 3.234,
+      "step": 29070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030109412276761635,
+      "loss": 3.3347,
+      "step": 29075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003010118582225069,
+      "loss": 3.334,
+      "step": 29080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000300929593601311,
+      "loss": 3.5063,
+      "step": 29085
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003008473289102143,
+      "loss": 3.2729,
+      "step": 29090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003007650641554029,
+      "loss": 3.3154,
+      "step": 29095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030068279934306253,
+      "loss": 3.2814,
+      "step": 29100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030060053447937915,
+      "loss": 3.2991,
+      "step": 29105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030051826957053874,
+      "loss": 3.2739,
+      "step": 29110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000300436004622727,
+      "loss": 3.3853,
+      "step": 29115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003003537396421301,
+      "loss": 3.3941,
+      "step": 29120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030027147463493364,
+      "loss": 3.2138,
+      "step": 29125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030018920960732376,
+      "loss": 3.2438,
+      "step": 29130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003001069445654863,
+      "loss": 3.1981,
+      "step": 29135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00030002467951560706,
+      "loss": 3.2727,
+      "step": 29140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029994241446387206,
+      "loss": 3.3465,
+      "step": 29145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002998601494164673,
+      "loss": 3.3148,
+      "step": 29150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002997778843795786,
+      "loss": 3.2841,
+      "step": 29155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029969561935939173,
+      "loss": 3.3489,
+      "step": 29160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002996133543620928,
+      "loss": 3.2372,
+      "step": 29165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002995310893938676,
+      "loss": 3.2644,
+      "step": 29170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029944882446090207,
+      "loss": 3.2412,
+      "step": 29175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002993665595693822,
+      "loss": 3.3525,
+      "step": 29180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002992842947254937,
+      "loss": 3.2611,
+      "step": 29185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029920202993542267,
+      "loss": 3.3687,
+      "step": 29190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002991197652053548,
+      "loss": 3.4074,
+      "step": 29195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002990375005414761,
+      "loss": 3.3301,
+      "step": 29200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002989552359499725,
+      "loss": 3.2824,
+      "step": 29205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002988729714370297,
+      "loss": 3.3212,
+      "step": 29210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002987907070088338,
+      "loss": 3.3241,
+      "step": 29215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002987084426715704,
+      "loss": 3.3126,
+      "step": 29220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002986261784314255,
+      "loss": 3.2104,
+      "step": 29225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002985439142945851,
+      "loss": 3.3995,
+      "step": 29230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002984616502672348,
+      "loss": 3.3807,
+      "step": 29235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029837938635556043,
+      "loss": 3.2622,
+      "step": 29240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002982971225657479,
+      "loss": 3.2761,
+      "step": 29245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029821485890398313,
+      "loss": 3.3754,
+      "step": 29250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029813259537645186,
+      "loss": 3.4256,
+      "step": 29255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002980503319893397,
+      "loss": 3.3391,
+      "step": 29260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029796806874883265,
+      "loss": 3.3717,
+      "step": 29265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029788580566111645,
+      "loss": 3.3585,
+      "step": 29270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002978035427323767,
+      "loss": 3.3437,
+      "step": 29275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002977212799687995,
+      "loss": 3.3646,
+      "step": 29280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029763901737657017,
+      "loss": 3.2832,
+      "step": 29285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002975567549618747,
+      "loss": 3.1794,
+      "step": 29290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002974744927308987,
+      "loss": 3.3764,
+      "step": 29295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000297392230689828,
+      "loss": 3.3258,
+      "step": 29300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000297309968844848,
+      "loss": 3.3501,
+      "step": 29305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002972277072021448,
+      "loss": 3.3172,
+      "step": 29310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002971454457679037,
+      "loss": 3.3571,
+      "step": 29315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002970631845483104,
+      "loss": 3.2708,
+      "step": 29320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002969809235495506,
+      "loss": 3.4065,
+      "step": 29325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029689866277780993,
+      "loss": 3.2756,
+      "step": 29330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000296816402239274,
+      "loss": 3.3239,
+      "step": 29335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029673414194012816,
+      "loss": 3.284,
+      "step": 29340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029665188188655807,
+      "loss": 3.3406,
+      "step": 29345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002965696220847494,
+      "loss": 3.338,
+      "step": 29350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002964873625408875,
+      "loss": 3.3738,
+      "step": 29355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002964051032611581,
+      "loss": 3.1847,
+      "step": 29360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002963228442517463,
+      "loss": 3.3021,
+      "step": 29365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002962405855188378,
+      "loss": 3.3367,
+      "step": 29370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029615832706861815,
+      "loss": 3.2833,
+      "step": 29375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029607606890727246,
+      "loss": 3.2644,
+      "step": 29380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002959938110409864,
+      "loss": 3.4083,
+      "step": 29385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002959115534759452,
+      "loss": 3.3432,
+      "step": 29390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002958292962183342,
+      "loss": 3.2759,
+      "step": 29395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002957470392743387,
+      "loss": 3.2871,
+      "step": 29400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029566478265014403,
+      "loss": 3.2679,
+      "step": 29405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029558252635193563,
+      "loss": 3.3223,
+      "step": 29410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002955002703858986,
+      "loss": 3.2751,
+      "step": 29415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000295418014758218,
+      "loss": 3.3319,
+      "step": 29420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029533575947507927,
+      "loss": 3.3044,
+      "step": 29425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002952535045426676,
+      "loss": 3.2715,
+      "step": 29430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029517124996716816,
+      "loss": 3.147,
+      "step": 29435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002950889957547658,
+      "loss": 3.3518,
+      "step": 29440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002950067419116457,
+      "loss": 3.2539,
+      "step": 29445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029492448844399327,
+      "loss": 3.26,
+      "step": 29450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002948422353579931,
+      "loss": 3.3907,
+      "step": 29455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002947599826598306,
+      "loss": 3.299,
+      "step": 29460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002946777303556904,
+      "loss": 3.3628,
+      "step": 29465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002945954784517575,
+      "loss": 3.2462,
+      "step": 29470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029451322695421713,
+      "loss": 3.3751,
+      "step": 29475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002944309758692538,
+      "loss": 3.2978,
+      "step": 29480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029434872520305277,
+      "loss": 3.3679,
+      "step": 29485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029426647496179847,
+      "loss": 3.2476,
+      "step": 29490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029418422515167596,
+      "loss": 3.3317,
+      "step": 29495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002941019757788698,
+      "loss": 3.3763,
+      "step": 29500
+    },
+    {
+      "epoch": 0.51,
+      "eval_loss": 3.340322494506836,
+      "eval_runtime": 149.8765,
+      "eval_samples_per_second": 12.283,
+      "eval_steps_per_second": 0.774,
+      "step": 29500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002940197268495649,
+      "loss": 3.2741,
+      "step": 29505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029393747836994595,
+      "loss": 3.2172,
+      "step": 29510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002938552303461975,
+      "loss": 3.282,
+      "step": 29515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002937729827845042,
+      "loss": 3.3107,
+      "step": 29520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029369073569105066,
+      "loss": 3.1937,
+      "step": 29525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029360848907202154,
+      "loss": 3.3407,
+      "step": 29530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002935262429336013,
+      "loss": 3.3098,
+      "step": 29535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002934439972819744,
+      "loss": 3.4239,
+      "step": 29540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002933617521233252,
+      "loss": 3.2941,
+      "step": 29545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002932795074638383,
+      "loss": 3.1586,
+      "step": 29550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000293197263309698,
+      "loss": 3.3078,
+      "step": 29555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002931150196670886,
+      "loss": 3.303,
+      "step": 29560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029303277654219457,
+      "loss": 3.1798,
+      "step": 29565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002929505339411999,
+      "loss": 3.3173,
+      "step": 29570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000292868291870289,
+      "loss": 3.4086,
+      "step": 29575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000292786050335646,
+      "loss": 3.4117,
+      "step": 29580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029270380934345495,
+      "loss": 3.1563,
+      "step": 29585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029262156889990024,
+      "loss": 3.3599,
+      "step": 29590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002925393290111656,
+      "loss": 3.3668,
+      "step": 29595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029245708968343514,
+      "loss": 3.3687,
+      "step": 29600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029237485092289294,
+      "loss": 3.2999,
+      "step": 29605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002922926127357229,
+      "loss": 3.2961,
+      "step": 29610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029221037512810886,
+      "loss": 3.2225,
+      "step": 29615
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029212813810623473,
+      "loss": 3.2368,
+      "step": 29620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002920459016762842,
+      "loss": 3.2402,
+      "step": 29625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002919636658444411,
+      "loss": 3.3306,
+      "step": 29630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029188143061688917,
+      "loss": 3.2542,
+      "step": 29635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029179919599981217,
+      "loss": 3.3902,
+      "step": 29640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002917169619993934,
+      "loss": 3.1883,
+      "step": 29645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002916347286218167,
+      "loss": 3.3464,
+      "step": 29650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002915524958732656,
+      "loss": 3.254,
+      "step": 29655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002914702637599234,
+      "loss": 3.3093,
+      "step": 29660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029138803228797384,
+      "loss": 3.3814,
+      "step": 29665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029130580146359997,
+      "loss": 3.3697,
+      "step": 29670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029122357129298537,
+      "loss": 3.3207,
+      "step": 29675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002911413417823131,
+      "loss": 3.3784,
+      "step": 29680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029105911293776654,
+      "loss": 3.326,
+      "step": 29685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029097688476552904,
+      "loss": 3.4872,
+      "step": 29690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002908946572717834,
+      "loss": 3.2418,
+      "step": 29695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002908124304627129,
+      "loss": 3.383,
+      "step": 29700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002907302043445005,
+      "loss": 3.4032,
+      "step": 29705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002906479789233293,
+      "loss": 3.2928,
+      "step": 29710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002905657542053822,
+      "loss": 3.2659,
+      "step": 29715
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000290483530196842,
+      "loss": 3.3656,
+      "step": 29720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002904013069038915,
+      "loss": 3.3738,
+      "step": 29725
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029031908433271354,
+      "loss": 3.3268,
+      "step": 29730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029023686248949083,
+      "loss": 3.2683,
+      "step": 29735
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029015464138040613,
+      "loss": 3.447,
+      "step": 29740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029007242101164183,
+      "loss": 3.2916,
+      "step": 29745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002899902013893806,
+      "loss": 3.2699,
+      "step": 29750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00028990798251980494,
+      "loss": 3.319,
+      "step": 29755
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028982576440909723,
+      "loss": 3.2191,
+      "step": 29760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028974354706344003,
+      "loss": 3.3557,
+      "step": 29765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002896613304890154,
+      "loss": 3.3007,
+      "step": 29770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028957911469200577,
+      "loss": 3.3596,
+      "step": 29775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028949689967859327,
+      "loss": 3.3322,
+      "step": 29780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002894146854549601,
+      "loss": 3.4363,
+      "step": 29785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002893324720272884,
+      "loss": 3.3202,
+      "step": 29790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002892502594017601,
+      "loss": 3.3451,
+      "step": 29795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002891680475845572,
+      "loss": 3.3622,
+      "step": 29800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002890858365818616,
+      "loss": 3.2766,
+      "step": 29805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002890036263998551,
+      "loss": 3.348,
+      "step": 29810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002889214170447196,
+      "loss": 3.3661,
+      "step": 29815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028883920852263686,
+      "loss": 3.275,
+      "step": 29820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002887570008397883,
+      "loss": 3.3184,
+      "step": 29825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028867479400235567,
+      "loss": 3.154,
+      "step": 29830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002885925880165205,
+      "loss": 3.2861,
+      "step": 29835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002885103828884642,
+      "loss": 3.2612,
+      "step": 29840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002884281786243684,
+      "loss": 3.3055,
+      "step": 29845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028834597523041413,
+      "loss": 3.3218,
+      "step": 29850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028826377271278284,
+      "loss": 3.3651,
+      "step": 29855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028818157107765563,
+      "loss": 3.3722,
+      "step": 29860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002880993703312137,
+      "loss": 3.3252,
+      "step": 29865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002880171704796383,
+      "loss": 3.2013,
+      "step": 29870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002879349715291101,
+      "loss": 3.3523,
+      "step": 29875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028785277348581024,
+      "loss": 3.2081,
+      "step": 29880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002877705763559196,
+      "loss": 3.3849,
+      "step": 29885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028768838014561884,
+      "loss": 3.3127,
+      "step": 29890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000287606184861089,
+      "loss": 3.4062,
+      "step": 29895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002875239905085104,
+      "loss": 3.2684,
+      "step": 29900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002874417970940638,
+      "loss": 3.4488,
+      "step": 29905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028735960462392964,
+      "loss": 3.3177,
+      "step": 29910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002872774131042885,
+      "loss": 3.2003,
+      "step": 29915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002871952225413208,
+      "loss": 3.2894,
+      "step": 29920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028711303294120654,
+      "loss": 3.2948,
+      "step": 29925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002870308443101262,
+      "loss": 3.2572,
+      "step": 29930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028694865665425997,
+      "loss": 3.309,
+      "step": 29935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028686646997978777,
+      "loss": 3.3262,
+      "step": 29940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002867842842928899,
+      "loss": 3.3213,
+      "step": 29945
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000286702099599746,
+      "loss": 3.1783,
+      "step": 29950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000286619915906536,
+      "loss": 3.3147,
+      "step": 29955
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002865377332194399,
+      "loss": 3.2999,
+      "step": 29960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002864555515446371,
+      "loss": 3.1541,
+      "step": 29965
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002863733708883076,
+      "loss": 3.2152,
+      "step": 29970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028629119125663064,
+      "loss": 3.2402,
+      "step": 29975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002862090126557859,
+      "loss": 3.4334,
+      "step": 29980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028612683509195267,
+      "loss": 3.2809,
+      "step": 29985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028604465857131033,
+      "loss": 3.4323,
+      "step": 29990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002859624831000383,
+      "loss": 3.3852,
+      "step": 29995
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002858803086843155,
+      "loss": 3.3623,
+      "step": 30000
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 3.3538341522216797,
+      "eval_runtime": 149.973,
+      "eval_samples_per_second": 12.276,
+      "eval_steps_per_second": 0.773,
+      "step": 30000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002857981353303211,
+      "loss": 3.3106,
+      "step": 30005
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002857159630442341,
+      "loss": 3.4119,
+      "step": 30010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028563379183223355,
+      "loss": 3.3264,
+      "step": 30015
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002855516217004982,
+      "loss": 3.2663,
+      "step": 30020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028546945265520694,
+      "loss": 3.349,
+      "step": 30025
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028538728470253825,
+      "loss": 3.3401,
+      "step": 30030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002853051178486709,
+      "loss": 3.4039,
+      "step": 30035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028522295209978336,
+      "loss": 3.3008,
+      "step": 30040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002851407874620541,
+      "loss": 3.356,
+      "step": 30045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028505862394166156,
+      "loss": 3.1896,
+      "step": 30050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028497646154478383,
+      "loss": 3.3458,
+      "step": 30055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002848943002775992,
+      "loss": 3.3157,
+      "step": 30060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002848121401462858,
+      "loss": 3.3277,
+      "step": 30065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028472998115702157,
+      "loss": 3.4176,
+      "step": 30070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028464782331598463,
+      "loss": 3.3079,
+      "step": 30075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028456566662935265,
+      "loss": 3.2774,
+      "step": 30080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002844835111033034,
+      "loss": 3.3172,
+      "step": 30085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028440135674401453,
+      "loss": 3.1986,
+      "step": 30090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002843192035576638,
+      "loss": 3.3215,
+      "step": 30095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002842370515504286,
+      "loss": 3.3483,
+      "step": 30100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028415490072848627,
+      "loss": 3.3005,
+      "step": 30105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002840727510980142,
+      "loss": 3.2421,
+      "step": 30110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002839906026651897,
+      "loss": 3.2194,
+      "step": 30115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002839084554361897,
+      "loss": 3.3219,
+      "step": 30120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002838263094171916,
+      "loss": 3.2303,
+      "step": 30125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028374416461437194,
+      "loss": 3.226,
+      "step": 30130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028366202103390786,
+      "loss": 3.2418,
+      "step": 30135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002835798786819761,
+      "loss": 3.1874,
+      "step": 30140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002834977375647532,
+      "loss": 3.3188,
+      "step": 30145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002834155976884161,
+      "loss": 3.1829,
+      "step": 30150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002833334590591408,
+      "loss": 3.3849,
+      "step": 30155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028325132168310413,
+      "loss": 3.3192,
+      "step": 30160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002831691855664821,
+      "loss": 3.3352,
+      "step": 30165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002830870507154511,
+      "loss": 3.3424,
+      "step": 30170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002830049171361873,
+      "loss": 3.23,
+      "step": 30175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028292278483486665,
+      "loss": 3.257,
+      "step": 30180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002828406538176649,
+      "loss": 3.3752,
+      "step": 30185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002827585240907581,
+      "loss": 3.2752,
+      "step": 30190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000282676395660322,
+      "loss": 3.2853,
+      "step": 30195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028259426853253215,
+      "loss": 3.3301,
+      "step": 30200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028251214271356406,
+      "loss": 3.258,
+      "step": 30205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002824300182095932,
+      "loss": 3.2661,
+      "step": 30210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028234789502679493,
+      "loss": 3.3113,
+      "step": 30215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002822657731713445,
+      "loss": 3.327,
+      "step": 30220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002821836526494171,
+      "loss": 3.2908,
+      "step": 30225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002821015334671877,
+      "loss": 3.3357,
+      "step": 30230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028201941563083113,
+      "loss": 3.394,
+      "step": 30235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002819372991465225,
+      "loss": 3.3686,
+      "step": 30240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002818551840204363,
+      "loss": 3.4033,
+      "step": 30245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028177307025874725,
+      "loss": 3.3786,
+      "step": 30250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028169095786763007,
+      "loss": 3.2397,
+      "step": 30255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000281608846853259,
+      "loss": 3.3121,
+      "step": 30260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002815267372218083,
+      "loss": 3.2628,
+      "step": 30265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028144462897945223,
+      "loss": 3.3744,
+      "step": 30270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002813625221323651,
+      "loss": 3.2426,
+      "step": 30275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002812804166867208,
+      "loss": 3.4355,
+      "step": 30280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002811983126486931,
+      "loss": 3.245,
+      "step": 30285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000281116210024456,
+      "loss": 3.3295,
+      "step": 30290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002810341088201831,
+      "loss": 3.3174,
+      "step": 30295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028095200904204804,
+      "loss": 3.1724,
+      "step": 30300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002808699106962244,
+      "loss": 3.2641,
+      "step": 30305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028078781378888524,
+      "loss": 3.2328,
+      "step": 30310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028070571832620404,
+      "loss": 3.2577,
+      "step": 30315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00028062362431435404,
+      "loss": 3.3785,
+      "step": 30320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002805415317595081,
+      "loss": 3.2063,
+      "step": 30325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002804594406678394,
+      "loss": 3.3733,
+      "step": 30330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002803773510455205,
+      "loss": 3.3888,
+      "step": 30335
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002802952628987243,
+      "loss": 3.2477,
+      "step": 30340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00028021317623362326,
+      "loss": 3.3568,
+      "step": 30345
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00028013109105638995,
+      "loss": 3.284,
+      "step": 30350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002800490073731969,
+      "loss": 3.3265,
+      "step": 30355
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002799669251902162,
+      "loss": 3.242,
+      "step": 30360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027988484451361994,
+      "loss": 3.391,
+      "step": 30365
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027980276534958037,
+      "loss": 3.3313,
+      "step": 30370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027972068770426936,
+      "loss": 3.2538,
+      "step": 30375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002796386115838587,
+      "loss": 3.3219,
+      "step": 30380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027955653699452006,
+      "loss": 3.3282,
+      "step": 30385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002794744639424251,
+      "loss": 3.3222,
+      "step": 30390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027939239243374517,
+      "loss": 3.3605,
+      "step": 30395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002793103224746518,
+      "loss": 3.296,
+      "step": 30400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027922825407131625,
+      "loss": 3.3659,
+      "step": 30405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002791461872299094,
+      "loss": 3.4053,
+      "step": 30410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027906412195660235,
+      "loss": 3.3414,
+      "step": 30415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027898205825756614,
+      "loss": 3.2646,
+      "step": 30420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027889999613897137,
+      "loss": 3.2449,
+      "step": 30425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027881793560698885,
+      "loss": 3.3722,
+      "step": 30430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027873587666778893,
+      "loss": 3.2959,
+      "step": 30435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027865381932754214,
+      "loss": 3.2453,
+      "step": 30440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002785717635924187,
+      "loss": 3.305,
+      "step": 30445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027848970946858884,
+      "loss": 3.337,
+      "step": 30450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002784076569622227,
+      "loss": 3.2714,
+      "step": 30455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027832560607948993,
+      "loss": 3.3498,
+      "step": 30460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027824355682656057,
+      "loss": 3.2591,
+      "step": 30465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002781615092096042,
+      "loss": 3.3538,
+      "step": 30470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002780794632347904,
+      "loss": 3.3029,
+      "step": 30475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027799741890828874,
+      "loss": 3.4062,
+      "step": 30480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027791537623626843,
+      "loss": 3.2511,
+      "step": 30485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002778333352248985,
+      "loss": 3.2801,
+      "step": 30490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027775129588034823,
+      "loss": 3.3308,
+      "step": 30495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027766925820878647,
+      "loss": 3.3449,
+      "step": 30500
+    },
+    {
+      "epoch": 0.53,
+      "eval_loss": 3.326901435852051,
+      "eval_runtime": 149.9761,
+      "eval_samples_per_second": 12.275,
+      "eval_steps_per_second": 0.773,
+      "step": 30500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027758722221638205,
+      "loss": 3.2604,
+      "step": 30505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002775051879093038,
+      "loss": 3.3171,
+      "step": 30510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027742315529371994,
+      "loss": 3.3948,
+      "step": 30515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027734112437579917,
+      "loss": 3.3968,
+      "step": 30520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027725909516170964,
+      "loss": 3.2516,
+      "step": 30525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027717706765761967,
+      "loss": 3.323,
+      "step": 30530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027709504186969737,
+      "loss": 3.3059,
+      "step": 30535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027701301780411045,
+      "loss": 3.2289,
+      "step": 30540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002769309954670267,
+      "loss": 3.3094,
+      "step": 30545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002768489748646139,
+      "loss": 3.3621,
+      "step": 30550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027676695600303955,
+      "loss": 3.409,
+      "step": 30555
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002766849388884711,
+      "loss": 3.3604,
+      "step": 30560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027660292352707576,
+      "loss": 3.2526,
+      "step": 30565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002765209099250206,
+      "loss": 3.2604,
+      "step": 30570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002764388980884726,
+      "loss": 3.269,
+      "step": 30575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002763568880235989,
+      "loss": 3.3991,
+      "step": 30580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027627487973656606,
+      "loss": 3.368,
+      "step": 30585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002761928732335405,
+      "loss": 3.2534,
+      "step": 30590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002761108685206889,
+      "loss": 3.2685,
+      "step": 30595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027602886560417763,
+      "loss": 3.3091,
+      "step": 30600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027594686449017276,
+      "loss": 3.3925,
+      "step": 30605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027586486518484056,
+      "loss": 3.3245,
+      "step": 30610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027578286769434663,
+      "loss": 3.3573,
+      "step": 30615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027570087202485703,
+      "loss": 3.3753,
+      "step": 30620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027561887818253726,
+      "loss": 3.287,
+      "step": 30625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002755368861735529,
+      "loss": 3.3379,
+      "step": 30630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002754548960040694,
+      "loss": 3.2588,
+      "step": 30635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002753729076802519,
+      "loss": 3.2747,
+      "step": 30640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002752909212082655,
+      "loss": 3.2554,
+      "step": 30645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027520893659427516,
+      "loss": 3.3417,
+      "step": 30650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027512695384444575,
+      "loss": 3.2509,
+      "step": 30655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027504497296494207,
+      "loss": 3.2155,
+      "step": 30660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027496299396192846,
+      "loss": 3.3851,
+      "step": 30665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002748810168415693,
+      "loss": 3.2289,
+      "step": 30670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000274799041610029,
+      "loss": 3.3077,
+      "step": 30675
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002747170682734716,
+      "loss": 3.2613,
+      "step": 30680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002746350968380613,
+      "loss": 3.3562,
+      "step": 30685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002745531273099615,
+      "loss": 3.2328,
+      "step": 30690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002744711596953361,
+      "loss": 3.3524,
+      "step": 30695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002743891940003487,
+      "loss": 3.3606,
+      "step": 30700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027430723023116264,
+      "loss": 3.2986,
+      "step": 30705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002742252683939412,
+      "loss": 3.2556,
+      "step": 30710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002741433084948476,
+      "loss": 3.28,
+      "step": 30715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027406135054004456,
+      "loss": 3.2094,
+      "step": 30720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027397939453569506,
+      "loss": 3.393,
+      "step": 30725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002738974404879617,
+      "loss": 3.2284,
+      "step": 30730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027381548840300706,
+      "loss": 3.2894,
+      "step": 30735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002737335382869936,
+      "loss": 3.3592,
+      "step": 30740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002736515901460834,
+      "loss": 3.2587,
+      "step": 30745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027356964398643854,
+      "loss": 3.2823,
+      "step": 30750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027348769981422094,
+      "loss": 3.3475,
+      "step": 30755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027340575763559256,
+      "loss": 3.3608,
+      "step": 30760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002733238174567149,
+      "loss": 3.3273,
+      "step": 30765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002732418792837494,
+      "loss": 3.241,
+      "step": 30770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027315994312285746,
+      "loss": 3.293,
+      "step": 30775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027307800898020027,
+      "loss": 3.1938,
+      "step": 30780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027299607686193876,
+      "loss": 3.3112,
+      "step": 30785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000272914146774234,
+      "loss": 3.3405,
+      "step": 30790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002728322187232465,
+      "loss": 3.2946,
+      "step": 30795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027275029271513695,
+      "loss": 3.3734,
+      "step": 30800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027266836875606567,
+      "loss": 3.3478,
+      "step": 30805
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000272586446852193,
+      "loss": 3.2108,
+      "step": 30810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002725045270096792,
+      "loss": 3.2706,
+      "step": 30815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002724226092346839,
+      "loss": 3.3201,
+      "step": 30820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002723406935333671,
+      "loss": 3.2457,
+      "step": 30825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002722587799118883,
+      "loss": 3.2723,
+      "step": 30830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027217686837640714,
+      "loss": 3.3165,
+      "step": 30835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027209495893308296,
+      "loss": 3.2665,
+      "step": 30840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027201305158807483,
+      "loss": 3.3003,
+      "step": 30845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002719311463475418,
+      "loss": 3.2617,
+      "step": 30850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002718492432176426,
+      "loss": 3.3741,
+      "step": 30855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002717673422045361,
+      "loss": 3.4239,
+      "step": 30860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002716854433143809,
+      "loss": 3.3527,
+      "step": 30865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002716035465533351,
+      "loss": 3.3153,
+      "step": 30870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027152165192755707,
+      "loss": 3.2969,
+      "step": 30875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027143975944320494,
+      "loss": 3.3132,
+      "step": 30880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027135786910643646,
+      "loss": 3.238,
+      "step": 30885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002712759809234096,
+      "loss": 3.3268,
+      "step": 30890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002711940949002816,
+      "loss": 3.3004,
+      "step": 30895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027111221104321004,
+      "loss": 3.3266,
+      "step": 30900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002710303293583522,
+      "loss": 3.3325,
+      "step": 30905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002709484498518651,
+      "loss": 3.3649,
+      "step": 30910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002708665725299058,
+      "loss": 3.3601,
+      "step": 30915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002707846973986308,
+      "loss": 3.3293,
+      "step": 30920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002707028244641969,
+      "loss": 3.2142,
+      "step": 30925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00027062095373276036,
+      "loss": 3.1972,
+      "step": 30930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002705390852104776,
+      "loss": 3.2593,
+      "step": 30935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002704572189035047,
+      "loss": 3.3917,
+      "step": 30940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002703753548179976,
+      "loss": 3.2728,
+      "step": 30945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002702934929601118,
+      "loss": 3.4312,
+      "step": 30950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002702116333360031,
+      "loss": 3.3181,
+      "step": 30955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000270129775951827,
+      "loss": 3.3649,
+      "step": 30960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002700479208137386,
+      "loss": 3.3491,
+      "step": 30965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002699660679278932,
+      "loss": 3.194,
+      "step": 30970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026988421730044545,
+      "loss": 3.3569,
+      "step": 30975
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002698023689375502,
+      "loss": 3.1581,
+      "step": 30980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002697205228453621,
+      "loss": 3.2575,
+      "step": 30985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026963867903003543,
+      "loss": 3.2312,
+      "step": 30990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026955683749772466,
+      "loss": 3.3679,
+      "step": 30995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002694749982545835,
+      "loss": 3.2296,
+      "step": 31000
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 3.32407808303833,
+      "eval_runtime": 149.9754,
+      "eval_samples_per_second": 12.275,
+      "eval_steps_per_second": 0.773,
+      "step": 31000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026939316130676616,
+      "loss": 3.2916,
+      "step": 31005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026931132666042614,
+      "loss": 3.316,
+      "step": 31010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002692294943217171,
+      "loss": 3.183,
+      "step": 31015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002691476642967925,
+      "loss": 3.3019,
+      "step": 31020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002690658365918054,
+      "loss": 3.1547,
+      "step": 31025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026898401121290875,
+      "loss": 3.3068,
+      "step": 31030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002689021881662555,
+      "loss": 3.2116,
+      "step": 31035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002688203674579984,
+      "loss": 3.2976,
+      "step": 31040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002687385490942899,
+      "loss": 3.3166,
+      "step": 31045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026865673308128224,
+      "loss": 3.2547,
+      "step": 31050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002685749194251276,
+      "loss": 3.3457,
+      "step": 31055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000268493108131978,
+      "loss": 3.2873,
+      "step": 31060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026841129920798513,
+      "loss": 3.3012,
+      "step": 31065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026832949265930087,
+      "loss": 3.1554,
+      "step": 31070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002682476884920763,
+      "loss": 3.4911,
+      "step": 31075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002681658867124628,
+      "loss": 3.2928,
+      "step": 31080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002680840873266115,
+      "loss": 3.0558,
+      "step": 31085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026800229034067326,
+      "loss": 3.2551,
+      "step": 31090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002679204957607989,
+      "loss": 3.2682,
+      "step": 31095
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002678387035931387,
+      "loss": 3.3801,
+      "step": 31100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002677569138438433,
+      "loss": 3.195,
+      "step": 31105
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026767512651906263,
+      "loss": 3.2651,
+      "step": 31110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002675933416249468,
+      "loss": 3.3951,
+      "step": 31115
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026751155916764573,
+      "loss": 3.3247,
+      "step": 31120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026742977915330884,
+      "loss": 3.3234,
+      "step": 31125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026734800158808555,
+      "loss": 3.3103,
+      "step": 31130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026726622647812526,
+      "loss": 3.3172,
+      "step": 31135
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026718445382957695,
+      "loss": 3.4098,
+      "step": 31140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026710268364858963,
+      "loss": 3.3378,
+      "step": 31145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026702091594131187,
+      "loss": 3.3174,
+      "step": 31150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026693915071389216,
+      "loss": 3.2108,
+      "step": 31155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026685738797247886,
+      "loss": 3.3841,
+      "step": 31160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026677562772322016,
+      "loss": 3.4254,
+      "step": 31165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002666938699722641,
+      "loss": 3.3229,
+      "step": 31170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002666121147257582,
+      "loss": 3.1803,
+      "step": 31175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026653036198985016,
+      "loss": 3.2278,
+      "step": 31180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026644861177068746,
+      "loss": 3.2767,
+      "step": 31185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002663668640744171,
+      "loss": 3.3793,
+      "step": 31190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002662851189071862,
+      "loss": 3.2449,
+      "step": 31195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026620337627514165,
+      "loss": 3.3237,
+      "step": 31200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026612163618443003,
+      "loss": 3.3144,
+      "step": 31205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002660398986411976,
+      "loss": 3.3152,
+      "step": 31210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002659581636515908,
+      "loss": 3.2679,
+      "step": 31215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002658764312217557,
+      "loss": 3.2625,
+      "step": 31220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002657947013578381,
+      "loss": 3.2968,
+      "step": 31225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002657129740659836,
+      "loss": 3.2104,
+      "step": 31230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002656312493523377,
+      "loss": 3.331,
+      "step": 31235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002655495272230457,
+      "loss": 3.2342,
+      "step": 31240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002654678076842528,
+      "loss": 3.2085,
+      "step": 31245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002653860907421038,
+      "loss": 3.311,
+      "step": 31250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002653043764027433,
+      "loss": 3.2037,
+      "step": 31255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002652226646723159,
+      "loss": 3.1527,
+      "step": 31260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002651409555569659,
+      "loss": 3.271,
+      "step": 31265
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002650592490628373,
+      "loss": 3.3012,
+      "step": 31270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026497754519607433,
+      "loss": 3.3312,
+      "step": 31275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002648958439628202,
+      "loss": 3.3229,
+      "step": 31280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026481414536921893,
+      "loss": 3.2941,
+      "step": 31285
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026473244942141344,
+      "loss": 3.2946,
+      "step": 31290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000264650756125547,
+      "loss": 3.3225,
+      "step": 31295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002645690654877627,
+      "loss": 3.2401,
+      "step": 31300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026448737751420305,
+      "loss": 3.328,
+      "step": 31305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002644056922110105,
+      "loss": 3.3317,
+      "step": 31310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002643240095843275,
+      "loss": 3.2121,
+      "step": 31315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002642423296402962,
+      "loss": 3.2404,
+      "step": 31320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026416065238505846,
+      "loss": 3.2064,
+      "step": 31325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000264078977824756,
+      "loss": 3.261,
+      "step": 31330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026399730596553023,
+      "loss": 3.174,
+      "step": 31335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002639156368135226,
+      "loss": 3.3145,
+      "step": 31340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026383397037487414,
+      "loss": 3.4055,
+      "step": 31345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026375230665572584,
+      "loss": 3.2682,
+      "step": 31350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002636706456622182,
+      "loss": 3.3942,
+      "step": 31355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002635889874004918,
+      "loss": 3.1798,
+      "step": 31360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026350733187668706,
+      "loss": 3.2755,
+      "step": 31365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026342567909694385,
+      "loss": 3.1718,
+      "step": 31370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026334402906740223,
+      "loss": 3.1946,
+      "step": 31375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026326238179420167,
+      "loss": 3.2054,
+      "step": 31380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002631807372834818,
+      "loss": 3.257,
+      "step": 31385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002630990955413817,
+      "loss": 3.3149,
+      "step": 31390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026301745657404045,
+      "loss": 3.3137,
+      "step": 31395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026293582038759716,
+      "loss": 3.3703,
+      "step": 31400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026285418698819004,
+      "loss": 3.2994,
+      "step": 31405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026277255638195774,
+      "loss": 3.2856,
+      "step": 31410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026269092857503836,
+      "loss": 3.3122,
+      "step": 31415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026260930357357,
+      "loss": 3.2749,
+      "step": 31420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026252768138369044,
+      "loss": 3.161,
+      "step": 31425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002624460620115372,
+      "loss": 3.2431,
+      "step": 31430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002623644454632475,
+      "loss": 3.2436,
+      "step": 31435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002622828317449587,
+      "loss": 3.3021,
+      "step": 31440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026220122086280764,
+      "loss": 3.2542,
+      "step": 31445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002621196128229311,
+      "loss": 3.2806,
+      "step": 31450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002620380076314656,
+      "loss": 3.2708,
+      "step": 31455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026195640529454726,
+      "loss": 3.3919,
+      "step": 31460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002618748058183123,
+      "loss": 3.2865,
+      "step": 31465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002617932092088965,
+      "loss": 3.2749,
+      "step": 31470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002617116154724356,
+      "loss": 3.2656,
+      "step": 31475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026163002461506503,
+      "loss": 3.212,
+      "step": 31480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00026154843664291997,
+      "loss": 3.2677,
+      "step": 31485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002614668515621354,
+      "loss": 3.2452,
+      "step": 31490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00026138526937884604,
+      "loss": 3.2805,
+      "step": 31495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002613036900991865,
+      "loss": 3.3033,
+      "step": 31500
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 3.316983938217163,
+      "eval_runtime": 150.2836,
+      "eval_samples_per_second": 12.25,
+      "eval_steps_per_second": 0.772,
+      "step": 31500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00026122211372929137,
+      "loss": 3.3315,
+      "step": 31505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00026114054027529443,
+      "loss": 3.3347,
+      "step": 31510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002610589697433297,
+      "loss": 3.3488,
+      "step": 31515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002609774021395308,
+      "loss": 3.3674,
+      "step": 31520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00026089583747003134,
+      "loss": 3.1443,
+      "step": 31525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002608142757409646,
+      "loss": 3.3235,
+      "step": 31530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00026073271695846337,
+      "loss": 3.2616,
+      "step": 31535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002606511611286606,
+      "loss": 3.328,
+      "step": 31540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002605696082576889,
+      "loss": 3.339,
+      "step": 31545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002604880583516805,
+      "loss": 3.2483,
+      "step": 31550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00026040651141676774,
+      "loss": 3.3018,
+      "step": 31555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002603249674590823,
+      "loss": 3.3246,
+      "step": 31560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00026024342648475594,
+      "loss": 3.4165,
+      "step": 31565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00026016188849992025,
+      "loss": 3.2257,
+      "step": 31570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002600803535107063,
+      "loss": 3.2111,
+      "step": 31575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002599988215232453,
+      "loss": 3.2358,
+      "step": 31580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025991729254366786,
+      "loss": 3.3897,
+      "step": 31585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025983576657810463,
+      "loss": 3.2663,
+      "step": 31590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002597542436326858,
+      "loss": 3.2689,
+      "step": 31595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025967272371354164,
+      "loss": 3.2168,
+      "step": 31600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025959120682680215,
+      "loss": 3.2391,
+      "step": 31605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025950969297859665,
+      "loss": 3.2852,
+      "step": 31610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025942818217505474,
+      "loss": 3.3367,
+      "step": 31615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002593466744223056,
+      "loss": 3.283,
+      "step": 31620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025926516972647827,
+      "loss": 3.3326,
+      "step": 31625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002591836680937014,
+      "loss": 3.3459,
+      "step": 31630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025910216953010343,
+      "loss": 3.4016,
+      "step": 31635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025902067404181276,
+      "loss": 3.2182,
+      "step": 31640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002589391816349574,
+      "loss": 3.2924,
+      "step": 31645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002588576923156651,
+      "loss": 3.3283,
+      "step": 31650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002587762060900634,
+      "loss": 3.346,
+      "step": 31655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025869472296428,
+      "loss": 3.1884,
+      "step": 31660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025861324294444157,
+      "loss": 3.2707,
+      "step": 31665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002585317660366752,
+      "loss": 3.2709,
+      "step": 31670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025845029224710745,
+      "loss": 3.3347,
+      "step": 31675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002583688215818648,
+      "loss": 3.268,
+      "step": 31680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025828735404707356,
+      "loss": 3.2481,
+      "step": 31685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002582058896488594,
+      "loss": 3.3265,
+      "step": 31690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002581244283933481,
+      "loss": 3.2392,
+      "step": 31695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002580429702866652,
+      "loss": 3.3131,
+      "step": 31700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000257961515334936,
+      "loss": 3.2571,
+      "step": 31705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002578800635442854,
+      "loss": 3.3224,
+      "step": 31710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025779861492083805,
+      "loss": 3.3081,
+      "step": 31715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002577171694707186,
+      "loss": 3.3205,
+      "step": 31720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025763572720005135,
+      "loss": 3.1753,
+      "step": 31725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025755428811496023,
+      "loss": 3.2786,
+      "step": 31730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002574728522215693,
+      "loss": 3.1329,
+      "step": 31735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025739141952600176,
+      "loss": 3.1854,
+      "step": 31740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002573099900343811,
+      "loss": 3.2668,
+      "step": 31745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002572285637528305,
+      "loss": 3.3433,
+      "step": 31750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025714714068747263,
+      "loss": 3.4145,
+      "step": 31755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025706572084443027,
+      "loss": 3.4166,
+      "step": 31760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025698430422982555,
+      "loss": 3.2479,
+      "step": 31765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025690289084978076,
+      "loss": 3.2204,
+      "step": 31770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025682148071041767,
+      "loss": 3.1905,
+      "step": 31775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002567400738178579,
+      "loss": 3.3027,
+      "step": 31780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002566586701782231,
+      "loss": 3.3398,
+      "step": 31785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000256577269797634,
+      "loss": 3.3163,
+      "step": 31790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002564958726822117,
+      "loss": 3.2045,
+      "step": 31795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002564144788380767,
+      "loss": 3.313,
+      "step": 31800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025633308827134973,
+      "loss": 3.2722,
+      "step": 31805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025625170098815065,
+      "loss": 3.3247,
+      "step": 31810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025617031699459947,
+      "loss": 3.3358,
+      "step": 31815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025608893629681576,
+      "loss": 3.0963,
+      "step": 31820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025600755890091897,
+      "loss": 3.2378,
+      "step": 31825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002559261848130284,
+      "loss": 3.2526,
+      "step": 31830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002558448140392629,
+      "loss": 3.3501,
+      "step": 31835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000255763446585741,
+      "loss": 3.1792,
+      "step": 31840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025568208245858116,
+      "loss": 3.27,
+      "step": 31845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025560072166390163,
+      "loss": 3.1158,
+      "step": 31850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002555193642078203,
+      "loss": 3.289,
+      "step": 31855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002554380100964549,
+      "loss": 3.2798,
+      "step": 31860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025535665933592265,
+      "loss": 3.3566,
+      "step": 31865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002552753119323409,
+      "loss": 3.3437,
+      "step": 31870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025519396789182644,
+      "loss": 3.2963,
+      "step": 31875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002551126272204959,
+      "loss": 3.3405,
+      "step": 31880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025503128992446586,
+      "loss": 3.2588,
+      "step": 31885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002549499560098524,
+      "loss": 3.2521,
+      "step": 31890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025486862548277126,
+      "loss": 3.2144,
+      "step": 31895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002547872983493382,
+      "loss": 3.3133,
+      "step": 31900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025470597461566864,
+      "loss": 3.2138,
+      "step": 31905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025462465428787763,
+      "loss": 3.2776,
+      "step": 31910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002545433373720801,
+      "loss": 3.2242,
+      "step": 31915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002544620238743906,
+      "loss": 3.2321,
+      "step": 31920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025438071380092355,
+      "loss": 3.2964,
+      "step": 31925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025429940715779304,
+      "loss": 3.2827,
+      "step": 31930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002542181039511128,
+      "loss": 3.265,
+      "step": 31935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025413680418699675,
+      "loss": 3.2511,
+      "step": 31940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025405550787155774,
+      "loss": 3.3294,
+      "step": 31945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002539742150109092,
+      "loss": 3.3519,
+      "step": 31950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002538929256111637,
+      "loss": 3.3194,
+      "step": 31955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000253811639678434,
+      "loss": 3.3208,
+      "step": 31960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025373035721883234,
+      "loss": 3.2799,
+      "step": 31965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002536490782384706,
+      "loss": 3.3969,
+      "step": 31970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025356780274346063,
+      "loss": 3.261,
+      "step": 31975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025348653073991386,
+      "loss": 3.1804,
+      "step": 31980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002534052622339417,
+      "loss": 3.322,
+      "step": 31985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025332399723165507,
+      "loss": 3.2808,
+      "step": 31990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002532427357391646,
+      "loss": 3.266,
+      "step": 31995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002531614777625807,
+      "loss": 3.3233,
+      "step": 32000
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 3.3087377548217773,
+      "eval_runtime": 149.8766,
+      "eval_samples_per_second": 12.283,
+      "eval_steps_per_second": 0.774,
+      "step": 32000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002530802233080136,
+      "loss": 3.4058,
+      "step": 32005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025299897238157337,
+      "loss": 3.3462,
+      "step": 32010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002529177249893696,
+      "loss": 3.3441,
+      "step": 32015
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002528364811375115,
+      "loss": 3.3583,
+      "step": 32020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002527552408321083,
+      "loss": 3.2635,
+      "step": 32025
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025267400407926886,
+      "loss": 3.2558,
+      "step": 32030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002525927708851018,
+      "loss": 3.3147,
+      "step": 32035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025251154125571545,
+      "loss": 3.2104,
+      "step": 32040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025243031519721777,
+      "loss": 3.4015,
+      "step": 32045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025234909271571663,
+      "loss": 3.1764,
+      "step": 32050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002522678738173195,
+      "loss": 3.282,
+      "step": 32055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00025218665850813355,
+      "loss": 3.2438,
+      "step": 32060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000252105446794266,
+      "loss": 3.2136,
+      "step": 32065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002520242386818233,
+      "loss": 3.3125,
+      "step": 32070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000251943034176912,
+      "loss": 3.3091,
+      "step": 32075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00025186183328563817,
+      "loss": 3.2398,
+      "step": 32080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002517806360141078,
+      "loss": 3.2099,
+      "step": 32085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00025169944236842664,
+      "loss": 3.3446,
+      "step": 32090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00025161825235469975,
+      "loss": 3.3282,
+      "step": 32095
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002515370659790323,
+      "loss": 3.2385,
+      "step": 32100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00025145588324752914,
+      "loss": 3.3214,
+      "step": 32105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002513747041662948,
+      "loss": 3.2566,
+      "step": 32110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00025129352874143343,
+      "loss": 3.329,
+      "step": 32115
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00025121235697904925,
+      "loss": 3.2985,
+      "step": 32120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00025113118888524564,
+      "loss": 3.2731,
+      "step": 32125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002510500244661263,
+      "loss": 3.2967,
+      "step": 32130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002509688637277941,
+      "loss": 3.2774,
+      "step": 32135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00025088770667635213,
+      "loss": 3.2615,
+      "step": 32140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000250806553317903,
+      "loss": 3.2842,
+      "step": 32145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00025072540365854895,
+      "loss": 3.1892,
+      "step": 32150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002506442577043919,
+      "loss": 3.1842,
+      "step": 32155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00025056311546153376,
+      "loss": 3.246,
+      "step": 32160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000250481976936076,
+      "loss": 3.3429,
+      "step": 32165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00025040084213411983,
+      "loss": 3.2132,
+      "step": 32170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002503197110617661,
+      "loss": 3.3142,
+      "step": 32175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002502385837251155,
+      "loss": 3.2998,
+      "step": 32180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00025015746013026834,
+      "loss": 3.1796,
+      "step": 32185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002500763402833248,
+      "loss": 3.2882,
+      "step": 32190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024999522419038466,
+      "loss": 3.2778,
+      "step": 32195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002499141118575473,
+      "loss": 3.2608,
+      "step": 32200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024983300329091206,
+      "loss": 3.2805,
+      "step": 32205
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024975189849657785,
+      "loss": 3.3096,
+      "step": 32210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002496707974806434,
+      "loss": 3.2804,
+      "step": 32215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002495897002492071,
+      "loss": 3.3137,
+      "step": 32220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002495086068083669,
+      "loss": 3.3505,
+      "step": 32225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002494275171642207,
+      "loss": 3.2128,
+      "step": 32230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024934643132286605,
+      "loss": 3.3917,
+      "step": 32235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024926534929040013,
+      "loss": 3.2804,
+      "step": 32240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024918427107292003,
+      "loss": 3.309,
+      "step": 32245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024910319667652227,
+      "loss": 3.2208,
+      "step": 32250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024902212610730326,
+      "loss": 3.3962,
+      "step": 32255
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002489410593713591,
+      "loss": 3.3368,
+      "step": 32260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002488599964747856,
+      "loss": 3.2554,
+      "step": 32265
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002487789374236784,
+      "loss": 3.3785,
+      "step": 32270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024869788222413254,
+      "loss": 3.2223,
+      "step": 32275
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002486168308822429,
+      "loss": 3.1955,
+      "step": 32280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002485357834041043,
+      "loss": 3.2661,
+      "step": 32285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002484547397958111,
+      "loss": 3.3976,
+      "step": 32290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002483737000634574,
+      "loss": 3.3036,
+      "step": 32295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024829266421313664,
+      "loss": 3.2412,
+      "step": 32300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024821163225094265,
+      "loss": 3.2499,
+      "step": 32305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024813060418296846,
+      "loss": 3.2339,
+      "step": 32310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000248049580015307,
+      "loss": 3.2515,
+      "step": 32315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000247968559754051,
+      "loss": 3.2655,
+      "step": 32320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002478875434052925,
+      "loss": 3.1095,
+      "step": 32325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002478065309751237,
+      "loss": 3.322,
+      "step": 32330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002477255224696364,
+      "loss": 3.2905,
+      "step": 32335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002476445178949218,
+      "loss": 3.2749,
+      "step": 32340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024756351725707107,
+      "loss": 3.2526,
+      "step": 32345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024748252056217533,
+      "loss": 3.3218,
+      "step": 32350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002474015278163248,
+      "loss": 3.3869,
+      "step": 32355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002473205390256097,
+      "loss": 3.2438,
+      "step": 32360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024723955419612016,
+      "loss": 3.3304,
+      "step": 32365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002471585733339458,
+      "loss": 3.2835,
+      "step": 32370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002470775964451759,
+      "loss": 3.1467,
+      "step": 32375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002469966235358995,
+      "loss": 3.322,
+      "step": 32380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002469156546122053,
+      "loss": 3.1773,
+      "step": 32385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024683468968018197,
+      "loss": 3.1448,
+      "step": 32390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002467537287459174,
+      "loss": 3.2498,
+      "step": 32395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024667277181549967,
+      "loss": 3.2981,
+      "step": 32400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024659181889501603,
+      "loss": 3.3788,
+      "step": 32405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000246510869990554,
+      "loss": 3.2933,
+      "step": 32410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002464299251082004,
+      "loss": 3.3849,
+      "step": 32415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002463489842540418,
+      "loss": 3.3092,
+      "step": 32420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024626804743416484,
+      "loss": 3.3249,
+      "step": 32425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002461871146546551,
+      "loss": 3.2616,
+      "step": 32430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024610618592159876,
+      "loss": 3.1845,
+      "step": 32435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002460252612410809,
+      "loss": 3.3216,
+      "step": 32440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002459443406191868,
+      "loss": 3.3316,
+      "step": 32445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002458634240620013,
+      "loss": 3.1841,
+      "step": 32450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024578251157560886,
+      "loss": 3.3593,
+      "step": 32455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024570160316609365,
+      "loss": 3.263,
+      "step": 32460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024562069883953956,
+      "loss": 3.2655,
+      "step": 32465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024553979860203034,
+      "loss": 3.1221,
+      "step": 32470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002454589024596492,
+      "loss": 3.1684,
+      "step": 32475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024537801041847894,
+      "loss": 3.2231,
+      "step": 32480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002452971224846024,
+      "loss": 3.3318,
+      "step": 32485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002452162386641019,
+      "loss": 3.2675,
+      "step": 32490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002451353589630595,
+      "loss": 3.2606,
+      "step": 32495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024505448338755706,
+      "loss": 3.2408,
+      "step": 32500
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 3.3065123558044434,
+      "eval_runtime": 149.9745,
+      "eval_samples_per_second": 12.275,
+      "eval_steps_per_second": 0.773,
+      "step": 32500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024497361194367564,
+      "loss": 3.2712,
+      "step": 32505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002448927446374967,
+      "loss": 3.2798,
+      "step": 32510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024481188147510097,
+      "loss": 3.2018,
+      "step": 32515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002447310224625689,
+      "loss": 3.2207,
+      "step": 32520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024465016760598075,
+      "loss": 3.3062,
+      "step": 32525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024456931691141625,
+      "loss": 3.2997,
+      "step": 32530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024448847038495506,
+      "loss": 3.2389,
+      "step": 32535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002444076280326764,
+      "loss": 3.2905,
+      "step": 32540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024432678986065914,
+      "loss": 3.1733,
+      "step": 32545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024424595587498217,
+      "loss": 3.2917,
+      "step": 32550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002441651260817235,
+      "loss": 3.262,
+      "step": 32555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002440843004869611,
+      "loss": 3.2987,
+      "step": 32560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002440034790967727,
+      "loss": 3.3115,
+      "step": 32565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024392266191723583,
+      "loss": 3.2615,
+      "step": 32570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002438418489544273,
+      "loss": 3.3166,
+      "step": 32575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024376104021442408,
+      "loss": 3.3056,
+      "step": 32580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024368023570330225,
+      "loss": 3.236,
+      "step": 32585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024359943542713805,
+      "loss": 3.1418,
+      "step": 32590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024351863939200735,
+      "loss": 3.3424,
+      "step": 32595
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024343784760398545,
+      "loss": 3.2941,
+      "step": 32600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024335706006914766,
+      "loss": 3.2504,
+      "step": 32605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024327627679356854,
+      "loss": 3.1296,
+      "step": 32610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024319549778332273,
+      "loss": 3.1769,
+      "step": 32615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024311472304448432,
+      "loss": 3.2691,
+      "step": 32620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024303395258312724,
+      "loss": 3.2445,
+      "step": 32625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024295318640532506,
+      "loss": 3.2382,
+      "step": 32630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024287242451715086,
+      "loss": 3.1444,
+      "step": 32635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002427916669246775,
+      "loss": 3.3026,
+      "step": 32640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024271091363397757,
+      "loss": 3.2935,
+      "step": 32645
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024263016465112344,
+      "loss": 3.1797,
+      "step": 32650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024254941998218694,
+      "loss": 3.2365,
+      "step": 32655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024246867963323952,
+      "loss": 3.3014,
+      "step": 32660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024238794361035253,
+      "loss": 3.2298,
+      "step": 32665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024230721191959703,
+      "loss": 3.2835,
+      "step": 32670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002422264845670434,
+      "loss": 3.2102,
+      "step": 32675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002421457615587622,
+      "loss": 3.329,
+      "step": 32680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024206504290082313,
+      "loss": 3.2735,
+      "step": 32685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002419843285992959,
+      "loss": 3.2008,
+      "step": 32690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002419036186602499,
+      "loss": 3.263,
+      "step": 32695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024182291308975394,
+      "loss": 3.3097,
+      "step": 32700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024174221189387693,
+      "loss": 3.202,
+      "step": 32705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024166151507868687,
+      "loss": 3.2625,
+      "step": 32710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000241580822650252,
+      "loss": 3.2967,
+      "step": 32715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002415001346146398,
+      "loss": 3.1632,
+      "step": 32720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024141945097791765,
+      "loss": 3.2288,
+      "step": 32725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024133877174615267,
+      "loss": 3.2721,
+      "step": 32730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002412580969254114,
+      "loss": 3.2332,
+      "step": 32735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024117742652176007,
+      "loss": 3.2993,
+      "step": 32740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024109676054126486,
+      "loss": 3.3246,
+      "step": 32745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024101609898999145,
+      "loss": 3.2481,
+      "step": 32750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024093544187400511,
+      "loss": 3.1872,
+      "step": 32755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024085478919937082,
+      "loss": 3.1915,
+      "step": 32760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002407741409721532,
+      "loss": 3.2906,
+      "step": 32765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024069349719841663,
+      "loss": 3.3152,
+      "step": 32770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024061285788422524,
+      "loss": 3.3742,
+      "step": 32775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024053222303564264,
+      "loss": 3.2066,
+      "step": 32780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024045159265873194,
+      "loss": 3.2123,
+      "step": 32785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024037096675955634,
+      "loss": 3.3466,
+      "step": 32790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024029034534417848,
+      "loss": 3.3364,
+      "step": 32795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024020972841866056,
+      "loss": 3.263,
+      "step": 32800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024012911598906466,
+      "loss": 3.138,
+      "step": 32805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024004850806145256,
+      "loss": 3.308,
+      "step": 32810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002399679046418853,
+      "loss": 3.3403,
+      "step": 32815
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002398873057364239,
+      "loss": 3.3254,
+      "step": 32820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023980671135112906,
+      "loss": 3.3335,
+      "step": 32825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023972612149206107,
+      "loss": 3.1811,
+      "step": 32830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023964553616527987,
+      "loss": 3.2545,
+      "step": 32835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023956495537684504,
+      "loss": 3.2524,
+      "step": 32840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023948437913281576,
+      "loss": 3.187,
+      "step": 32845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000239403807439251,
+      "loss": 3.234,
+      "step": 32850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023932324030220947,
+      "loss": 3.3456,
+      "step": 32855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002392426777277494,
+      "loss": 3.3587,
+      "step": 32860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023916211972192843,
+      "loss": 3.1395,
+      "step": 32865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023908156629080427,
+      "loss": 3.3324,
+      "step": 32870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002390010174404342,
+      "loss": 3.2773,
+      "step": 32875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023892047317687493,
+      "loss": 3.3637,
+      "step": 32880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023883993350618316,
+      "loss": 3.216,
+      "step": 32885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023875939843441484,
+      "loss": 3.2905,
+      "step": 32890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023867886796762598,
+      "loss": 3.2365,
+      "step": 32895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002385983421118719,
+      "loss": 3.2825,
+      "step": 32900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002385178208732078,
+      "loss": 3.2922,
+      "step": 32905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023843730425768863,
+      "loss": 3.1794,
+      "step": 32910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023835679227136858,
+      "loss": 3.2932,
+      "step": 32915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023827628492030184,
+      "loss": 3.3405,
+      "step": 32920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002381957822105421,
+      "loss": 3.3375,
+      "step": 32925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023811528414814285,
+      "loss": 3.1655,
+      "step": 32930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002380347907391571,
+      "loss": 3.2732,
+      "step": 32935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002379543019896375,
+      "loss": 3.2293,
+      "step": 32940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023787381790563636,
+      "loss": 3.2525,
+      "step": 32945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023779333849320572,
+      "loss": 3.2663,
+      "step": 32950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002377128637583973,
+      "loss": 3.328,
+      "step": 32955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023763239370726237,
+      "loss": 3.2032,
+      "step": 32960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023755192834585165,
+      "loss": 3.2314,
+      "step": 32965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002374714676802159,
+      "loss": 3.3097,
+      "step": 32970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023739101171640537,
+      "loss": 3.2052,
+      "step": 32975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023731056046046985,
+      "loss": 3.3136,
+      "step": 32980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023723011391845903,
+      "loss": 3.3039,
+      "step": 32985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002371496720964218,
+      "loss": 3.2194,
+      "step": 32990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002370692350004072,
+      "loss": 3.2503,
+      "step": 32995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023698880263646357,
+      "loss": 3.2548,
+      "step": 33000
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 3.29677677154541,
+      "eval_runtime": 149.7729,
+      "eval_samples_per_second": 12.292,
+      "eval_steps_per_second": 0.775,
+      "step": 33000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023690837501063905,
+      "loss": 3.2068,
+      "step": 33005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002368279521289815,
+      "loss": 3.3154,
+      "step": 33010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002367475339975381,
+      "loss": 3.3981,
+      "step": 33015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023666712062235602,
+      "loss": 3.1803,
+      "step": 33020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002365867120094818,
+      "loss": 3.2611,
+      "step": 33025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002365063081649619,
+      "loss": 3.331,
+      "step": 33030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023642590909484236,
+      "loss": 3.2699,
+      "step": 33035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023634551480516852,
+      "loss": 3.2775,
+      "step": 33040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023626512530198574,
+      "loss": 3.2124,
+      "step": 33045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023618474059133887,
+      "loss": 3.267,
+      "step": 33050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023610436067927253,
+      "loss": 3.3642,
+      "step": 33055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023602398557183074,
+      "loss": 3.288,
+      "step": 33060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002359436152750575,
+      "loss": 3.2004,
+      "step": 33065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023586324979499597,
+      "loss": 3.1586,
+      "step": 33070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023578288913768942,
+      "loss": 3.2883,
+      "step": 33075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023570253330918044,
+      "loss": 3.2835,
+      "step": 33080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023562218231551143,
+      "loss": 3.328,
+      "step": 33085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002355418361627245,
+      "loss": 3.1934,
+      "step": 33090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023546149485686104,
+      "loss": 3.3223,
+      "step": 33095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002353811584039625,
+      "loss": 3.26,
+      "step": 33100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023530082681006957,
+      "loss": 3.3357,
+      "step": 33105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023522050008122292,
+      "loss": 3.3325,
+      "step": 33110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002351401782234628,
+      "loss": 3.2175,
+      "step": 33115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002350598612428288,
+      "loss": 3.2222,
+      "step": 33120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023497954914536033,
+      "loss": 3.3419,
+      "step": 33125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023489924193709658,
+      "loss": 3.2388,
+      "step": 33130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023481893962407622,
+      "loss": 3.2658,
+      "step": 33135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023473864221233766,
+      "loss": 3.3275,
+      "step": 33140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023465834970791863,
+      "loss": 3.3339,
+      "step": 33145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002345780621168568,
+      "loss": 3.2137,
+      "step": 33150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023449777944518948,
+      "loss": 3.1605,
+      "step": 33155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002344175016989534,
+      "loss": 3.2205,
+      "step": 33160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002343372288841852,
+      "loss": 3.3544,
+      "step": 33165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023425696100692078,
+      "loss": 3.3108,
+      "step": 33170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023417669807319598,
+      "loss": 3.2136,
+      "step": 33175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023409644008904611,
+      "loss": 3.2587,
+      "step": 33180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023401618706050617,
+      "loss": 3.3205,
+      "step": 33185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023393593899361095,
+      "loss": 3.2509,
+      "step": 33190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023385569589439445,
+      "loss": 3.2005,
+      "step": 33195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023377545776889064,
+      "loss": 3.2256,
+      "step": 33200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000233695224623133,
+      "loss": 3.231,
+      "step": 33205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002336149964631546,
+      "loss": 3.2971,
+      "step": 33210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023353477329498843,
+      "loss": 3.2918,
+      "step": 33215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023345455512466658,
+      "loss": 3.2987,
+      "step": 33220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00023337434195822108,
+      "loss": 3.3542,
+      "step": 33225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023329413380168365,
+      "loss": 3.3258,
+      "step": 33230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023321393066108557,
+      "loss": 3.2357,
+      "step": 33235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023313373254245763,
+      "loss": 3.2245,
+      "step": 33240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023305353945183023,
+      "loss": 3.3036,
+      "step": 33245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023297335139523356,
+      "loss": 3.3151,
+      "step": 33250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023289316837869743,
+      "loss": 3.2807,
+      "step": 33255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023281299040825104,
+      "loss": 3.2591,
+      "step": 33260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002327328174899236,
+      "loss": 3.3142,
+      "step": 33265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023265264962974339,
+      "loss": 3.2627,
+      "step": 33270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023257248683373878,
+      "loss": 3.3279,
+      "step": 33275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023249232910793768,
+      "loss": 3.234,
+      "step": 33280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002324121764583674,
+      "loss": 3.2011,
+      "step": 33285
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023233202889105507,
+      "loss": 3.2268,
+      "step": 33290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023225188641202746,
+      "loss": 3.1476,
+      "step": 33295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023217174902731078,
+      "loss": 3.2587,
+      "step": 33300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023209161674293087,
+      "loss": 3.1833,
+      "step": 33305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023201148956491336,
+      "loss": 3.1849,
+      "step": 33310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023193136749928342,
+      "loss": 3.2485,
+      "step": 33315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023185125055206592,
+      "loss": 3.2812,
+      "step": 33320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023177113872928498,
+      "loss": 3.1795,
+      "step": 33325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023169103203696473,
+      "loss": 3.3434,
+      "step": 33330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023161093048112887,
+      "loss": 3.3059,
+      "step": 33335
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023153083406780045,
+      "loss": 3.3019,
+      "step": 33340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023145074280300258,
+      "loss": 3.2374,
+      "step": 33345
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023137065669275739,
+      "loss": 3.2438,
+      "step": 33350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002312905757430871,
+      "loss": 3.1102,
+      "step": 33355
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023121049996001344,
+      "loss": 3.3015,
+      "step": 33360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002311304293495576,
+      "loss": 3.284,
+      "step": 33365
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023105036391774065,
+      "loss": 3.3114,
+      "step": 33370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023097030367058284,
+      "loss": 3.2978,
+      "step": 33375
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023089024861410446,
+      "loss": 3.2116,
+      "step": 33380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023081019875432517,
+      "loss": 3.3001,
+      "step": 33385
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023073015409726434,
+      "loss": 3.317,
+      "step": 33390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023065011464894103,
+      "loss": 3.2739,
+      "step": 33395
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023057008041537362,
+      "loss": 3.2719,
+      "step": 33400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002304900514025803,
+      "loss": 3.2217,
+      "step": 33405
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023041002761657888,
+      "loss": 3.32,
+      "step": 33410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00023033000906338676,
+      "loss": 3.2321,
+      "step": 33415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000230249995749021,
+      "loss": 3.2446,
+      "step": 33420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000230169987679498,
+      "loss": 3.2752,
+      "step": 33425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000230089984860834,
+      "loss": 3.1693,
+      "step": 33430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002300099872990449,
+      "loss": 3.3488,
+      "step": 33435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002299299950001461,
+      "loss": 3.2554,
+      "step": 33440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022985000797015262,
+      "loss": 3.2451,
+      "step": 33445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022977002621507893,
+      "loss": 3.0189,
+      "step": 33450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022969004974093932,
+      "loss": 3.3202,
+      "step": 33455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002296100785537477,
+      "loss": 3.2214,
+      "step": 33460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022953011265951738,
+      "loss": 3.13,
+      "step": 33465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002294501520642616,
+      "loss": 3.2021,
+      "step": 33470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022937019677399267,
+      "loss": 3.2777,
+      "step": 33475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022929024679472303,
+      "loss": 3.279,
+      "step": 33480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022921030213246437,
+      "loss": 3.3105,
+      "step": 33485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002291303627932282,
+      "loss": 3.2978,
+      "step": 33490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002290504287830257,
+      "loss": 3.2142,
+      "step": 33495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022897050010786725,
+      "loss": 3.2172,
+      "step": 33500
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 3.2933146953582764,
+      "eval_runtime": 149.7693,
+      "eval_samples_per_second": 12.292,
+      "eval_steps_per_second": 0.775,
+      "step": 33500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022889057677376312,
+      "loss": 3.3861,
+      "step": 33505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002288106587867232,
+      "loss": 3.1333,
+      "step": 33510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022873074615275695,
+      "loss": 3.2619,
+      "step": 33515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022865083887787328,
+      "loss": 3.2127,
+      "step": 33520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022857093696808102,
+      "loss": 3.3108,
+      "step": 33525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022849104042938805,
+      "loss": 3.3339,
+      "step": 33530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002284111492678024,
+      "loss": 3.2757,
+      "step": 33535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002283312634893315,
+      "loss": 3.2759,
+      "step": 33540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002282513830999822,
+      "loss": 3.2947,
+      "step": 33545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022817150810576134,
+      "loss": 3.2764,
+      "step": 33550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022809163851267482,
+      "loss": 3.2314,
+      "step": 33555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022801177432672863,
+      "loss": 3.2722,
+      "step": 33560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000227931915553928,
+      "loss": 3.1763,
+      "step": 33565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022785206220027798,
+      "loss": 3.3469,
+      "step": 33570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022777221427178326,
+      "loss": 3.1805,
+      "step": 33575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002276923717744478,
+      "loss": 3.1481,
+      "step": 33580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022761253471427536,
+      "loss": 3.2782,
+      "step": 33585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022753270309726933,
+      "loss": 3.3483,
+      "step": 33590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022745287692943266,
+      "loss": 3.2516,
+      "step": 33595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022737305621676787,
+      "loss": 3.2987,
+      "step": 33600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022729324096527708,
+      "loss": 3.2064,
+      "step": 33605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022721343118096185,
+      "loss": 3.3007,
+      "step": 33610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022713362686982357,
+      "loss": 3.2484,
+      "step": 33615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002270538280378632,
+      "loss": 3.2258,
+      "step": 33620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022697403469108116,
+      "loss": 3.306,
+      "step": 33625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022689424683547735,
+      "loss": 3.1254,
+      "step": 33630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022681446447705153,
+      "loss": 3.1907,
+      "step": 33635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022673468762180295,
+      "loss": 3.2499,
+      "step": 33640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022665491627573033,
+      "loss": 3.2333,
+      "step": 33645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022657515044483223,
+      "loss": 3.1616,
+      "step": 33650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002264953901351064,
+      "loss": 3.1961,
+      "step": 33655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002264156353525506,
+      "loss": 3.3953,
+      "step": 33660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022633588610316185,
+      "loss": 3.2886,
+      "step": 33665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022625614239293692,
+      "loss": 3.3128,
+      "step": 33670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022617640422787226,
+      "loss": 3.318,
+      "step": 33675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022609667161396357,
+      "loss": 3.1784,
+      "step": 33680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022601694455720648,
+      "loss": 3.3266,
+      "step": 33685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022593722306359594,
+      "loss": 3.2716,
+      "step": 33690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002258575071391267,
+      "loss": 3.2966,
+      "step": 33695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022577779678979303,
+      "loss": 3.2999,
+      "step": 33700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002256980920215886,
+      "loss": 3.1279,
+      "step": 33705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022561839284050678,
+      "loss": 3.3071,
+      "step": 33710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022553869925254064,
+      "loss": 3.2625,
+      "step": 33715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002254590112636828,
+      "loss": 3.2814,
+      "step": 33720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002253793288799253,
+      "loss": 3.25,
+      "step": 33725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022529965210725972,
+      "loss": 3.2336,
+      "step": 33730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002252199809516775,
+      "loss": 3.3184,
+      "step": 33735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022514031541916955,
+      "loss": 3.3121,
+      "step": 33740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022506065551572614,
+      "loss": 3.319,
+      "step": 33745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022498100124733738,
+      "loss": 3.2058,
+      "step": 33750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000224901352619993,
+      "loss": 3.2683,
+      "step": 33755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022482170963968195,
+      "loss": 3.3285,
+      "step": 33760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000224742072312393,
+      "loss": 3.3569,
+      "step": 33765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002246624406441145,
+      "loss": 3.2333,
+      "step": 33770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022458281464083442,
+      "loss": 3.1978,
+      "step": 33775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002245031943085402,
+      "loss": 3.1741,
+      "step": 33780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022442357965321889,
+      "loss": 3.0947,
+      "step": 33785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022434397068085693,
+      "loss": 3.2276,
+      "step": 33790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022426436739744068,
+      "loss": 3.1653,
+      "step": 33795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002241847698089559,
+      "loss": 3.3112,
+      "step": 33800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022410517792138801,
+      "loss": 3.1834,
+      "step": 33805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022402559174072162,
+      "loss": 3.2776,
+      "step": 33810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002239460112729413,
+      "loss": 3.1643,
+      "step": 33815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022386643652403133,
+      "loss": 3.3707,
+      "step": 33820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022378686749997505,
+      "loss": 3.2157,
+      "step": 33825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002237073042067559,
+      "loss": 3.2477,
+      "step": 33830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022362774665035632,
+      "loss": 3.2571,
+      "step": 33835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022354819483675885,
+      "loss": 3.169,
+      "step": 33840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022346864877194526,
+      "loss": 3.2855,
+      "step": 33845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022338910846189707,
+      "loss": 3.2139,
+      "step": 33850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022330957391259544,
+      "loss": 3.331,
+      "step": 33855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002232300451300207,
+      "loss": 3.3466,
+      "step": 33860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022315052212015316,
+      "loss": 3.2267,
+      "step": 33865
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022307100488897243,
+      "loss": 3.1652,
+      "step": 33870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002229914934424579,
+      "loss": 3.2467,
+      "step": 33875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022291198778658853,
+      "loss": 3.3519,
+      "step": 33880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022283248792734256,
+      "loss": 3.1367,
+      "step": 33885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022275299387069798,
+      "loss": 3.2817,
+      "step": 33890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022267350562263233,
+      "loss": 3.1939,
+      "step": 33895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022259402318912283,
+      "loss": 3.1678,
+      "step": 33900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022251454657614615,
+      "loss": 3.2199,
+      "step": 33905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022243507578967837,
+      "loss": 3.289,
+      "step": 33910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022235561083569532,
+      "loss": 3.3205,
+      "step": 33915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022227615172017254,
+      "loss": 3.2538,
+      "step": 33920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002221966984490847,
+      "loss": 3.1713,
+      "step": 33925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022211725102840659,
+      "loss": 3.1795,
+      "step": 33930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022203780946411188,
+      "loss": 3.2756,
+      "step": 33935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002219583737621744,
+      "loss": 3.2947,
+      "step": 33940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022187894392856726,
+      "loss": 3.2601,
+      "step": 33945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002217995199692631,
+      "loss": 3.2433,
+      "step": 33950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022172010189023446,
+      "loss": 3.1556,
+      "step": 33955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022164068969745282,
+      "loss": 3.3248,
+      "step": 33960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022156128339688978,
+      "loss": 3.2672,
+      "step": 33965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022148188299451623,
+      "loss": 3.1232,
+      "step": 33970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002214024884963026,
+      "loss": 3.3216,
+      "step": 33975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022132309990821914,
+      "loss": 3.3364,
+      "step": 33980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022124371723623542,
+      "loss": 3.3008,
+      "step": 33985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002211643404863204,
+      "loss": 3.1925,
+      "step": 33990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022108496966444295,
+      "loss": 3.1433,
+      "step": 33995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002210056047765714,
+      "loss": 3.1537,
+      "step": 34000
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 3.282989501953125,
+      "eval_runtime": 151.8721,
+      "eval_samples_per_second": 12.122,
+      "eval_steps_per_second": 0.764,
+      "step": 34000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022092624582867348,
+      "loss": 3.2759,
+      "step": 34005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022084689282671677,
+      "loss": 3.3142,
+      "step": 34010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022076754577666794,
+      "loss": 3.3801,
+      "step": 34015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022068820468449366,
+      "loss": 3.283,
+      "step": 34020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022060886955615986,
+      "loss": 3.2798,
+      "step": 34025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022052954039763218,
+      "loss": 3.1747,
+      "step": 34030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022045021721487593,
+      "loss": 3.2212,
+      "step": 34035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022037090001385552,
+      "loss": 3.3031,
+      "step": 34040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002202915888005354,
+      "loss": 3.219,
+      "step": 34045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022021228358087922,
+      "loss": 3.2706,
+      "step": 34050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022013298436085045,
+      "loss": 3.2221,
+      "step": 34055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022005369114641205,
+      "loss": 3.2401,
+      "step": 34060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002199744039435263,
+      "loss": 3.2445,
+      "step": 34065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002198951227581552,
+      "loss": 3.2795,
+      "step": 34070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002198158475962603,
+      "loss": 3.3098,
+      "step": 34075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021973657846380282,
+      "loss": 3.2402,
+      "step": 34080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021965731536674333,
+      "loss": 3.2837,
+      "step": 34085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002195780583110419,
+      "loss": 3.2242,
+      "step": 34090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002194988073026583,
+      "loss": 3.3176,
+      "step": 34095
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021941956234755185,
+      "loss": 3.2141,
+      "step": 34100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021934032345168132,
+      "loss": 3.2779,
+      "step": 34105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021926109062100525,
+      "loss": 3.189,
+      "step": 34110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021918186386148122,
+      "loss": 3.2255,
+      "step": 34115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021910264317906685,
+      "loss": 3.2848,
+      "step": 34120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021902342857971918,
+      "loss": 3.125,
+      "step": 34125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002189442200693946,
+      "loss": 3.2491,
+      "step": 34130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002188650176540494,
+      "loss": 3.2903,
+      "step": 34135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021878582133963895,
+      "loss": 3.1633,
+      "step": 34140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002187066311321186,
+      "loss": 3.183,
+      "step": 34145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021862744703744286,
+      "loss": 3.227,
+      "step": 34150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002185482690615661,
+      "loss": 3.1407,
+      "step": 34155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002184690972104422,
+      "loss": 3.259,
+      "step": 34160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002183899314900243,
+      "loss": 3.2113,
+      "step": 34165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002183107719062652,
+      "loss": 3.2791,
+      "step": 34170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021823161846511746,
+      "loss": 3.2624,
+      "step": 34175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021815247117253298,
+      "loss": 3.2633,
+      "step": 34180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002180733300344632,
+      "loss": 3.1451,
+      "step": 34185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021799419505685915,
+      "loss": 3.2792,
+      "step": 34190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021791506624567128,
+      "loss": 3.2983,
+      "step": 34195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002178359436068498,
+      "loss": 3.0929,
+      "step": 34200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021775682714634428,
+      "loss": 3.2645,
+      "step": 34205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021767771687010386,
+      "loss": 3.2737,
+      "step": 34210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002175986127840773,
+      "loss": 3.2691,
+      "step": 34215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021751951489421266,
+      "loss": 3.2498,
+      "step": 34220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021744042320645787,
+      "loss": 3.3095,
+      "step": 34225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021736133772676006,
+      "loss": 3.3028,
+      "step": 34230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021728225846106612,
+      "loss": 3.2369,
+      "step": 34235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021720318541532262,
+      "loss": 3.245,
+      "step": 34240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021712411859547512,
+      "loss": 3.2824,
+      "step": 34245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021704505800746913,
+      "loss": 3.3172,
+      "step": 34250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021696600365724966,
+      "loss": 3.2327,
+      "step": 34255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002168869555507612,
+      "loss": 3.198,
+      "step": 34260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021680791369394783,
+      "loss": 3.2636,
+      "step": 34265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021672887809275285,
+      "loss": 3.317,
+      "step": 34270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021664984875311948,
+      "loss": 3.2125,
+      "step": 34275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002165708256809904,
+      "loss": 3.2339,
+      "step": 34280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002164918088823076,
+      "loss": 3.2303,
+      "step": 34285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002164127983630129,
+      "loss": 3.2303,
+      "step": 34290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021633379412904727,
+      "loss": 3.3098,
+      "step": 34295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021625479618635153,
+      "loss": 3.296,
+      "step": 34300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021617580454086599,
+      "loss": 3.1825,
+      "step": 34305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002160968191985303,
+      "loss": 3.2411,
+      "step": 34310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021601784016528396,
+      "loss": 3.2966,
+      "step": 34315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021593886744706548,
+      "loss": 3.2849,
+      "step": 34320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021585990104981345,
+      "loss": 3.2534,
+      "step": 34325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021578094097946557,
+      "loss": 3.2557,
+      "step": 34330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021570198724195933,
+      "loss": 3.2512,
+      "step": 34335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021562303984323176,
+      "loss": 3.2774,
+      "step": 34340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021554409878921904,
+      "loss": 3.2232,
+      "step": 34345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021546516408585726,
+      "loss": 3.28,
+      "step": 34350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002153862357390819,
+      "loss": 3.2287,
+      "step": 34355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000215307313754828,
+      "loss": 3.2912,
+      "step": 34360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021522839813903013,
+      "loss": 3.2309,
+      "step": 34365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021514948889762224,
+      "loss": 3.2931,
+      "step": 34370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021507058603653783,
+      "loss": 3.2682,
+      "step": 34375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00021499168956171016,
+      "loss": 3.2441,
+      "step": 34380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021491279947907178,
+      "loss": 3.3051,
+      "step": 34385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002148339157945549,
+      "loss": 3.2857,
+      "step": 34390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000214755038514091,
+      "loss": 3.274,
+      "step": 34395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021467616764361128,
+      "loss": 3.2249,
+      "step": 34400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002145973031890465,
+      "loss": 3.2408,
+      "step": 34405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021451844515632686,
+      "loss": 3.2944,
+      "step": 34410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021443959355138214,
+      "loss": 3.2375,
+      "step": 34415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002143607483801414,
+      "loss": 3.3012,
+      "step": 34420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002142819096485335,
+      "loss": 3.3001,
+      "step": 34425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021420307736248667,
+      "loss": 3.2736,
+      "step": 34430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002141242515279287,
+      "loss": 3.3536,
+      "step": 34435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021404543215078697,
+      "loss": 3.2831,
+      "step": 34440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021396661923698827,
+      "loss": 3.2456,
+      "step": 34445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021388781279245886,
+      "loss": 3.2275,
+      "step": 34450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021380901282312456,
+      "loss": 3.2121,
+      "step": 34455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021373021933491075,
+      "loss": 3.3182,
+      "step": 34460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021365143233374238,
+      "loss": 3.2926,
+      "step": 34465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021357265182554385,
+      "loss": 3.2554,
+      "step": 34470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021349387781623887,
+      "loss": 3.2501,
+      "step": 34475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002134151103117509,
+      "loss": 3.2919,
+      "step": 34480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021333634931800297,
+      "loss": 3.2388,
+      "step": 34485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002132575948409174,
+      "loss": 3.3324,
+      "step": 34490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002131788468864162,
+      "loss": 3.1517,
+      "step": 34495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002131001054604207,
+      "loss": 3.1907,
+      "step": 34500
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 3.2766971588134766,
+      "eval_runtime": 150.1846,
+      "eval_samples_per_second": 12.258,
+      "eval_steps_per_second": 0.772,
+      "step": 34500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021302137056885198,
+      "loss": 3.1657,
+      "step": 34505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021294264221763032,
+      "loss": 3.2792,
+      "step": 34510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002128639204126758,
+      "loss": 3.3099,
+      "step": 34515
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021278520515990808,
+      "loss": 3.345,
+      "step": 34520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002127064964652459,
+      "loss": 3.2582,
+      "step": 34525
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002126277943346077,
+      "loss": 3.2024,
+      "step": 34530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021254909877391163,
+      "loss": 3.2492,
+      "step": 34535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021247040978907524,
+      "loss": 3.2715,
+      "step": 34540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021239172738601539,
+      "loss": 3.2769,
+      "step": 34545
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021231305157064868,
+      "loss": 3.1182,
+      "step": 34550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021223438234889102,
+      "loss": 3.2583,
+      "step": 34555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021215571972665808,
+      "loss": 3.3762,
+      "step": 34560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021207706370986483,
+      "loss": 3.3264,
+      "step": 34565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021199841430442586,
+      "loss": 3.2889,
+      "step": 34570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021191977151625505,
+      "loss": 3.2835,
+      "step": 34575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021184113535126597,
+      "loss": 3.1812,
+      "step": 34580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021176250581537176,
+      "loss": 3.2657,
+      "step": 34585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021168388291448485,
+      "loss": 3.2983,
+      "step": 34590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002116052666545175,
+      "loss": 3.2066,
+      "step": 34595
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002115266570413809,
+      "loss": 3.2254,
+      "step": 34600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021144805408098634,
+      "loss": 3.2641,
+      "step": 34605
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021136945777924427,
+      "loss": 3.2156,
+      "step": 34610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021129086814206474,
+      "loss": 3.3646,
+      "step": 34615
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021121228517535738,
+      "loss": 3.1861,
+      "step": 34620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021113370888503106,
+      "loss": 3.2872,
+      "step": 34625
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021105513927699446,
+      "loss": 3.1778,
+      "step": 34630
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002109765763571555,
+      "loss": 3.3283,
+      "step": 34635
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021089802013142173,
+      "loss": 3.1781,
+      "step": 34640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021081947060570035,
+      "loss": 3.2741,
+      "step": 34645
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021074092778589767,
+      "loss": 3.252,
+      "step": 34650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002106623916779197,
+      "loss": 3.2506,
+      "step": 34655
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021058386228767206,
+      "loss": 3.2566,
+      "step": 34660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021050533962105975,
+      "loss": 3.2547,
+      "step": 34665
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021042682368398717,
+      "loss": 3.2342,
+      "step": 34670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021034831448235857,
+      "loss": 3.2487,
+      "step": 34675
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021026981202207712,
+      "loss": 3.308,
+      "step": 34680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021019131630904602,
+      "loss": 3.3426,
+      "step": 34685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00021011282734916757,
+      "loss": 3.2593,
+      "step": 34690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002100343451483439,
+      "loss": 3.2209,
+      "step": 34695
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020995586971247647,
+      "loss": 3.1753,
+      "step": 34700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020987740104746607,
+      "loss": 3.3474,
+      "step": 34705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020979893915921332,
+      "loss": 3.2738,
+      "step": 34710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020972048405361798,
+      "loss": 3.1795,
+      "step": 34715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020964203573657959,
+      "loss": 3.1401,
+      "step": 34720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020956359421399714,
+      "loss": 3.2246,
+      "step": 34725
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002094851594917689,
+      "loss": 3.2825,
+      "step": 34730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020940673157579268,
+      "loss": 3.2241,
+      "step": 34735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020932831047196595,
+      "loss": 3.2764,
+      "step": 34740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002092498961861857,
+      "loss": 3.2176,
+      "step": 34745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002091714887243482,
+      "loss": 3.2499,
+      "step": 34750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020909308809234913,
+      "loss": 3.2197,
+      "step": 34755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000209014694296084,
+      "loss": 3.2597,
+      "step": 34760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020893630734144755,
+      "loss": 3.1445,
+      "step": 34765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020885792723433406,
+      "loss": 3.2218,
+      "step": 34770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002087795539806375,
+      "loss": 3.2191,
+      "step": 34775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020870118758625086,
+      "loss": 3.3003,
+      "step": 34780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020862282805706706,
+      "loss": 3.293,
+      "step": 34785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002085444753989782,
+      "loss": 3.2902,
+      "step": 34790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020846612961787612,
+      "loss": 3.2605,
+      "step": 34795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002083877907196521,
+      "loss": 3.2403,
+      "step": 34800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020830945871019656,
+      "loss": 3.1848,
+      "step": 34805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020823113359539993,
+      "loss": 3.2072,
+      "step": 34810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020815281538115164,
+      "loss": 3.235,
+      "step": 34815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020807450407334097,
+      "loss": 3.2259,
+      "step": 34820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002079961996778566,
+      "loss": 3.1593,
+      "step": 34825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002079179022005864,
+      "loss": 3.2826,
+      "step": 34830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020783961164741804,
+      "loss": 3.2567,
+      "step": 34835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020776132802423851,
+      "loss": 3.2129,
+      "step": 34840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020768305133693447,
+      "loss": 3.2653,
+      "step": 34845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020760478159139196,
+      "loss": 3.1523,
+      "step": 34850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002075265187934962,
+      "loss": 3.284,
+      "step": 34855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002074482629491323,
+      "loss": 3.1118,
+      "step": 34860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002073700140641848,
+      "loss": 3.2638,
+      "step": 34865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020729177214453745,
+      "loss": 3.1445,
+      "step": 34870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002072135371960738,
+      "loss": 3.2557,
+      "step": 34875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020713530922467652,
+      "loss": 3.2218,
+      "step": 34880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020705708823622805,
+      "loss": 3.3213,
+      "step": 34885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020697887423661032,
+      "loss": 3.2697,
+      "step": 34890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020690066723170447,
+      "loss": 3.2742,
+      "step": 34895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002068224672273914,
+      "loss": 3.2105,
+      "step": 34900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020674427422955115,
+      "loss": 3.2227,
+      "step": 34905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020666608824406363,
+      "loss": 3.2625,
+      "step": 34910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020658790927680787,
+      "loss": 3.1875,
+      "step": 34915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020650973733366262,
+      "loss": 3.233,
+      "step": 34920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020643157242050602,
+      "loss": 3.2415,
+      "step": 34925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020635341454321575,
+      "loss": 3.2621,
+      "step": 34930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020627526370766862,
+      "loss": 3.3244,
+      "step": 34935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002061971199197413,
+      "loss": 3.2816,
+      "step": 34940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020611898318530995,
+      "loss": 3.1986,
+      "step": 34945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020604085351024986,
+      "loss": 3.2335,
+      "step": 34950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020596273090043614,
+      "loss": 3.2378,
+      "step": 34955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020588461536174304,
+      "loss": 3.1589,
+      "step": 34960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002058065069000445,
+      "loss": 3.2316,
+      "step": 34965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000205728405521214,
+      "loss": 3.2408,
+      "step": 34970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002056503112311242,
+      "loss": 3.2371,
+      "step": 34975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020557222403564757,
+      "loss": 3.2762,
+      "step": 34980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002054941439406556,
+      "loss": 3.2734,
+      "step": 34985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002054160709520198,
+      "loss": 3.1853,
+      "step": 34990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020533800507561067,
+      "loss": 3.2824,
+      "step": 34995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020525994631729837,
+      "loss": 3.19,
+      "step": 35000
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 3.272491693496704,
+      "eval_runtime": 149.9711,
+      "eval_samples_per_second": 12.276,
+      "eval_steps_per_second": 0.773,
+      "step": 35000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002051818946829527,
+      "loss": 3.2679,
+      "step": 35005
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020510385017844258,
+      "loss": 3.1934,
+      "step": 35010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020502581280963656,
+      "loss": 3.2659,
+      "step": 35015
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002049477825824026,
+      "loss": 3.1719,
+      "step": 35020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002048697595026084,
+      "loss": 3.2524,
+      "step": 35025
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020479174357612077,
+      "loss": 3.147,
+      "step": 35030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020471373480880593,
+      "loss": 3.0605,
+      "step": 35035
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020463573320652996,
+      "loss": 3.2339,
+      "step": 35040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020455773877515814,
+      "loss": 3.3027,
+      "step": 35045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020447975152055518,
+      "loss": 3.2568,
+      "step": 35050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020440177144858554,
+      "loss": 3.2592,
+      "step": 35055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020432379856511257,
+      "loss": 3.2078,
+      "step": 35060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002042458328759996,
+      "loss": 3.1336,
+      "step": 35065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020416787438710937,
+      "loss": 3.2229,
+      "step": 35070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020408992310430382,
+      "loss": 3.3117,
+      "step": 35075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020401197903344463,
+      "loss": 3.2499,
+      "step": 35080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020393404218039252,
+      "loss": 3.2453,
+      "step": 35085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002038561125510082,
+      "loss": 3.1057,
+      "step": 35090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002037781901511514,
+      "loss": 3.2666,
+      "step": 35095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020370027498668158,
+      "loss": 3.0863,
+      "step": 35100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002036223670634577,
+      "loss": 3.2022,
+      "step": 35105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002035444663873378,
+      "loss": 3.3174,
+      "step": 35110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020346657296417968,
+      "loss": 3.242,
+      "step": 35115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020338868679984048,
+      "loss": 3.2349,
+      "step": 35120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020331080790017702,
+      "loss": 3.3197,
+      "step": 35125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020323293627104528,
+      "loss": 3.202,
+      "step": 35130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002031550719183008,
+      "loss": 3.0995,
+      "step": 35135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020307721484779854,
+      "loss": 3.0756,
+      "step": 35140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000202999365065393,
+      "loss": 3.2164,
+      "step": 35145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002029215225769381,
+      "loss": 3.1971,
+      "step": 35150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020284368738828716,
+      "loss": 3.1615,
+      "step": 35155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020276585950529312,
+      "loss": 3.2523,
+      "step": 35160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000202688038933808,
+      "loss": 3.3051,
+      "step": 35165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020261022567968372,
+      "loss": 3.1922,
+      "step": 35170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002025324197487713,
+      "loss": 3.2387,
+      "step": 35175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020245462114692138,
+      "loss": 3.3196,
+      "step": 35180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002023768298799842,
+      "loss": 3.1601,
+      "step": 35185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000202299045953809,
+      "loss": 3.2727,
+      "step": 35190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020222126937424477,
+      "loss": 3.2983,
+      "step": 35195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020214350014714,
+      "loss": 3.294,
+      "step": 35200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020206573827834254,
+      "loss": 3.2709,
+      "step": 35205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002019879837736997,
+      "loss": 3.2138,
+      "step": 35210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020191023663905814,
+      "loss": 3.2238,
+      "step": 35215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020183249688026405,
+      "loss": 3.2299,
+      "step": 35220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002017547645031631,
+      "loss": 3.2146,
+      "step": 35225
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020167703951360039,
+      "loss": 3.1378,
+      "step": 35230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002015993219174205,
+      "loss": 3.3198,
+      "step": 35235
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002015216117204671,
+      "loss": 3.1608,
+      "step": 35240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020144390892858392,
+      "loss": 3.2125,
+      "step": 35245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002013662135476137,
+      "loss": 3.2722,
+      "step": 35250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020128852558339868,
+      "loss": 2.9673,
+      "step": 35255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020121084504178078,
+      "loss": 3.0968,
+      "step": 35260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020113317192860093,
+      "loss": 3.2643,
+      "step": 35265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020105550624969988,
+      "loss": 3.0082,
+      "step": 35270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020097784801091767,
+      "loss": 3.228,
+      "step": 35275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020090019721809375,
+      "loss": 3.2276,
+      "step": 35280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020082255387706733,
+      "loss": 3.1601,
+      "step": 35285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020074491799367642,
+      "loss": 3.0885,
+      "step": 35290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000200667289573759,
+      "loss": 3.2486,
+      "step": 35295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020058966862315238,
+      "loss": 3.2406,
+      "step": 35300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020051205514769313,
+      "loss": 3.1888,
+      "step": 35305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020043444915321762,
+      "loss": 3.1852,
+      "step": 35310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020035685064556118,
+      "loss": 3.2038,
+      "step": 35315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002002792596305589,
+      "loss": 3.2482,
+      "step": 35320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002002016761140452,
+      "loss": 3.2151,
+      "step": 35325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020012410010185406,
+      "loss": 3.2906,
+      "step": 35330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002000465315998188,
+      "loss": 3.2794,
+      "step": 35335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000199968970613772,
+      "loss": 3.1682,
+      "step": 35340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019989141714954592,
+      "loss": 3.2599,
+      "step": 35345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001998138712129723,
+      "loss": 3.2017,
+      "step": 35350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019973633280988206,
+      "loss": 3.3271,
+      "step": 35355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001996588019461059,
+      "loss": 3.1123,
+      "step": 35360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019958127862747344,
+      "loss": 3.2482,
+      "step": 35365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019950376285981426,
+      "loss": 3.3316,
+      "step": 35370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019942625464895703,
+      "loss": 3.2091,
+      "step": 35375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019934875400072998,
+      "loss": 3.34,
+      "step": 35380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019927126092096094,
+      "loss": 3.1714,
+      "step": 35385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019919377541547677,
+      "loss": 3.142,
+      "step": 35390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019911629749010408,
+      "loss": 3.2751,
+      "step": 35395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019903882715066875,
+      "loss": 3.2666,
+      "step": 35400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019896136440299622,
+      "loss": 3.3066,
+      "step": 35405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001988839092529113,
+      "loss": 3.296,
+      "step": 35410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019880646170623829,
+      "loss": 3.3092,
+      "step": 35415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019872902176880058,
+      "loss": 3.2801,
+      "step": 35420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019865158944642146,
+      "loss": 3.1055,
+      "step": 35425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001985741647449235,
+      "loss": 3.2495,
+      "step": 35430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001984967476701284,
+      "loss": 3.3053,
+      "step": 35435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019841933822785788,
+      "loss": 3.2339,
+      "step": 35440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001983419364239324,
+      "loss": 3.2144,
+      "step": 35445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019826454226417236,
+      "loss": 3.2215,
+      "step": 35450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019818715575439729,
+      "loss": 3.2162,
+      "step": 35455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019810977690042634,
+      "loss": 3.3478,
+      "step": 35460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001980324057080781,
+      "loss": 3.3434,
+      "step": 35465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019795504218317024,
+      "loss": 3.2074,
+      "step": 35470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001978776863315203,
+      "loss": 3.2835,
+      "step": 35475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019780033815894495,
+      "loss": 3.2004,
+      "step": 35480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019772299767126039,
+      "loss": 3.3079,
+      "step": 35485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019764566487428235,
+      "loss": 3.2486,
+      "step": 35490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001975683397738257,
+      "loss": 3.3032,
+      "step": 35495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001974910223757049,
+      "loss": 3.2286,
+      "step": 35500
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 3.266571283340454,
+      "eval_runtime": 150.2623,
+      "eval_samples_per_second": 12.252,
+      "eval_steps_per_second": 0.772,
+      "step": 35500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019741371268573388,
+      "loss": 3.1576,
+      "step": 35505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001973364107097259,
+      "loss": 3.2607,
+      "step": 35510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019725911645349382,
+      "loss": 3.1451,
+      "step": 35515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001971818299228495,
+      "loss": 3.2446,
+      "step": 35520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019710455112360468,
+      "loss": 3.3296,
+      "step": 35525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019702728006157033,
+      "loss": 3.3272,
+      "step": 35530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019695001674255669,
+      "loss": 3.2667,
+      "step": 35535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019687276117237384,
+      "loss": 3.3462,
+      "step": 35540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019679551335683064,
+      "loss": 3.226,
+      "step": 35545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019671827330173597,
+      "loss": 3.2596,
+      "step": 35550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001966410410128979,
+      "loss": 3.2741,
+      "step": 35555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019656381649612375,
+      "loss": 3.1513,
+      "step": 35560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019648659975722064,
+      "loss": 3.2949,
+      "step": 35565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019640939080199462,
+      "loss": 3.2267,
+      "step": 35570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019633218963625157,
+      "loss": 3.2633,
+      "step": 35575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001962549962657965,
+      "loss": 3.2205,
+      "step": 35580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019617781069643402,
+      "loss": 3.2456,
+      "step": 35585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019610063293396823,
+      "loss": 3.1869,
+      "step": 35590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001960234629842023,
+      "loss": 3.2434,
+      "step": 35595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019594630085293906,
+      "loss": 3.1685,
+      "step": 35600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019586914654598066,
+      "loss": 3.2325,
+      "step": 35605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001957920000691289,
+      "loss": 3.2937,
+      "step": 35610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019571486142818458,
+      "loss": 3.2478,
+      "step": 35615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019563773062894834,
+      "loss": 3.2374,
+      "step": 35620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019556060767721983,
+      "loss": 3.1036,
+      "step": 35625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001954834925787984,
+      "loss": 3.1363,
+      "step": 35630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019540638533948267,
+      "loss": 3.1992,
+      "step": 35635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019532928596507071,
+      "loss": 3.2498,
+      "step": 35640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019525219446136013,
+      "loss": 3.3513,
+      "step": 35645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019517511083414757,
+      "loss": 3.2661,
+      "step": 35650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019509803508922953,
+      "loss": 3.2577,
+      "step": 35655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019502096723240158,
+      "loss": 3.1112,
+      "step": 35660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019494390726945888,
+      "loss": 3.2319,
+      "step": 35665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019486685520619606,
+      "loss": 3.2341,
+      "step": 35670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001947898110484069,
+      "loss": 3.1483,
+      "step": 35675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001947127748018847,
+      "loss": 3.2717,
+      "step": 35680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001946357464724222,
+      "loss": 3.2706,
+      "step": 35685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019455872606581167,
+      "loss": 3.2078,
+      "step": 35690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019448171358784464,
+      "loss": 3.2178,
+      "step": 35695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001944047090443118,
+      "loss": 3.1824,
+      "step": 35700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019432771244100373,
+      "loss": 3.2617,
+      "step": 35705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019425072378371014,
+      "loss": 3.2644,
+      "step": 35710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019417374307822012,
+      "loss": 3.2957,
+      "step": 35715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019409677033032242,
+      "loss": 3.2295,
+      "step": 35720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019401980554580468,
+      "loss": 3.2526,
+      "step": 35725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019394284873045447,
+      "loss": 3.2768,
+      "step": 35730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001938658998900585,
+      "loss": 3.1854,
+      "step": 35735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001937889590304029,
+      "loss": 3.0675,
+      "step": 35740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001937120261572734,
+      "loss": 3.3383,
+      "step": 35745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019363510127645467,
+      "loss": 3.1992,
+      "step": 35750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019355818439373127,
+      "loss": 3.262,
+      "step": 35755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019348127551488682,
+      "loss": 3.2469,
+      "step": 35760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019340437464570462,
+      "loss": 3.2135,
+      "step": 35765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019332748179196724,
+      "loss": 3.2866,
+      "step": 35770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019325059695945648,
+      "loss": 3.2102,
+      "step": 35775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019317372015395366,
+      "loss": 3.2206,
+      "step": 35780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019309685138123962,
+      "loss": 3.1866,
+      "step": 35785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019301999064709455,
+      "loss": 3.2248,
+      "step": 35790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019294313795729793,
+      "loss": 3.1514,
+      "step": 35795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019286629331762868,
+      "loss": 3.1978,
+      "step": 35800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019278945673386504,
+      "loss": 3.2829,
+      "step": 35805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001927126282117848,
+      "loss": 3.2408,
+      "step": 35810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001926358077571652,
+      "loss": 3.2113,
+      "step": 35815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019255899537578262,
+      "loss": 3.223,
+      "step": 35820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019248219107341287,
+      "loss": 3.1289,
+      "step": 35825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001924053948558313,
+      "loss": 3.2393,
+      "step": 35830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019232860672881272,
+      "loss": 3.1435,
+      "step": 35835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019225182669813104,
+      "loss": 3.1946,
+      "step": 35840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001921750547695598,
+      "loss": 3.1769,
+      "step": 35845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019209829094887196,
+      "loss": 3.198,
+      "step": 35850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001920215352418396,
+      "loss": 3.2421,
+      "step": 35855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019194478765423433,
+      "loss": 3.0731,
+      "step": 35860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019186804819182725,
+      "loss": 3.2731,
+      "step": 35865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019179131686038883,
+      "loss": 3.1812,
+      "step": 35870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019171459366568893,
+      "loss": 3.2939,
+      "step": 35875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019163787861349647,
+      "loss": 3.1732,
+      "step": 35880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019156117170958023,
+      "loss": 3.2444,
+      "step": 35885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019148447295970817,
+      "loss": 3.1963,
+      "step": 35890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019140778236964752,
+      "loss": 3.3122,
+      "step": 35895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019133109994516528,
+      "loss": 3.2267,
+      "step": 35900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019125442569202725,
+      "loss": 3.2838,
+      "step": 35905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019117775961599912,
+      "loss": 3.3105,
+      "step": 35910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001911011017228458,
+      "loss": 3.2765,
+      "step": 35915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019102445201833152,
+      "loss": 3.3233,
+      "step": 35920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019094781050822007,
+      "loss": 3.2274,
+      "step": 35925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001908711771982742,
+      "loss": 3.2296,
+      "step": 35930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001907945520942566,
+      "loss": 3.1826,
+      "step": 35935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000190717935201929,
+      "loss": 3.3203,
+      "step": 35940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001906413265270526,
+      "loss": 3.1889,
+      "step": 35945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019056472607538807,
+      "loss": 2.9914,
+      "step": 35950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001904881338526953,
+      "loss": 3.2326,
+      "step": 35955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019041154986473356,
+      "loss": 3.2476,
+      "step": 35960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019033497411726158,
+      "loss": 3.1318,
+      "step": 35965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019025840661603764,
+      "loss": 3.266,
+      "step": 35970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019018184736681906,
+      "loss": 3.1782,
+      "step": 35975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001901052963753628,
+      "loss": 3.1955,
+      "step": 35980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001900287536474249,
+      "loss": 3.2449,
+      "step": 35985
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001899522191887612,
+      "loss": 2.9662,
+      "step": 35990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001898756930051267,
+      "loss": 3.1855,
+      "step": 35995
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018979917510227574,
+      "loss": 3.1923,
+      "step": 36000
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 3.261223554611206,
+      "eval_runtime": 150.469,
+      "eval_samples_per_second": 12.235,
+      "eval_steps_per_second": 0.771,
+      "step": 36000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018972266548596193,
+      "loss": 3.2353,
+      "step": 36005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018964616416193848,
+      "loss": 3.2502,
+      "step": 36010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018956967113595806,
+      "loss": 3.2247,
+      "step": 36015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018949318641377233,
+      "loss": 3.2834,
+      "step": 36020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018941671000113276,
+      "loss": 3.2689,
+      "step": 36025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018934024190378972,
+      "loss": 3.2487,
+      "step": 36030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018926378212749347,
+      "loss": 3.1112,
+      "step": 36035
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018918733067799324,
+      "loss": 3.1531,
+      "step": 36040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001891108875610378,
+      "loss": 3.2902,
+      "step": 36045
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018903445278237543,
+      "loss": 3.278,
+      "step": 36050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018895802634775346,
+      "loss": 3.1309,
+      "step": 36055
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001888816082629188,
+      "loss": 3.1394,
+      "step": 36060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018880519853361775,
+      "loss": 3.2527,
+      "step": 36065
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018872879716559584,
+      "loss": 3.1727,
+      "step": 36070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018865240416459823,
+      "loss": 3.2331,
+      "step": 36075
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018857601953636922,
+      "loss": 3.1652,
+      "step": 36080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018849964328665242,
+      "loss": 3.3314,
+      "step": 36085
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018842327542119096,
+      "loss": 3.2538,
+      "step": 36090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018834691594572746,
+      "loss": 3.2119,
+      "step": 36095
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001882705648660036,
+      "loss": 3.2479,
+      "step": 36100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001881942221877608,
+      "loss": 3.2294,
+      "step": 36105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001881178879167393,
+      "loss": 3.1991,
+      "step": 36110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001880415620586794,
+      "loss": 3.2136,
+      "step": 36115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001879652446193201,
+      "loss": 3.1497,
+      "step": 36120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001878889356044003,
+      "loss": 3.301,
+      "step": 36125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001878126350196581,
+      "loss": 3.1176,
+      "step": 36130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018773634287083072,
+      "loss": 3.3595,
+      "step": 36135
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018766005916365493,
+      "loss": 3.2179,
+      "step": 36140
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000187583783903867,
+      "loss": 3.2545,
+      "step": 36145
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018750751709720247,
+      "loss": 3.2765,
+      "step": 36150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001874312587493961,
+      "loss": 3.2513,
+      "step": 36155
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018735500886618215,
+      "loss": 3.1349,
+      "step": 36160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018727876745329425,
+      "loss": 3.278,
+      "step": 36165
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018720253451646524,
+      "loss": 3.1892,
+      "step": 36170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018712631006142773,
+      "loss": 3.2387,
+      "step": 36175
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018705009409391324,
+      "loss": 3.2239,
+      "step": 36180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001869738866196527,
+      "loss": 3.1069,
+      "step": 36185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018689768764437663,
+      "loss": 3.3351,
+      "step": 36190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018682149717381489,
+      "loss": 3.2367,
+      "step": 36195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018674531521369644,
+      "loss": 3.2287,
+      "step": 36200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018666914176975,
+      "loss": 3.3816,
+      "step": 36205
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018659297684770317,
+      "loss": 3.2868,
+      "step": 36210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018651682045328334,
+      "loss": 3.2455,
+      "step": 36215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018644067259221696,
+      "loss": 3.2449,
+      "step": 36220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018636453327023004,
+      "loss": 3.2737,
+      "step": 36225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018628840249304793,
+      "loss": 3.2192,
+      "step": 36230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001862122802663951,
+      "loss": 3.2535,
+      "step": 36235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018613616659599567,
+      "loss": 3.1544,
+      "step": 36240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018606006148757292,
+      "loss": 3.2894,
+      "step": 36245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001859839649468496,
+      "loss": 3.1255,
+      "step": 36250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018590787697954795,
+      "loss": 3.132,
+      "step": 36255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018583179759138916,
+      "loss": 3.1732,
+      "step": 36260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018575572678809402,
+      "loss": 3.2779,
+      "step": 36265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018567966457538274,
+      "loss": 3.2261,
+      "step": 36270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018560361095897485,
+      "loss": 3.1753,
+      "step": 36275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018552756594458923,
+      "loss": 3.2143,
+      "step": 36280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018545152953794385,
+      "loss": 3.2923,
+      "step": 36285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018537550174475639,
+      "loss": 3.1306,
+      "step": 36290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018529948257074385,
+      "loss": 3.2825,
+      "step": 36295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018522347202162232,
+      "loss": 3.3198,
+      "step": 36300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018514747010310744,
+      "loss": 3.289,
+      "step": 36305
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018507147682091435,
+      "loss": 3.267,
+      "step": 36310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018499549218075703,
+      "loss": 3.3149,
+      "step": 36315
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018491951618834947,
+      "loss": 3.1908,
+      "step": 36320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018484354884940437,
+      "loss": 3.2189,
+      "step": 36325
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018476759016963426,
+      "loss": 3.1788,
+      "step": 36330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018469164015475092,
+      "loss": 3.232,
+      "step": 36335
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018461569881046526,
+      "loss": 3.3037,
+      "step": 36340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018453976614248762,
+      "loss": 3.2038,
+      "step": 36345
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018446384215652786,
+      "loss": 3.124,
+      "step": 36350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018438792685829513,
+      "loss": 3.2536,
+      "step": 36355
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018431202025349783,
+      "loss": 3.1973,
+      "step": 36360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018423612234784364,
+      "loss": 3.2563,
+      "step": 36365
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018416023314703973,
+      "loss": 3.2235,
+      "step": 36370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018408435265679268,
+      "loss": 3.3684,
+      "step": 36375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001840084808828082,
+      "loss": 3.2167,
+      "step": 36380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001839326178307916,
+      "loss": 3.2903,
+      "step": 36385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018385676350644723,
+      "loss": 3.0828,
+      "step": 36390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001837809179154791,
+      "loss": 3.3216,
+      "step": 36395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001837050810635902,
+      "loss": 3.2892,
+      "step": 36400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018362925295648322,
+      "loss": 3.1734,
+      "step": 36405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018355343359986018,
+      "loss": 3.1735,
+      "step": 36410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000183477622999422,
+      "loss": 3.1576,
+      "step": 36415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018340182116086947,
+      "loss": 3.2127,
+      "step": 36420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018332602808990238,
+      "loss": 3.1634,
+      "step": 36425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018325024379222006,
+      "loss": 3.3005,
+      "step": 36430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018317446827352115,
+      "loss": 3.2287,
+      "step": 36435
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018309870153950347,
+      "loss": 3.23,
+      "step": 36440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018302294359586426,
+      "loss": 3.208,
+      "step": 36445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001829471944483002,
+      "loss": 3.2683,
+      "step": 36450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018287145410250723,
+      "loss": 3.1638,
+      "step": 36455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018279572256418077,
+      "loss": 3.2345,
+      "step": 36460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018271999983901515,
+      "loss": 3.2936,
+      "step": 36465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018264428593270452,
+      "loss": 3.2147,
+      "step": 36470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001825685808509422,
+      "loss": 3.1711,
+      "step": 36475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018249288459942073,
+      "loss": 3.1448,
+      "step": 36480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001824171971838322,
+      "loss": 3.2079,
+      "step": 36485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018234151860986773,
+      "loss": 3.1737,
+      "step": 36490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018226584888321812,
+      "loss": 3.2266,
+      "step": 36495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018219018800957334,
+      "loss": 3.154,
+      "step": 36500
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 3.2557082176208496,
+      "eval_runtime": 151.364,
+      "eval_samples_per_second": 12.163,
+      "eval_steps_per_second": 0.766,
+      "step": 36500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018211453599462256,
+      "loss": 3.2151,
+      "step": 36505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018203889284405467,
+      "loss": 3.3157,
+      "step": 36510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018196325856355736,
+      "loss": 3.1872,
+      "step": 36515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018188763315881817,
+      "loss": 3.2263,
+      "step": 36520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018181201663552353,
+      "loss": 3.2157,
+      "step": 36525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018173640899935958,
+      "loss": 3.2489,
+      "step": 36530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001816608102560117,
+      "loss": 3.2881,
+      "step": 36535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001815852204111643,
+      "loss": 3.2426,
+      "step": 36540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018150963947050145,
+      "loss": 3.2455,
+      "step": 36545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018143406743970642,
+      "loss": 3.3231,
+      "step": 36550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018135850432446197,
+      "loss": 3.3465,
+      "step": 36555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018128295013044993,
+      "loss": 3.2252,
+      "step": 36560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001812074048633517,
+      "loss": 3.2743,
+      "step": 36565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018113186852884773,
+      "loss": 3.2134,
+      "step": 36570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018105634113261805,
+      "loss": 3.1525,
+      "step": 36575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018098082268034197,
+      "loss": 3.2405,
+      "step": 36580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018090531317769805,
+      "loss": 3.1579,
+      "step": 36585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001808298126303643,
+      "loss": 3.3345,
+      "step": 36590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018075432104401782,
+      "loss": 3.2023,
+      "step": 36595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018067883842433534,
+      "loss": 3.3086,
+      "step": 36600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018060336477699262,
+      "loss": 3.2832,
+      "step": 36605
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018052790010766496,
+      "loss": 3.311,
+      "step": 36610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018045244442202707,
+      "loss": 3.263,
+      "step": 36615
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018037699772575262,
+      "loss": 3.2083,
+      "step": 36620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001803015600245148,
+      "loss": 3.3219,
+      "step": 36625
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018022613132398624,
+      "loss": 3.271,
+      "step": 36630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018015071162983884,
+      "loss": 3.1227,
+      "step": 36635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018007530094774375,
+      "loss": 3.2264,
+      "step": 36640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017999989928337132,
+      "loss": 3.2169,
+      "step": 36645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017992450664239149,
+      "loss": 3.1578,
+      "step": 36650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001798491230304734,
+      "loss": 3.1818,
+      "step": 36655
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017977374845328547,
+      "loss": 3.1602,
+      "step": 36660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001796983829164956,
+      "loss": 3.1913,
+      "step": 36665
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001796230264257707,
+      "loss": 3.3033,
+      "step": 36670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017954767898677726,
+      "loss": 3.2693,
+      "step": 36675
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017947234060518115,
+      "loss": 3.2891,
+      "step": 36680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017939701128664727,
+      "loss": 3.2005,
+      "step": 36685
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017932169103684016,
+      "loss": 3.1168,
+      "step": 36690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001792463798614233,
+      "loss": 3.2606,
+      "step": 36695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001791710777660599,
+      "loss": 3.1485,
+      "step": 36700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017909578475641213,
+      "loss": 3.0826,
+      "step": 36705
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017902050083814174,
+      "loss": 3.2234,
+      "step": 36710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017894522601690974,
+      "loss": 3.36,
+      "step": 36715
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017886996029837634,
+      "loss": 3.2038,
+      "step": 36720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001787947036882011,
+      "loss": 3.1815,
+      "step": 36725
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017871945619204294,
+      "loss": 3.265,
+      "step": 36730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001786442178155602,
+      "loss": 3.2302,
+      "step": 36735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017856898856441036,
+      "loss": 3.1937,
+      "step": 36740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017849376844425024,
+      "loss": 3.1804,
+      "step": 36745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017841855746073595,
+      "loss": 3.2216,
+      "step": 36750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017834335561952306,
+      "loss": 3.3077,
+      "step": 36755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001782681629262664,
+      "loss": 3.1248,
+      "step": 36760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001781929793866201,
+      "loss": 3.1452,
+      "step": 36765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017811780500623733,
+      "loss": 3.2719,
+      "step": 36770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000178042639790771,
+      "loss": 3.2455,
+      "step": 36775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017796748374587323,
+      "loss": 3.1196,
+      "step": 36780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001778923368771952,
+      "loss": 3.329,
+      "step": 36785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001778171991903876,
+      "loss": 3.1396,
+      "step": 36790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017774207069110064,
+      "loss": 3.2559,
+      "step": 36795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017766695138498332,
+      "loss": 3.2072,
+      "step": 36800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017759184127768424,
+      "loss": 3.3362,
+      "step": 36805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017751674037485136,
+      "loss": 3.2532,
+      "step": 36810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017744164868213193,
+      "loss": 3.2233,
+      "step": 36815
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017736656620517246,
+      "loss": 3.2606,
+      "step": 36820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001772914929496187,
+      "loss": 3.1516,
+      "step": 36825
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017721642892111573,
+      "loss": 3.1829,
+      "step": 36830
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017714137412530806,
+      "loss": 3.2415,
+      "step": 36835
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017706632856783952,
+      "loss": 3.2393,
+      "step": 36840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017699129225435303,
+      "loss": 3.192,
+      "step": 36845
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017691626519049085,
+      "loss": 3.3051,
+      "step": 36850
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017684124738189476,
+      "loss": 3.2575,
+      "step": 36855
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017676623883420574,
+      "loss": 3.1901,
+      "step": 36860
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017669123955306394,
+      "loss": 3.1218,
+      "step": 36865
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017661624954410905,
+      "loss": 3.2743,
+      "step": 36870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001765412688129798,
+      "loss": 3.2173,
+      "step": 36875
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001764662973653145,
+      "loss": 3.1673,
+      "step": 36880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017639133520675043,
+      "loss": 3.3145,
+      "step": 36885
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017631638234292448,
+      "loss": 3.2983,
+      "step": 36890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017624143877947282,
+      "loss": 3.1276,
+      "step": 36895
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017616650452203066,
+      "loss": 3.0952,
+      "step": 36900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017609157957623265,
+      "loss": 3.1819,
+      "step": 36905
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017601666394771287,
+      "loss": 3.13,
+      "step": 36910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017594175764210453,
+      "loss": 3.1295,
+      "step": 36915
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017586686066504035,
+      "loss": 3.2718,
+      "step": 36920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017579197302215204,
+      "loss": 3.3466,
+      "step": 36925
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001757170947190707,
+      "loss": 3.1818,
+      "step": 36930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017564222576142694,
+      "loss": 3.2946,
+      "step": 36935
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001755673661548505,
+      "loss": 3.2415,
+      "step": 36940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001754925159049705,
+      "loss": 3.2438,
+      "step": 36945
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001754176750174151,
+      "loss": 3.2137,
+      "step": 36950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017534284349781203,
+      "loss": 3.1935,
+      "step": 36955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017526802135178836,
+      "loss": 3.281,
+      "step": 36960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017519320858497018,
+      "loss": 3.185,
+      "step": 36965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001751184052029832,
+      "loss": 3.2558,
+      "step": 36970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001750436112114521,
+      "loss": 3.3523,
+      "step": 36975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017496882661600095,
+      "loss": 3.2183,
+      "step": 36980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017489405142225332,
+      "loss": 3.308,
+      "step": 36985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001748192856358319,
+      "loss": 3.2556,
+      "step": 36990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017474452926235864,
+      "loss": 3.2036,
+      "step": 36995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017466978230745486,
+      "loss": 3.2368,
+      "step": 37000
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 3.2513468265533447,
+      "eval_runtime": 149.7793,
+      "eval_samples_per_second": 12.291,
+      "eval_steps_per_second": 0.774,
+      "step": 37000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017459504477674104,
+      "loss": 3.2079,
+      "step": 37005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017452031667583728,
+      "loss": 3.0015,
+      "step": 37010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017444559801036262,
+      "loss": 3.2196,
+      "step": 37015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017437088878593542,
+      "loss": 3.2114,
+      "step": 37020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001742961890081738,
+      "loss": 3.3074,
+      "step": 37025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017422149868269427,
+      "loss": 3.1725,
+      "step": 37030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017414681781511356,
+      "loss": 3.2638,
+      "step": 37035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017407214641104715,
+      "loss": 3.1967,
+      "step": 37040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017399748447610987,
+      "loss": 3.1839,
+      "step": 37045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017392283201591625,
+      "loss": 3.251,
+      "step": 37050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001738481890360794,
+      "loss": 3.1605,
+      "step": 37055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001737735555422121,
+      "loss": 3.2344,
+      "step": 37060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017369893153992662,
+      "loss": 3.1954,
+      "step": 37065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017362431703483425,
+      "loss": 3.2549,
+      "step": 37070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017354971203254553,
+      "loss": 3.2404,
+      "step": 37075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017347511653867045,
+      "loss": 3.249,
+      "step": 37080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017340053055881803,
+      "loss": 3.2248,
+      "step": 37085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017332595409859706,
+      "loss": 3.2577,
+      "step": 37090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001732513871636151,
+      "loss": 3.249,
+      "step": 37095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017317682975947926,
+      "loss": 3.2418,
+      "step": 37100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001731022818917959,
+      "loss": 3.1526,
+      "step": 37105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017302774356617052,
+      "loss": 3.2873,
+      "step": 37110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017295321478820805,
+      "loss": 3.1383,
+      "step": 37115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017287869556351283,
+      "loss": 3.2247,
+      "step": 37120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001728041858976882,
+      "loss": 3.1668,
+      "step": 37125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017272968579633695,
+      "loss": 3.2901,
+      "step": 37130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017265519526506108,
+      "loss": 3.2611,
+      "step": 37135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017258071430946174,
+      "loss": 3.1702,
+      "step": 37140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001725062429351398,
+      "loss": 3.2486,
+      "step": 37145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017243178114769505,
+      "loss": 3.2305,
+      "step": 37150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017235732895272648,
+      "loss": 3.2179,
+      "step": 37155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001722828863558327,
+      "loss": 3.2362,
+      "step": 37160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017220845336261118,
+      "loss": 3.2262,
+      "step": 37165
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017213402997865915,
+      "loss": 3.1945,
+      "step": 37170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001720596162095729,
+      "loss": 3.2624,
+      "step": 37175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001719852120609476,
+      "loss": 3.0769,
+      "step": 37180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017191081753837836,
+      "loss": 3.1758,
+      "step": 37185
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017183643264745925,
+      "loss": 3.2337,
+      "step": 37190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017176205739378344,
+      "loss": 3.1392,
+      "step": 37195
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001716876917829439,
+      "loss": 3.2752,
+      "step": 37200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017161333582053217,
+      "loss": 3.3017,
+      "step": 37205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001715389895121397,
+      "loss": 3.1933,
+      "step": 37210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017146465286335688,
+      "loss": 3.2897,
+      "step": 37215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017139032587977333,
+      "loss": 3.2704,
+      "step": 37220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017131600856697834,
+      "loss": 3.221,
+      "step": 37225
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017124170093055983,
+      "loss": 3.2712,
+      "step": 37230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017116740297610565,
+      "loss": 3.1784,
+      "step": 37235
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001710931147092025,
+      "loss": 3.1287,
+      "step": 37240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017101883613543636,
+      "loss": 3.3075,
+      "step": 37245
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017094456726039284,
+      "loss": 3.2108,
+      "step": 37250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017087030808965654,
+      "loss": 3.167,
+      "step": 37255
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017079605862881114,
+      "loss": 3.0637,
+      "step": 37260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017072181888344004,
+      "loss": 3.2903,
+      "step": 37265
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001706475888591256,
+      "loss": 3.3106,
+      "step": 37270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001705733685614495,
+      "loss": 3.2914,
+      "step": 37275
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017049915799599298,
+      "loss": 3.2675,
+      "step": 37280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017042495716833588,
+      "loss": 3.2125,
+      "step": 37285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017035076608405803,
+      "loss": 3.197,
+      "step": 37290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017027658474873817,
+      "loss": 3.1445,
+      "step": 37295
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017020241316795416,
+      "loss": 3.2512,
+      "step": 37300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001701282513472837,
+      "loss": 3.2431,
+      "step": 37305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017005409929230297,
+      "loss": 3.1504,
+      "step": 37310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016997995700858808,
+      "loss": 3.185,
+      "step": 37315
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016990582450171412,
+      "loss": 3.0925,
+      "step": 37320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016983170177725532,
+      "loss": 3.0949,
+      "step": 37325
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016975758884078566,
+      "loss": 3.2816,
+      "step": 37330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016968348569787773,
+      "loss": 3.3073,
+      "step": 37335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001696093923541037,
+      "loss": 3.3084,
+      "step": 37340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001695353088150352,
+      "loss": 3.1617,
+      "step": 37345
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016946123508624288,
+      "loss": 3.2592,
+      "step": 37350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001693871711732967,
+      "loss": 3.2328,
+      "step": 37355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016931311708176587,
+      "loss": 3.217,
+      "step": 37360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001692390728172187,
+      "loss": 3.2719,
+      "step": 37365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016916503838522331,
+      "loss": 3.069,
+      "step": 37370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001690910137913465,
+      "loss": 3.3553,
+      "step": 37375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016901699904115454,
+      "loss": 3.1695,
+      "step": 37380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016894299414021294,
+      "loss": 3.2769,
+      "step": 37385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016886899909408643,
+      "loss": 3.126,
+      "step": 37390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016879501390833928,
+      "loss": 3.265,
+      "step": 37395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001687210385885347,
+      "loss": 3.2026,
+      "step": 37400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016864707314023517,
+      "loss": 3.2575,
+      "step": 37405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001685731175690026,
+      "loss": 3.2137,
+      "step": 37410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016849917188039803,
+      "loss": 3.2626,
+      "step": 37415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001684252360799817,
+      "loss": 3.2544,
+      "step": 37420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016835131017331337,
+      "loss": 3.3139,
+      "step": 37425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016827739416595187,
+      "loss": 3.2438,
+      "step": 37430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016820348806345525,
+      "loss": 3.1817,
+      "step": 37435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001681295918713809,
+      "loss": 3.2638,
+      "step": 37440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016805570559528526,
+      "loss": 3.2733,
+      "step": 37445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001679818292407245,
+      "loss": 3.2396,
+      "step": 37450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001679079628132536,
+      "loss": 3.2589,
+      "step": 37455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016783410631842691,
+      "loss": 3.2473,
+      "step": 37460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001677602597617981,
+      "loss": 3.19,
+      "step": 37465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001676864231489199,
+      "loss": 3.2841,
+      "step": 37470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016761259648534477,
+      "loss": 3.3538,
+      "step": 37475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016753877977662384,
+      "loss": 3.3771,
+      "step": 37480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001674649730283079,
+      "loss": 3.1082,
+      "step": 37485
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016739117624594665,
+      "loss": 3.1594,
+      "step": 37490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001673173894350894,
+      "loss": 3.2678,
+      "step": 37495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016724361260128434,
+      "loss": 3.2161,
+      "step": 37500
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 3.246263027191162,
+      "eval_runtime": 150.2834,
+      "eval_samples_per_second": 12.25,
+      "eval_steps_per_second": 0.772,
+      "step": 37500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001671698457500794,
+      "loss": 3.1662,
+      "step": 37505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016709608888702124,
+      "loss": 3.2584,
+      "step": 37510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016702234201765613,
+      "loss": 3.2162,
+      "step": 37515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016694860514752938,
+      "loss": 3.1176,
+      "step": 37520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016687487828218557,
+      "loss": 3.1931,
+      "step": 37525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016680116142716873,
+      "loss": 3.2113,
+      "step": 37530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016672745458802193,
+      "loss": 3.2477,
+      "step": 37535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016665375777028755,
+      "loss": 3.1498,
+      "step": 37540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016658007097950715,
+      "loss": 3.3016,
+      "step": 37545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016650639422122158,
+      "loss": 3.194,
+      "step": 37550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001664327275009711,
+      "loss": 3.137,
+      "step": 37555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016635907082429514,
+      "loss": 3.2968,
+      "step": 37560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001662854241967319,
+      "loss": 3.2957,
+      "step": 37565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016621178762381964,
+      "loss": 3.2532,
+      "step": 37570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016613816111109525,
+      "loss": 3.1561,
+      "step": 37575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000166064544664095,
+      "loss": 3.1695,
+      "step": 37580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016599093828835484,
+      "loss": 3.1958,
+      "step": 37585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016591734198940906,
+      "loss": 3.18,
+      "step": 37590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016584375577279219,
+      "loss": 3.1963,
+      "step": 37595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016577017964403728,
+      "loss": 3.1112,
+      "step": 37600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016569661360867687,
+      "loss": 3.1904,
+      "step": 37605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016562305767224304,
+      "loss": 3.2197,
+      "step": 37610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016554951184026646,
+      "loss": 3.2805,
+      "step": 37615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016547597611827746,
+      "loss": 3.237,
+      "step": 37620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016540245051180572,
+      "loss": 3.1968,
+      "step": 37625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016532893502637993,
+      "loss": 3.2983,
+      "step": 37630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016525542966752804,
+      "loss": 3.2977,
+      "step": 37635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016518193444077728,
+      "loss": 3.2517,
+      "step": 37640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000165108449351654,
+      "loss": 3.1764,
+      "step": 37645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001650349744056841,
+      "loss": 3.1407,
+      "step": 37650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001649615096083925,
+      "loss": 3.3731,
+      "step": 37655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016488805496530328,
+      "loss": 3.2043,
+      "step": 37660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001648146104819399,
+      "loss": 3.214,
+      "step": 37665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001647411761638249,
+      "loss": 3.1209,
+      "step": 37670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016466775201648032,
+      "loss": 3.2196,
+      "step": 37675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016459433804542727,
+      "loss": 3.3123,
+      "step": 37680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016452093425618606,
+      "loss": 3.245,
+      "step": 37685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016444754065427625,
+      "loss": 3.1629,
+      "step": 37690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016437415724521672,
+      "loss": 3.2326,
+      "step": 37695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016430078403452535,
+      "loss": 3.1992,
+      "step": 37700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016422742102771969,
+      "loss": 3.2132,
+      "step": 37705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016415406823031616,
+      "loss": 3.179,
+      "step": 37710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016408072564783047,
+      "loss": 3.2493,
+      "step": 37715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016400739328577767,
+      "loss": 3.203,
+      "step": 37720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016393407114967184,
+      "loss": 3.1359,
+      "step": 37725
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016386075924502657,
+      "loss": 3.2161,
+      "step": 37730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001637874575773546,
+      "loss": 3.1806,
+      "step": 37735
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016371416615216765,
+      "loss": 3.2406,
+      "step": 37740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016364088497497706,
+      "loss": 3.1397,
+      "step": 37745
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016356761405129287,
+      "loss": 3.1923,
+      "step": 37750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001634943533866251,
+      "loss": 3.3604,
+      "step": 37755
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016342110298648232,
+      "loss": 3.2568,
+      "step": 37760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016334786285637264,
+      "loss": 3.2019,
+      "step": 37765
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016327463300180335,
+      "loss": 3.1786,
+      "step": 37770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016320141342828095,
+      "loss": 3.0524,
+      "step": 37775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016312820414131107,
+      "loss": 3.3086,
+      "step": 37780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016305500514639888,
+      "loss": 3.1939,
+      "step": 37785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016298181644904854,
+      "loss": 3.307,
+      "step": 37790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016290863805476331,
+      "loss": 3.2686,
+      "step": 37795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016283546996904595,
+      "loss": 3.1524,
+      "step": 37800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001627623121973982,
+      "loss": 3.1401,
+      "step": 37805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001626891647453213,
+      "loss": 3.1872,
+      "step": 37810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016261602761831554,
+      "loss": 3.4207,
+      "step": 37815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016254290082188044,
+      "loss": 3.2811,
+      "step": 37820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016246978436151472,
+      "loss": 3.2386,
+      "step": 37825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016239667824271626,
+      "loss": 3.2014,
+      "step": 37830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001623235824709825,
+      "loss": 3.1844,
+      "step": 37835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016225049705180988,
+      "loss": 3.1639,
+      "step": 37840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016217742199069375,
+      "loss": 3.2266,
+      "step": 37845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016210435729312918,
+      "loss": 3.1546,
+      "step": 37850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001620313029646103,
+      "loss": 3.2589,
+      "step": 37855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016195825901063024,
+      "loss": 3.1717,
+      "step": 37860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016188522543668187,
+      "loss": 3.1791,
+      "step": 37865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016181220224825647,
+      "loss": 3.2002,
+      "step": 37870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016173918945084538,
+      "loss": 3.0784,
+      "step": 37875
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016166618704993867,
+      "loss": 3.2606,
+      "step": 37880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016159319505102565,
+      "loss": 3.2025,
+      "step": 37885
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016152021345959528,
+      "loss": 3.119,
+      "step": 37890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016144724228113488,
+      "loss": 3.3311,
+      "step": 37895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016137428152113197,
+      "loss": 3.2303,
+      "step": 37900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001613013311850726,
+      "loss": 3.306,
+      "step": 37905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016122839127844218,
+      "loss": 3.2077,
+      "step": 37910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001611554618067258,
+      "loss": 3.1467,
+      "step": 37915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000161082542775407,
+      "loss": 3.2522,
+      "step": 37920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001610096341899689,
+      "loss": 3.2779,
+      "step": 37925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016093673605589416,
+      "loss": 3.1941,
+      "step": 37930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016086384837866414,
+      "loss": 3.1619,
+      "step": 37935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016079097116375958,
+      "loss": 3.2503,
+      "step": 37940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016071810441666077,
+      "loss": 3.2323,
+      "step": 37945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016064524814284644,
+      "loss": 3.1909,
+      "step": 37950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016057240234779543,
+      "loss": 3.1922,
+      "step": 37955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001604995670369852,
+      "loss": 3.2015,
+      "step": 37960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016042674221589242,
+      "loss": 3.1244,
+      "step": 37965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016035392788999364,
+      "loss": 3.2906,
+      "step": 37970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016028112406476353,
+      "loss": 3.2155,
+      "step": 37975
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016020833074567695,
+      "loss": 3.2215,
+      "step": 37980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001601355479382075,
+      "loss": 3.2019,
+      "step": 37985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016006277564782796,
+      "loss": 3.1993,
+      "step": 37990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015999001388001074,
+      "loss": 3.3097,
+      "step": 37995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015991726264022677,
+      "loss": 3.1445,
+      "step": 38000
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 3.2400145530700684,
+      "eval_runtime": 149.8729,
+      "eval_samples_per_second": 12.284,
+      "eval_steps_per_second": 0.774,
+      "step": 38000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001598445219339467,
+      "loss": 3.2372,
+      "step": 38005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015977179176664035,
+      "loss": 3.2157,
+      "step": 38010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015969907214377662,
+      "loss": 3.2457,
+      "step": 38015
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015962636307082363,
+      "loss": 3.2293,
+      "step": 38020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015955366455324874,
+      "loss": 3.1265,
+      "step": 38025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001594809765965184,
+      "loss": 3.2626,
+      "step": 38030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015940829920609858,
+      "loss": 3.2296,
+      "step": 38035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015933563238745412,
+      "loss": 3.2178,
+      "step": 38040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015926297614604924,
+      "loss": 3.2407,
+      "step": 38045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015919033048734726,
+      "loss": 3.2555,
+      "step": 38050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015911769541681066,
+      "loss": 3.2065,
+      "step": 38055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015904507093990152,
+      "loss": 3.2034,
+      "step": 38060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015897245706208063,
+      "loss": 3.3155,
+      "step": 38065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015889985378880818,
+      "loss": 3.2745,
+      "step": 38070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015882726112554364,
+      "loss": 3.2874,
+      "step": 38075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015875467907774558,
+      "loss": 3.1474,
+      "step": 38080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015868210765087167,
+      "loss": 3.2925,
+      "step": 38085
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015860954685037912,
+      "loss": 3.1178,
+      "step": 38090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000158536996681724,
+      "loss": 3.2575,
+      "step": 38095
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001584644571503618,
+      "loss": 3.2266,
+      "step": 38100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000158391928261747,
+      "loss": 3.0909,
+      "step": 38105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015831941002133343,
+      "loss": 3.1911,
+      "step": 38110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001582469024345742,
+      "loss": 3.1309,
+      "step": 38115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015817440550692152,
+      "loss": 3.1826,
+      "step": 38120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001581019192438265,
+      "loss": 3.212,
+      "step": 38125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015802944365074005,
+      "loss": 3.3265,
+      "step": 38130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015795697873311184,
+      "loss": 3.1416,
+      "step": 38135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015788452449639077,
+      "loss": 3.0383,
+      "step": 38140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015781208094602537,
+      "loss": 3.159,
+      "step": 38145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001577396480874625,
+      "loss": 3.2361,
+      "step": 38150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001576672259261492,
+      "loss": 3.2279,
+      "step": 38155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015759481446753098,
+      "loss": 3.248,
+      "step": 38160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015752241371705286,
+      "loss": 3.1824,
+      "step": 38165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015745002368015913,
+      "loss": 3.1641,
+      "step": 38170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015737764436229306,
+      "loss": 3.1412,
+      "step": 38175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001573052757688972,
+      "loss": 3.2713,
+      "step": 38180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015723291790541334,
+      "loss": 3.226,
+      "step": 38185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015716057077728223,
+      "loss": 3.3351,
+      "step": 38190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001570882343899443,
+      "loss": 3.2517,
+      "step": 38195
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015701590874883883,
+      "loss": 3.1853,
+      "step": 38200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015694359385940403,
+      "loss": 3.1761,
+      "step": 38205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001568712897270779,
+      "loss": 3.1338,
+      "step": 38210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001567989963572973,
+      "loss": 3.2637,
+      "step": 38215
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015672671375549814,
+      "loss": 3.2493,
+      "step": 38220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015665444192711604,
+      "loss": 3.2718,
+      "step": 38225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015658218087758507,
+      "loss": 3.2829,
+      "step": 38230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015650993061233924,
+      "loss": 3.3045,
+      "step": 38235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015643769113681124,
+      "loss": 3.2175,
+      "step": 38240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000156365462456433,
+      "loss": 3.2631,
+      "step": 38245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015629324457663605,
+      "loss": 3.3236,
+      "step": 38250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015622103750285047,
+      "loss": 3.2581,
+      "step": 38255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015614884124050607,
+      "loss": 3.287,
+      "step": 38260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015607665579503164,
+      "loss": 3.3278,
+      "step": 38265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015600448117185492,
+      "loss": 3.1982,
+      "step": 38270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015593231737640352,
+      "loss": 3.2156,
+      "step": 38275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015586016441410341,
+      "loss": 3.2382,
+      "step": 38280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015578802229038007,
+      "loss": 3.2535,
+      "step": 38285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015571589101065855,
+      "loss": 3.3024,
+      "step": 38290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015564377058036258,
+      "loss": 3.1432,
+      "step": 38295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015557166100491518,
+      "loss": 3.1777,
+      "step": 38300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015549956228973875,
+      "loss": 3.2004,
+      "step": 38305
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015542747444025455,
+      "loss": 3.0625,
+      "step": 38310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015535539746188348,
+      "loss": 3.1938,
+      "step": 38315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015528333136004523,
+      "loss": 3.221,
+      "step": 38320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015521127614015877,
+      "loss": 3.2017,
+      "step": 38325
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015513923180764233,
+      "loss": 3.2288,
+      "step": 38330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015506719836791313,
+      "loss": 3.2243,
+      "step": 38335
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015499517582638798,
+      "loss": 3.232,
+      "step": 38340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015492316418848248,
+      "loss": 3.1439,
+      "step": 38345
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001548511634596115,
+      "loss": 3.1259,
+      "step": 38350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015477917364518915,
+      "loss": 3.1856,
+      "step": 38355
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001547071947506287,
+      "loss": 3.2,
+      "step": 38360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015463522678134248,
+      "loss": 3.1343,
+      "step": 38365
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015456326974274233,
+      "loss": 3.2869,
+      "step": 38370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015449132364023897,
+      "loss": 3.0858,
+      "step": 38375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015441938847924233,
+      "loss": 3.1788,
+      "step": 38380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001543474642651616,
+      "loss": 3.2329,
+      "step": 38385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015427555100340498,
+      "loss": 3.196,
+      "step": 38390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001542036486993802,
+      "loss": 3.1304,
+      "step": 38395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015413175735849388,
+      "loss": 3.2239,
+      "step": 38400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015405987698615184,
+      "loss": 3.0827,
+      "step": 38405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001539880075877591,
+      "loss": 3.2181,
+      "step": 38410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015391614916871978,
+      "loss": 3.2241,
+      "step": 38415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015384430173443747,
+      "loss": 3.2218,
+      "step": 38420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015377246529031465,
+      "loss": 3.098,
+      "step": 38425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015370063984175304,
+      "loss": 3.2373,
+      "step": 38430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015362882539415352,
+      "loss": 3.2509,
+      "step": 38435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001535570219529162,
+      "loss": 3.1837,
+      "step": 38440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001534852295234402,
+      "loss": 3.2769,
+      "step": 38445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001534134481111242,
+      "loss": 3.1648,
+      "step": 38450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015334167772136566,
+      "loss": 3.1912,
+      "step": 38455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015326991835956132,
+      "loss": 3.1258,
+      "step": 38460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015319817003110715,
+      "loss": 3.2684,
+      "step": 38465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015312643274139812,
+      "loss": 3.2462,
+      "step": 38470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001530547064958288,
+      "loss": 3.1875,
+      "step": 38475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015298299129979238,
+      "loss": 3.2456,
+      "step": 38480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015291128715868164,
+      "loss": 3.1945,
+      "step": 38485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015283959407788825,
+      "loss": 3.2784,
+      "step": 38490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015276791206280307,
+      "loss": 3.0896,
+      "step": 38495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001526962411188165,
+      "loss": 3.1892,
+      "step": 38500
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 3.235760450363159,
+      "eval_runtime": 149.9771,
+      "eval_samples_per_second": 12.275,
+      "eval_steps_per_second": 0.773,
+      "step": 38500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001526245812513178,
+      "loss": 3.1669,
+      "step": 38505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015255293246569504,
+      "loss": 3.2019,
+      "step": 38510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015248129476733626,
+      "loss": 3.2363,
+      "step": 38515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015240966816162807,
+      "loss": 3.2336,
+      "step": 38520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001523380526539563,
+      "loss": 3.2384,
+      "step": 38525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001522664482497065,
+      "loss": 3.1823,
+      "step": 38530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015219485495426242,
+      "loss": 3.2383,
+      "step": 38535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015212327277300784,
+      "loss": 3.2228,
+      "step": 38540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015205170171132536,
+      "loss": 3.0597,
+      "step": 38545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001519801417745965,
+      "loss": 3.2417,
+      "step": 38550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015190859296820264,
+      "loss": 3.2426,
+      "step": 38555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001518370552975234,
+      "loss": 3.1934,
+      "step": 38560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015176552876793847,
+      "loss": 3.2366,
+      "step": 38565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000151694013384826,
+      "loss": 3.1862,
+      "step": 38570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015162250915356357,
+      "loss": 3.2587,
+      "step": 38575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015155101607952827,
+      "loss": 3.1882,
+      "step": 38580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015147953416809565,
+      "loss": 3.2082,
+      "step": 38585
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015140806342464077,
+      "loss": 3.204,
+      "step": 38590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015133660385453814,
+      "loss": 3.2401,
+      "step": 38595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015126515546316102,
+      "loss": 3.1105,
+      "step": 38600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015119371825588192,
+      "loss": 3.2315,
+      "step": 38605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015112229223807262,
+      "loss": 3.2599,
+      "step": 38610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015105087741510386,
+      "loss": 3.1626,
+      "step": 38615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015097947379234583,
+      "loss": 3.0691,
+      "step": 38620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015090808137516767,
+      "loss": 3.1392,
+      "step": 38625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001508367001689377,
+      "loss": 3.2198,
+      "step": 38630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015076533017902347,
+      "loss": 3.1879,
+      "step": 38635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015069397141079154,
+      "loss": 3.228,
+      "step": 38640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015062262386960772,
+      "loss": 3.2194,
+      "step": 38645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001505512875608371,
+      "loss": 3.2814,
+      "step": 38650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015047996248984376,
+      "loss": 3.1777,
+      "step": 38655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015040864866199097,
+      "loss": 3.221,
+      "step": 38660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015033734608264115,
+      "loss": 3.2044,
+      "step": 38665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015026605475715581,
+      "loss": 3.2661,
+      "step": 38670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015019477469089586,
+      "loss": 3.2046,
+      "step": 38675
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001501235058892211,
+      "loss": 3.1087,
+      "step": 38680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015005224835749063,
+      "loss": 3.2581,
+      "step": 38685
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014998100210106258,
+      "loss": 3.1716,
+      "step": 38690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014990976712529424,
+      "loss": 3.1427,
+      "step": 38695
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014983854343554232,
+      "loss": 3.1025,
+      "step": 38700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001497673310371624,
+      "loss": 3.1843,
+      "step": 38705
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001496961299355092,
+      "loss": 3.2923,
+      "step": 38710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001496249401359368,
+      "loss": 3.2623,
+      "step": 38715
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014955376164379818,
+      "loss": 3.1085,
+      "step": 38720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001494825944644456,
+      "loss": 3.2231,
+      "step": 38725
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001494114386032306,
+      "loss": 3.2339,
+      "step": 38730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014934029406550367,
+      "loss": 3.1435,
+      "step": 38735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001492691608566145,
+      "loss": 3.2924,
+      "step": 38740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014919803898191198,
+      "loss": 3.3259,
+      "step": 38745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014912692844674393,
+      "loss": 3.2163,
+      "step": 38750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001490558292564578,
+      "loss": 3.1284,
+      "step": 38755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001489847414163997,
+      "loss": 3.2856,
+      "step": 38760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001489136649319151,
+      "loss": 3.2836,
+      "step": 38765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014884259980834866,
+      "loss": 3.1405,
+      "step": 38770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001487715460510439,
+      "loss": 3.1253,
+      "step": 38775
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001487005036653439,
+      "loss": 3.2167,
+      "step": 38780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001486294726565908,
+      "loss": 3.1719,
+      "step": 38785
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014855845303012535,
+      "loss": 3.1742,
+      "step": 38790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001484874447912882,
+      "loss": 3.1847,
+      "step": 38795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014841644794541872,
+      "loss": 3.2533,
+      "step": 38800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014834546249785542,
+      "loss": 3.1358,
+      "step": 38805
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014827448845393634,
+      "loss": 3.2148,
+      "step": 38810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014820352581899787,
+      "loss": 3.1691,
+      "step": 38815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014813257459837649,
+      "loss": 3.1114,
+      "step": 38820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014806163479740714,
+      "loss": 3.1567,
+      "step": 38825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014799070642142412,
+      "loss": 3.189,
+      "step": 38830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014791978947576115,
+      "loss": 3.1744,
+      "step": 38835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014784888396575038,
+      "loss": 3.2016,
+      "step": 38840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014777798989672387,
+      "loss": 3.1499,
+      "step": 38845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014770710727401241,
+      "loss": 3.2737,
+      "step": 38850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014763623610294588,
+      "loss": 3.2374,
+      "step": 38855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014756537638885376,
+      "loss": 3.2429,
+      "step": 38860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014749452813706405,
+      "loss": 3.3057,
+      "step": 38865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014742369135290409,
+      "loss": 3.2106,
+      "step": 38870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014735286604170074,
+      "loss": 3.1396,
+      "step": 38875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014728205220877953,
+      "loss": 3.3025,
+      "step": 38880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014721124985946521,
+      "loss": 3.1966,
+      "step": 38885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014714045899908215,
+      "loss": 3.229,
+      "step": 38890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001470696796329529,
+      "loss": 3.2342,
+      "step": 38895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001469989117664001,
+      "loss": 3.2281,
+      "step": 38900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014692815540474506,
+      "loss": 3.1531,
+      "step": 38905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014685741055330809,
+      "loss": 3.2777,
+      "step": 38910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014678667721740925,
+      "loss": 3.0968,
+      "step": 38915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014671595540236687,
+      "loss": 3.2377,
+      "step": 38920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014664524511349918,
+      "loss": 3.2756,
+      "step": 38925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014657454635612314,
+      "loss": 3.2936,
+      "step": 38930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014650385913555475,
+      "loss": 3.1701,
+      "step": 38935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014643318345710976,
+      "loss": 3.2626,
+      "step": 38940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014636251932610226,
+      "loss": 3.1047,
+      "step": 38945
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001462918667478458,
+      "loss": 3.2449,
+      "step": 38950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014622122572765335,
+      "loss": 3.1803,
+      "step": 38955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014615059627083658,
+      "loss": 3.199,
+      "step": 38960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014607997838270652,
+      "loss": 3.2114,
+      "step": 38965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014600937206857328,
+      "loss": 3.1264,
+      "step": 38970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014593877733374595,
+      "loss": 3.1626,
+      "step": 38975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001458681941835331,
+      "loss": 3.1494,
+      "step": 38980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014579762262324216,
+      "loss": 3.1752,
+      "step": 38985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014572706265817976,
+      "loss": 3.2887,
+      "step": 38990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014565651429365153,
+      "loss": 3.1689,
+      "step": 38995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001455859775349624,
+      "loss": 3.1576,
+      "step": 39000
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 3.232856512069702,
+      "eval_runtime": 149.8711,
+      "eval_samples_per_second": 12.284,
+      "eval_steps_per_second": 0.774,
+      "step": 39000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014551545238741643,
+      "loss": 3.1477,
+      "step": 39005
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014544493885631677,
+      "loss": 3.1183,
+      "step": 39010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001453744369469656,
+      "loss": 3.2491,
+      "step": 39015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014530394666466436,
+      "loss": 3.2484,
+      "step": 39020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014523346801471347,
+      "loss": 3.1706,
+      "step": 39025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014516300100241256,
+      "loss": 3.2258,
+      "step": 39030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014509254563306053,
+      "loss": 3.212,
+      "step": 39035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014502210191195522,
+      "loss": 3.1339,
+      "step": 39040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014495166984439356,
+      "loss": 3.0903,
+      "step": 39045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014488124943567174,
+      "loss": 3.2413,
+      "step": 39050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014481084069108488,
+      "loss": 3.0555,
+      "step": 39055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014474044361592755,
+      "loss": 3.1508,
+      "step": 39060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001446700582154932,
+      "loss": 3.2852,
+      "step": 39065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001445996844950744,
+      "loss": 3.1775,
+      "step": 39070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001445293224599629,
+      "loss": 3.1098,
+      "step": 39075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014445897211544948,
+      "loss": 3.2777,
+      "step": 39080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014438863346682437,
+      "loss": 3.2472,
+      "step": 39085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014431830651937658,
+      "loss": 3.1457,
+      "step": 39090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001442479912783941,
+      "loss": 3.1604,
+      "step": 39095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014417768774916456,
+      "loss": 3.2105,
+      "step": 39100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014410739593697432,
+      "loss": 3.2452,
+      "step": 39105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014403711584710886,
+      "loss": 3.1147,
+      "step": 39110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014396684748485308,
+      "loss": 3.138,
+      "step": 39115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014389659085549075,
+      "loss": 3.1713,
+      "step": 39120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014382634596430476,
+      "loss": 3.2346,
+      "step": 39125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014375611281657716,
+      "loss": 3.1103,
+      "step": 39130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014368589141758903,
+      "loss": 3.2913,
+      "step": 39135
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014361568177262088,
+      "loss": 3.0587,
+      "step": 39140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014354548388695201,
+      "loss": 3.183,
+      "step": 39145
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014347529776586094,
+      "loss": 3.2254,
+      "step": 39150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001434051234146253,
+      "loss": 3.2179,
+      "step": 39155
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014333496083852174,
+      "loss": 3.1301,
+      "step": 39160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014326481004282632,
+      "loss": 3.0906,
+      "step": 39165
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014319467103281406,
+      "loss": 3.0157,
+      "step": 39170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001431245438137587,
+      "loss": 3.2439,
+      "step": 39175
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014305442839093375,
+      "loss": 3.1518,
+      "step": 39180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014298432476961148,
+      "loss": 3.1854,
+      "step": 39185
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014291423295506318,
+      "loss": 3.2038,
+      "step": 39190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014284415295255973,
+      "loss": 3.226,
+      "step": 39195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014277408476737034,
+      "loss": 3.2791,
+      "step": 39200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014270402840476412,
+      "loss": 3.3441,
+      "step": 39205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014263398387000885,
+      "loss": 3.131,
+      "step": 39210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014256395116837137,
+      "loss": 3.2233,
+      "step": 39215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014249393030511814,
+      "loss": 3.1594,
+      "step": 39220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014242392128551405,
+      "loss": 3.2299,
+      "step": 39225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001423539241148234,
+      "loss": 3.2021,
+      "step": 39230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014228393879830984,
+      "loss": 3.2461,
+      "step": 39235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014221396534123582,
+      "loss": 3.1472,
+      "step": 39240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014214400374886293,
+      "loss": 3.1902,
+      "step": 39245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014207405402645197,
+      "loss": 3.2461,
+      "step": 39250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001420041161792627,
+      "loss": 3.2336,
+      "step": 39255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014193419021255428,
+      "loss": 3.2692,
+      "step": 39260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014186427613158468,
+      "loss": 3.1938,
+      "step": 39265
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014179437394161107,
+      "loss": 3.1897,
+      "step": 39270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014172448364788972,
+      "loss": 3.2023,
+      "step": 39275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014165460525567602,
+      "loss": 3.1319,
+      "step": 39280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014158473877022454,
+      "loss": 3.3315,
+      "step": 39285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014151488419678887,
+      "loss": 3.2045,
+      "step": 39290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014144504154062166,
+      "loss": 3.298,
+      "step": 39295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014137521080697478,
+      "loss": 3.2035,
+      "step": 39300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001413053920010991,
+      "loss": 3.2147,
+      "step": 39305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014123558512824452,
+      "loss": 3.2121,
+      "step": 39310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001411657901936604,
+      "loss": 3.1644,
+      "step": 39315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014109600720259484,
+      "loss": 3.2276,
+      "step": 39320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014102623616029516,
+      "loss": 3.2369,
+      "step": 39325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014095647707200782,
+      "loss": 3.0479,
+      "step": 39330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001408867299429782,
+      "loss": 3.174,
+      "step": 39335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014081699477845112,
+      "loss": 3.248,
+      "step": 39340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014074727158367028,
+      "loss": 3.1176,
+      "step": 39345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014067756036387842,
+      "loss": 3.218,
+      "step": 39350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014060786112431755,
+      "loss": 3.2828,
+      "step": 39355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001405381738702285,
+      "loss": 3.1822,
+      "step": 39360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001404684986068517,
+      "loss": 3.2422,
+      "step": 39365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014039883533942615,
+      "loss": 3.2066,
+      "step": 39370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001403291840731903,
+      "loss": 3.2333,
+      "step": 39375
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001402595448133815,
+      "loss": 3.3032,
+      "step": 39380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014018991756523628,
+      "loss": 3.2043,
+      "step": 39385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014012030233399013,
+      "loss": 3.1163,
+      "step": 39390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014005069912487793,
+      "loss": 3.1967,
+      "step": 39395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001399811079431335,
+      "loss": 3.3533,
+      "step": 39400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013991152879398963,
+      "loss": 3.1829,
+      "step": 39405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013984196168267837,
+      "loss": 3.245,
+      "step": 39410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001397724066144307,
+      "loss": 3.1686,
+      "step": 39415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013970286359447697,
+      "loss": 3.2045,
+      "step": 39420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013963333262804643,
+      "loss": 3.3767,
+      "step": 39425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001395638137203674,
+      "loss": 3.2496,
+      "step": 39430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013949430687666732,
+      "loss": 3.2372,
+      "step": 39435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001394248121021727,
+      "loss": 3.1356,
+      "step": 39440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013935532940210937,
+      "loss": 3.2368,
+      "step": 39445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013928585878170212,
+      "loss": 3.2078,
+      "step": 39450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001392164002461744,
+      "loss": 3.2284,
+      "step": 39455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013914695380074947,
+      "loss": 3.312,
+      "step": 39460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001390775194506493,
+      "loss": 3.0963,
+      "step": 39465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013900809720109478,
+      "loss": 3.1184,
+      "step": 39470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013893868705730657,
+      "loss": 3.0958,
+      "step": 39475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001388692890245034,
+      "loss": 3.2181,
+      "step": 39480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013879990310790405,
+      "loss": 3.2215,
+      "step": 39485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013873052931272583,
+      "loss": 3.2281,
+      "step": 39490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001386611676441852,
+      "loss": 3.2299,
+      "step": 39495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013859181810749817,
+      "loss": 3.2065,
+      "step": 39500
+    },
+    {
+      "epoch": 0.68,
+      "eval_loss": 3.226215362548828,
+      "eval_runtime": 149.8725,
+      "eval_samples_per_second": 12.284,
+      "eval_steps_per_second": 0.774,
+      "step": 39500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013852248070787897,
+      "loss": 3.1109,
+      "step": 39505
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013845315545054179,
+      "loss": 3.1988,
+      "step": 39510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013838384234069937,
+      "loss": 3.2595,
+      "step": 39515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013831454138356366,
+      "loss": 3.1322,
+      "step": 39520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013824525258434602,
+      "loss": 3.2045,
+      "step": 39525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001381759759482563,
+      "loss": 3.2601,
+      "step": 39530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013810671148050373,
+      "loss": 3.249,
+      "step": 39535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013803745918629688,
+      "loss": 3.3018,
+      "step": 39540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013796821907084307,
+      "loss": 3.1712,
+      "step": 39545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013789899113934878,
+      "loss": 3.1496,
+      "step": 39550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013782977539701958,
+      "loss": 3.1739,
+      "step": 39555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013776057184906008,
+      "loss": 3.3045,
+      "step": 39560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001376913805006742,
+      "loss": 3.2423,
+      "step": 39565
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013762220135706468,
+      "loss": 3.1416,
+      "step": 39570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013755303442343337,
+      "loss": 3.2441,
+      "step": 39575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013748387970498156,
+      "loss": 3.2493,
+      "step": 39580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001374147372069089,
+      "loss": 3.224,
+      "step": 39585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013734560693441484,
+      "loss": 3.28,
+      "step": 39590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013727648889269754,
+      "loss": 3.2218,
+      "step": 39595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013720738308695428,
+      "loss": 3.1533,
+      "step": 39600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013713828952238168,
+      "loss": 3.2559,
+      "step": 39605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013706920820417498,
+      "loss": 3.1483,
+      "step": 39610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001370001391375287,
+      "loss": 3.2323,
+      "step": 39615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013693108232763667,
+      "loss": 3.1266,
+      "step": 39620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013686203777969156,
+      "loss": 3.1826,
+      "step": 39625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013679300549888514,
+      "loss": 3.2438,
+      "step": 39630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013672398549040827,
+      "loss": 3.1957,
+      "step": 39635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013665497775945082,
+      "loss": 3.2472,
+      "step": 39640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013658598231120205,
+      "loss": 3.0956,
+      "step": 39645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013651699915084986,
+      "loss": 3.2534,
+      "step": 39650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013644802828358154,
+      "loss": 3.1855,
+      "step": 39655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001363790697145833,
+      "loss": 3.2826,
+      "step": 39660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013631012344904038,
+      "loss": 3.176,
+      "step": 39665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013624118949213737,
+      "loss": 3.1286,
+      "step": 39670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013617226784905765,
+      "loss": 3.2014,
+      "step": 39675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001361033585249838,
+      "loss": 3.1897,
+      "step": 39680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013603446152509743,
+      "loss": 3.3175,
+      "step": 39685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013596557685457923,
+      "loss": 3.1729,
+      "step": 39690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001358967045186089,
+      "loss": 3.216,
+      "step": 39695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013582784452236546,
+      "loss": 3.1958,
+      "step": 39700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013575899687102674,
+      "loss": 3.2406,
+      "step": 39705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013569016156976977,
+      "loss": 3.1854,
+      "step": 39710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013562133862377054,
+      "loss": 3.2161,
+      "step": 39715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001355525280382041,
+      "loss": 3.0194,
+      "step": 39720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013548372981824492,
+      "loss": 3.0951,
+      "step": 39725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001354149439690662,
+      "loss": 3.1332,
+      "step": 39730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013534617049584002,
+      "loss": 3.2012,
+      "step": 39735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013527740940373807,
+      "loss": 3.2941,
+      "step": 39740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013520866069793077,
+      "loss": 3.2922,
+      "step": 39745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001351399243835875,
+      "loss": 3.2285,
+      "step": 39750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013507120046587728,
+      "loss": 3.1846,
+      "step": 39755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013500248894996727,
+      "loss": 3.2841,
+      "step": 39760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013493378984102462,
+      "loss": 3.2459,
+      "step": 39765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013486510314421503,
+      "loss": 3.2431,
+      "step": 39770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013479642886470326,
+      "loss": 3.1897,
+      "step": 39775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001347277670076536,
+      "loss": 3.1831,
+      "step": 39780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013465911757822865,
+      "loss": 3.2116,
+      "step": 39785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013459048058159085,
+      "loss": 3.0972,
+      "step": 39790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013452185602290113,
+      "loss": 3.2108,
+      "step": 39795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001344532439073197,
+      "loss": 3.2728,
+      "step": 39800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013438464424000602,
+      "loss": 3.188,
+      "step": 39805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001343160570261184,
+      "loss": 3.1234,
+      "step": 39810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000134247482270814,
+      "loss": 3.1929,
+      "step": 39815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013417891997924956,
+      "loss": 3.1941,
+      "step": 39820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013411037015658052,
+      "loss": 3.2134,
+      "step": 39825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013404183280796137,
+      "loss": 3.1629,
+      "step": 39830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001339733079385461,
+      "loss": 3.315,
+      "step": 39835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013390479555348698,
+      "loss": 3.1403,
+      "step": 39840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013383629565793612,
+      "loss": 3.1475,
+      "step": 39845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013376780825704422,
+      "loss": 3.1861,
+      "step": 39850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013369933335596112,
+      "loss": 3.2026,
+      "step": 39855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013363087095983608,
+      "loss": 3.2458,
+      "step": 39860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013356242107381673,
+      "loss": 3.0653,
+      "step": 39865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001334939837030504,
+      "loss": 3.1658,
+      "step": 39870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001334255588526832,
+      "loss": 3.2229,
+      "step": 39875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013335714652786018,
+      "loss": 3.3034,
+      "step": 39880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001332887467337259,
+      "loss": 3.1056,
+      "step": 39885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013322035947542344,
+      "loss": 3.2603,
+      "step": 39890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013315198475809505,
+      "loss": 3.1717,
+      "step": 39895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013308362258688245,
+      "loss": 3.1708,
+      "step": 39900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000133015272966926,
+      "loss": 3.0321,
+      "step": 39905
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013294693590336522,
+      "loss": 3.1176,
+      "step": 39910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013287861140133875,
+      "loss": 3.1338,
+      "step": 39915
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001328102994659841,
+      "loss": 3.1293,
+      "step": 39920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001327420001024382,
+      "loss": 3.1591,
+      "step": 39925
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013267371331583672,
+      "loss": 3.2051,
+      "step": 39930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013260543911131447,
+      "loss": 3.2059,
+      "step": 39935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013253717749400531,
+      "loss": 3.1315,
+      "step": 39940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013246892846904205,
+      "loss": 3.2699,
+      "step": 39945
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001324006920415569,
+      "loss": 3.2865,
+      "step": 39950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013233246821668077,
+      "loss": 3.1292,
+      "step": 39955
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001322642569995438,
+      "loss": 3.237,
+      "step": 39960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000132196058395275,
+      "loss": 3.1434,
+      "step": 39965
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013212787240900265,
+      "loss": 3.2707,
+      "step": 39970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013205969904585385,
+      "loss": 3.2334,
+      "step": 39975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013199153831095511,
+      "loss": 3.1555,
+      "step": 39980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001319233902094317,
+      "loss": 3.1761,
+      "step": 39985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013185525474640792,
+      "loss": 3.3449,
+      "step": 39990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013178713192700726,
+      "loss": 3.1922,
+      "step": 39995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013171902175635212,
+      "loss": 3.2252,
+      "step": 40000
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 3.221252679824829,
+      "eval_runtime": 149.7766,
+      "eval_samples_per_second": 12.292,
+      "eval_steps_per_second": 0.774,
+      "step": 40000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001316509242395642,
+      "loss": 3.2361,
+      "step": 40005
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013158283938176403,
+      "loss": 3.1515,
+      "step": 40010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013151476718807118,
+      "loss": 3.2082,
+      "step": 40015
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013144670766360438,
+      "loss": 3.1433,
+      "step": 40020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013137866081348122,
+      "loss": 3.0056,
+      "step": 40025
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013131062664281876,
+      "loss": 3.1968,
+      "step": 40030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001312426051567326,
+      "loss": 3.1794,
+      "step": 40035
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013117459636033767,
+      "loss": 3.086,
+      "step": 40040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013110660025874787,
+      "loss": 3.2869,
+      "step": 40045
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013103861685707623,
+      "loss": 3.2714,
+      "step": 40050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001309706461604345,
+      "loss": 3.3275,
+      "step": 40055
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013090268817393409,
+      "loss": 3.1481,
+      "step": 40060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013083474290268487,
+      "loss": 3.1364,
+      "step": 40065
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013076681035179604,
+      "loss": 3.2324,
+      "step": 40070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001306988905263758,
+      "loss": 3.2283,
+      "step": 40075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013063098343153116,
+      "loss": 3.2573,
+      "step": 40080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001305630890723687,
+      "loss": 3.1392,
+      "step": 40085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013049520745399363,
+      "loss": 3.3101,
+      "step": 40090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013042733858151023,
+      "loss": 3.1184,
+      "step": 40095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013035948246002191,
+      "loss": 3.2271,
+      "step": 40100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013029163909463102,
+      "loss": 3.2393,
+      "step": 40105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013022380849043926,
+      "loss": 3.2122,
+      "step": 40110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013015599065254707,
+      "loss": 3.2095,
+      "step": 40115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013008818558605376,
+      "loss": 3.2102,
+      "step": 40120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00013002039329605824,
+      "loss": 3.0793,
+      "step": 40125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00012995261378765798,
+      "loss": 3.1341,
+      "step": 40130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001298848470659496,
+      "loss": 3.2486,
+      "step": 40135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001298170931360291,
+      "loss": 3.1172,
+      "step": 40140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00012974935200299077,
+      "loss": 3.0749,
+      "step": 40145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001296816236719288,
+      "loss": 3.1997,
+      "step": 40150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00012961390814793583,
+      "loss": 3.1917,
+      "step": 40155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012954620543610367,
+      "loss": 3.2264,
+      "step": 40160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012947851554152354,
+      "loss": 3.2491,
+      "step": 40165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012941083846928488,
+      "loss": 3.1931,
+      "step": 40170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012934317422447702,
+      "loss": 3.218,
+      "step": 40175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001292755228121879,
+      "loss": 3.2307,
+      "step": 40180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012920788423750434,
+      "loss": 3.2845,
+      "step": 40185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012914025850551284,
+      "loss": 3.1866,
+      "step": 40190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012907264562129814,
+      "loss": 3.1706,
+      "step": 40195
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012900504558994438,
+      "loss": 3.2382,
+      "step": 40200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012893745841653499,
+      "loss": 3.1685,
+      "step": 40205
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012886988410615207,
+      "loss": 3.1907,
+      "step": 40210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012880232266387678,
+      "loss": 3.2388,
+      "step": 40215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001287347740947895,
+      "loss": 3.2157,
+      "step": 40220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012866723840396934,
+      "loss": 3.1247,
+      "step": 40225
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001285997155964949,
+      "loss": 3.2234,
+      "step": 40230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012853220567744344,
+      "loss": 3.2356,
+      "step": 40235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012846470865189138,
+      "loss": 3.2794,
+      "step": 40240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001283972245249141,
+      "loss": 3.1414,
+      "step": 40245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001283297533015861,
+      "loss": 3.1587,
+      "step": 40250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012826229498698075,
+      "loss": 3.2113,
+      "step": 40255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012819484958617077,
+      "loss": 3.2561,
+      "step": 40260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001281274171042276,
+      "loss": 3.1379,
+      "step": 40265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012805999754622182,
+      "loss": 3.1382,
+      "step": 40270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001279925909172231,
+      "loss": 3.2634,
+      "step": 40275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012792519722229989,
+      "loss": 3.2751,
+      "step": 40280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012785781646652006,
+      "loss": 3.0939,
+      "step": 40285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012779044865495024,
+      "loss": 3.2182,
+      "step": 40290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001277230937926561,
+      "loss": 3.0571,
+      "step": 40295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012765575188470246,
+      "loss": 3.1138,
+      "step": 40300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012758842293615285,
+      "loss": 3.1653,
+      "step": 40305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012752110695207037,
+      "loss": 3.2168,
+      "step": 40310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001274538039375167,
+      "loss": 3.1544,
+      "step": 40315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012738651389755274,
+      "loss": 3.2413,
+      "step": 40320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012731923683723828,
+      "loss": 3.2262,
+      "step": 40325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012725197276163222,
+      "loss": 3.1795,
+      "step": 40330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012718472167579236,
+      "loss": 3.1318,
+      "step": 40335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012711748358477592,
+      "loss": 3.2327,
+      "step": 40340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012705025849363868,
+      "loss": 3.3024,
+      "step": 40345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012698304640743565,
+      "loss": 3.156,
+      "step": 40350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012691584733122083,
+      "loss": 3.0815,
+      "step": 40355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012684866127004715,
+      "loss": 3.1712,
+      "step": 40360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012678148822896685,
+      "loss": 3.1828,
+      "step": 40365
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001267143282130309,
+      "loss": 3.2036,
+      "step": 40370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001266471812272894,
+      "loss": 3.1898,
+      "step": 40375
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012658004727679145,
+      "loss": 3.206,
+      "step": 40380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001265129263665851,
+      "loss": 3.2307,
+      "step": 40385
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012644581850171764,
+      "loss": 3.1573,
+      "step": 40390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012637872368723534,
+      "loss": 3.1681,
+      "step": 40395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000126311641928183,
+      "loss": 3.2237,
+      "step": 40400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001262445732296051,
+      "loss": 3.1012,
+      "step": 40405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012617751759654488,
+      "loss": 3.1711,
+      "step": 40410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012611047503404435,
+      "loss": 3.2486,
+      "step": 40415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012604344554714518,
+      "loss": 3.1923,
+      "step": 40420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012597642914088716,
+      "loss": 3.2736,
+      "step": 40425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012590942582030992,
+      "loss": 3.2386,
+      "step": 40430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012584243559045168,
+      "loss": 3.1725,
+      "step": 40435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001257754584563496,
+      "loss": 3.1664,
+      "step": 40440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001257084944230404,
+      "loss": 3.1511,
+      "step": 40445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012564154349555893,
+      "loss": 3.0612,
+      "step": 40450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012557460567893999,
+      "loss": 3.3118,
+      "step": 40455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012550768097821673,
+      "loss": 3.0661,
+      "step": 40460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001254407693984215,
+      "loss": 3.2882,
+      "step": 40465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012537387094458603,
+      "loss": 3.1967,
+      "step": 40470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012530698562174044,
+      "loss": 3.1915,
+      "step": 40475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012524011343491412,
+      "loss": 3.1552,
+      "step": 40480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012517325438913572,
+      "loss": 3.0782,
+      "step": 40485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012510640848943265,
+      "loss": 3.1107,
+      "step": 40490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012503957574083133,
+      "loss": 3.1968,
+      "step": 40495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001249727561483573,
+      "loss": 3.1461,
+      "step": 40500
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 3.2169432640075684,
+      "eval_runtime": 149.8523,
+      "eval_samples_per_second": 12.285,
+      "eval_steps_per_second": 0.774,
+      "step": 40500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012490594971703483,
+      "loss": 2.9939,
+      "step": 40505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012483915645188776,
+      "loss": 3.0932,
+      "step": 40510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012477237635793843,
+      "loss": 3.1378,
+      "step": 40515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012470560944020825,
+      "loss": 3.1682,
+      "step": 40520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012463885570371808,
+      "loss": 3.0671,
+      "step": 40525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012457211515348705,
+      "loss": 3.0951,
+      "step": 40530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000124505387794534,
+      "loss": 3.2597,
+      "step": 40535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001244386736318764,
+      "loss": 3.1131,
+      "step": 40540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001243719726705307,
+      "loss": 3.0366,
+      "step": 40545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012430528491551277,
+      "loss": 3.148,
+      "step": 40550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012423861037183693,
+      "loss": 3.1749,
+      "step": 40555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012417194904451668,
+      "loss": 3.1621,
+      "step": 40560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012410530093856487,
+      "loss": 3.1866,
+      "step": 40565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000124038666058993,
+      "loss": 3.2424,
+      "step": 40570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012397204441081164,
+      "loss": 3.3355,
+      "step": 40575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012390543599903042,
+      "loss": 3.1592,
+      "step": 40580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012383884082865781,
+      "loss": 3.105,
+      "step": 40585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001237722589047017,
+      "loss": 3.2395,
+      "step": 40590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012370569023216856,
+      "loss": 3.1991,
+      "step": 40595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012363913481606403,
+      "loss": 3.1987,
+      "step": 40600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012357259266139272,
+      "loss": 3.0725,
+      "step": 40605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012350606377315815,
+      "loss": 3.2065,
+      "step": 40610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012343954815636314,
+      "loss": 3.2754,
+      "step": 40615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001233730458160093,
+      "loss": 3.1027,
+      "step": 40620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001233065567570972,
+      "loss": 3.244,
+      "step": 40625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012324008098462652,
+      "loss": 3.1625,
+      "step": 40630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012317361850359583,
+      "loss": 3.1442,
+      "step": 40635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001231071693190027,
+      "loss": 3.2685,
+      "step": 40640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012304073343584403,
+      "loss": 3.1912,
+      "step": 40645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012297431085911526,
+      "loss": 3.1647,
+      "step": 40650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012290790159381106,
+      "loss": 3.1712,
+      "step": 40655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012284150564492513,
+      "loss": 3.3168,
+      "step": 40660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001227751230174499,
+      "loss": 3.2013,
+      "step": 40665
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001227087537163773,
+      "loss": 3.2584,
+      "step": 40670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012264239774669776,
+      "loss": 3.2529,
+      "step": 40675
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012257605511340104,
+      "loss": 3.1225,
+      "step": 40680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012250972582147562,
+      "loss": 3.0349,
+      "step": 40685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012244340987590914,
+      "loss": 3.1949,
+      "step": 40690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012237710728168835,
+      "loss": 3.2146,
+      "step": 40695
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012231081804379892,
+      "loss": 3.2274,
+      "step": 40700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001222445421672251,
+      "loss": 3.1047,
+      "step": 40705
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012217827965695083,
+      "loss": 3.276,
+      "step": 40710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012211203051795864,
+      "loss": 3.1447,
+      "step": 40715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012204579475522996,
+      "loss": 3.1452,
+      "step": 40720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012197957237374574,
+      "loss": 3.187,
+      "step": 40725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012191336337848511,
+      "loss": 3.1845,
+      "step": 40730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012184716777442699,
+      "loss": 3.1049,
+      "step": 40735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012178098556654884,
+      "loss": 3.1475,
+      "step": 40740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012171481675982707,
+      "loss": 3.0952,
+      "step": 40745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012164866135923754,
+      "loss": 3.204,
+      "step": 40750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001215825193697546,
+      "loss": 3.3169,
+      "step": 40755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012151639079635188,
+      "loss": 3.324,
+      "step": 40760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012145027564400183,
+      "loss": 3.1329,
+      "step": 40765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012138417391767586,
+      "loss": 3.2409,
+      "step": 40770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012131808562234478,
+      "loss": 3.1679,
+      "step": 40775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012125201076297802,
+      "loss": 3.2133,
+      "step": 40780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012118594934454375,
+      "loss": 3.2173,
+      "step": 40785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001211199013720098,
+      "loss": 2.9781,
+      "step": 40790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001210538668503425,
+      "loss": 3.2979,
+      "step": 40795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012098784578450726,
+      "loss": 3.1413,
+      "step": 40800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001209218381794688,
+      "loss": 3.1493,
+      "step": 40805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012085584404019011,
+      "loss": 3.1366,
+      "step": 40810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012078986337163399,
+      "loss": 3.1048,
+      "step": 40815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012072389617876167,
+      "loss": 3.1932,
+      "step": 40820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012065794246653348,
+      "loss": 3.1397,
+      "step": 40825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012059200223990913,
+      "loss": 3.1325,
+      "step": 40830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012052607550384668,
+      "loss": 3.1377,
+      "step": 40835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012046016226330344,
+      "loss": 3.1563,
+      "step": 40840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012039426252323596,
+      "loss": 3.246,
+      "step": 40845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012032837628859946,
+      "loss": 3.0618,
+      "step": 40850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001202625035643483,
+      "loss": 3.1616,
+      "step": 40855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012019664435543569,
+      "loss": 3.1188,
+      "step": 40860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012013079866681387,
+      "loss": 3.2018,
+      "step": 40865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012006496650343427,
+      "loss": 3.1535,
+      "step": 40870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011999914787024704,
+      "loss": 3.1538,
+      "step": 40875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011993334277220142,
+      "loss": 3.2631,
+      "step": 40880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011986755121424558,
+      "loss": 3.1214,
+      "step": 40885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011980177320132663,
+      "loss": 3.1057,
+      "step": 40890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001197360087383909,
+      "loss": 3.1804,
+      "step": 40895
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001196702578303835,
+      "loss": 3.0932,
+      "step": 40900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011960452048224853,
+      "loss": 3.1851,
+      "step": 40905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011953879669892913,
+      "loss": 3.1617,
+      "step": 40910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011947308648536734,
+      "loss": 3.1573,
+      "step": 40915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011940738984650419,
+      "loss": 3.1449,
+      "step": 40920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011934170678727987,
+      "loss": 3.1135,
+      "step": 40925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011927603731263337,
+      "loss": 3.1522,
+      "step": 40930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011921038142750264,
+      "loss": 3.1436,
+      "step": 40935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001191447391368247,
+      "loss": 3.1961,
+      "step": 40940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011907911044553541,
+      "loss": 3.2341,
+      "step": 40945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011901349535856993,
+      "loss": 3.3183,
+      "step": 40950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011894789388086205,
+      "loss": 3.1467,
+      "step": 40955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011888230601734466,
+      "loss": 3.099,
+      "step": 40960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011881673177294967,
+      "loss": 3.1508,
+      "step": 40965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011875117115260775,
+      "loss": 3.2019,
+      "step": 40970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011868562416124903,
+      "loss": 3.2991,
+      "step": 40975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011862009080380213,
+      "loss": 3.0779,
+      "step": 40980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011855457108519482,
+      "loss": 3.1401,
+      "step": 40985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001184890650103539,
+      "loss": 3.1752,
+      "step": 40990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011842357258420507,
+      "loss": 3.241,
+      "step": 40995
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011835809381167289,
+      "loss": 3.2169,
+      "step": 41000
+    },
+    {
+      "epoch": 0.71,
+      "eval_loss": 3.2123522758483887,
+      "eval_runtime": 149.7773,
+      "eval_samples_per_second": 12.292,
+      "eval_steps_per_second": 0.774,
+      "step": 41000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011829262869768125,
+      "loss": 3.1618,
+      "step": 41005
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011822717724715272,
+      "loss": 3.1831,
+      "step": 41010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001181617394650089,
+      "loss": 3.1159,
+      "step": 41015
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011809631535617035,
+      "loss": 3.2193,
+      "step": 41020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011803090492555653,
+      "loss": 3.2243,
+      "step": 41025
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001179655081780862,
+      "loss": 3.1949,
+      "step": 41030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011790012511867675,
+      "loss": 3.2647,
+      "step": 41035
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011783475575224461,
+      "loss": 3.1304,
+      "step": 41040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001177694000837053,
+      "loss": 3.045,
+      "step": 41045
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011770405811797305,
+      "loss": 3.1695,
+      "step": 41050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011763872985996152,
+      "loss": 3.1831,
+      "step": 41055
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011757341531458297,
+      "loss": 3.0754,
+      "step": 41060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001175081144867485,
+      "loss": 3.0875,
+      "step": 41065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011744282738136864,
+      "loss": 3.1625,
+      "step": 41070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011737755400335259,
+      "loss": 3.2232,
+      "step": 41075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011731229435760843,
+      "loss": 3.2507,
+      "step": 41080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011724704844904368,
+      "loss": 3.1869,
+      "step": 41085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011718181628256413,
+      "loss": 3.2847,
+      "step": 41090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011711659786307514,
+      "loss": 3.199,
+      "step": 41095
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011705139319548073,
+      "loss": 3.1731,
+      "step": 41100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011698620228468384,
+      "loss": 3.0682,
+      "step": 41105
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011692102513558682,
+      "loss": 3.1529,
+      "step": 41110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011685586175309026,
+      "loss": 3.31,
+      "step": 41115
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011679071214209438,
+      "loss": 3.1318,
+      "step": 41120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011672557630749801,
+      "loss": 3.2206,
+      "step": 41125
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011666045425419895,
+      "loss": 3.1701,
+      "step": 41130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011659534598709431,
+      "loss": 3.1972,
+      "step": 41135
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011653025151107965,
+      "loss": 3.1705,
+      "step": 41140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001164651708310497,
+      "loss": 3.1075,
+      "step": 41145
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011640010395189838,
+      "loss": 3.2107,
+      "step": 41150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001163350508785183,
+      "loss": 3.1441,
+      "step": 41155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011627001161580116,
+      "loss": 3.2071,
+      "step": 41160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011620498616863753,
+      "loss": 3.0613,
+      "step": 41165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011613997454191691,
+      "loss": 3.2416,
+      "step": 41170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011607497674052802,
+      "loss": 3.1232,
+      "step": 41175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011600999276935832,
+      "loss": 3.1565,
+      "step": 41180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011594502263329418,
+      "loss": 3.1478,
+      "step": 41185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001158800663372211,
+      "loss": 3.0881,
+      "step": 41190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011581512388602334,
+      "loss": 3.174,
+      "step": 41195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001157501952845844,
+      "loss": 3.0763,
+      "step": 41200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011568528053778651,
+      "loss": 3.153,
+      "step": 41205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011562037965051084,
+      "loss": 3.2635,
+      "step": 41210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001155554926276379,
+      "loss": 3.2358,
+      "step": 41215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011549061947404656,
+      "loss": 3.167,
+      "step": 41220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011542576019461489,
+      "loss": 3.144,
+      "step": 41225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011536091479422028,
+      "loss": 3.1572,
+      "step": 41230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011529608327773859,
+      "loss": 3.128,
+      "step": 41235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011523126565004486,
+      "loss": 3.2118,
+      "step": 41240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011516646191601301,
+      "loss": 3.273,
+      "step": 41245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011510167208051588,
+      "loss": 3.1155,
+      "step": 41250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011503689614842551,
+      "loss": 3.2507,
+      "step": 41255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011497213412461262,
+      "loss": 3.231,
+      "step": 41260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000114907386013947,
+      "loss": 3.1433,
+      "step": 41265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011484265182129739,
+      "loss": 3.1993,
+      "step": 41270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011477793155153134,
+      "loss": 3.3162,
+      "step": 41275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011471322520951567,
+      "loss": 3.2125,
+      "step": 41280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001146485328001159,
+      "loss": 3.1264,
+      "step": 41285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011458385432819654,
+      "loss": 3.2733,
+      "step": 41290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011451918979862108,
+      "loss": 3.2649,
+      "step": 41295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011445453921625198,
+      "loss": 3.2322,
+      "step": 41300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011438990258595055,
+      "loss": 3.1406,
+      "step": 41305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001143252799125773,
+      "loss": 3.1362,
+      "step": 41310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011426067120099141,
+      "loss": 3.1726,
+      "step": 41315
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011419607645605117,
+      "loss": 3.1266,
+      "step": 41320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011413149568261374,
+      "loss": 3.1741,
+      "step": 41325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011406692888553518,
+      "loss": 3.1985,
+      "step": 41330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011400237606967078,
+      "loss": 3.0828,
+      "step": 41335
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011393783723987457,
+      "loss": 3.2476,
+      "step": 41340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011387331240099926,
+      "loss": 3.2025,
+      "step": 41345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011380880155789706,
+      "loss": 3.3122,
+      "step": 41350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011374430471541874,
+      "loss": 3.1677,
+      "step": 41355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011367982187841411,
+      "loss": 2.9866,
+      "step": 41360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011361535305173218,
+      "loss": 3.1992,
+      "step": 41365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011355089824022028,
+      "loss": 3.1555,
+      "step": 41370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011348645744872542,
+      "loss": 3.2704,
+      "step": 41375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011342203068209307,
+      "loss": 3.1413,
+      "step": 41380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011335761794516771,
+      "loss": 3.1366,
+      "step": 41385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011329321924279318,
+      "loss": 3.1669,
+      "step": 41390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011322883457981149,
+      "loss": 3.1667,
+      "step": 41395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011316446396106438,
+      "loss": 3.1504,
+      "step": 41400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011310010739139204,
+      "loss": 3.2718,
+      "step": 41405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011303576487563366,
+      "loss": 3.1071,
+      "step": 41410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011297143641862781,
+      "loss": 3.2227,
+      "step": 41415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011290712202521133,
+      "loss": 3.2243,
+      "step": 41420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011284282170022033,
+      "loss": 3.185,
+      "step": 41425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001127785354484901,
+      "loss": 3.1816,
+      "step": 41430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011271426327485448,
+      "loss": 3.204,
+      "step": 41435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011265000518414633,
+      "loss": 3.1829,
+      "step": 41440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011258576118119786,
+      "loss": 3.0784,
+      "step": 41445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011252153127083944,
+      "loss": 3.0765,
+      "step": 41450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011245731545790121,
+      "loss": 3.0026,
+      "step": 41455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011239311374721167,
+      "loss": 3.1984,
+      "step": 41460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001123289261435984,
+      "loss": 3.1799,
+      "step": 41465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011226475265188832,
+      "loss": 3.2619,
+      "step": 41470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011220059327690649,
+      "loss": 3.1734,
+      "step": 41475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011213644802347769,
+      "loss": 3.1921,
+      "step": 41480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011207231689642522,
+      "loss": 3.1058,
+      "step": 41485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011200819990057129,
+      "loss": 3.0331,
+      "step": 41490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011194409704073752,
+      "loss": 3.0883,
+      "step": 41495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011188000832174376,
+      "loss": 3.1265,
+      "step": 41500
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 3.207731008529663,
+      "eval_runtime": 149.6797,
+      "eval_samples_per_second": 12.3,
+      "eval_steps_per_second": 0.775,
+      "step": 41500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011181593374840916,
+      "loss": 3.0709,
+      "step": 41505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011175187332555206,
+      "loss": 3.2107,
+      "step": 41510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011168782705798929,
+      "loss": 3.1451,
+      "step": 41515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011162379495053683,
+      "loss": 3.118,
+      "step": 41520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011155977700800959,
+      "loss": 3.0218,
+      "step": 41525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011149577323522128,
+      "loss": 3.0796,
+      "step": 41530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011143178363698489,
+      "loss": 3.2099,
+      "step": 41535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011136780821811194,
+      "loss": 3.0831,
+      "step": 41540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011130384698341308,
+      "loss": 3.2227,
+      "step": 41545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011123989993769789,
+      "loss": 3.151,
+      "step": 41550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011117596708577473,
+      "loss": 3.1856,
+      "step": 41555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011111204843245124,
+      "loss": 3.086,
+      "step": 41560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011104814398253367,
+      "loss": 3.2135,
+      "step": 41565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011098425374082733,
+      "loss": 3.1735,
+      "step": 41570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001109203777121364,
+      "loss": 3.2179,
+      "step": 41575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011085651590126405,
+      "loss": 3.2251,
+      "step": 41580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011079266831301225,
+      "loss": 3.1628,
+      "step": 41585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011072883495218224,
+      "loss": 3.2514,
+      "step": 41590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011066501582357384,
+      "loss": 3.2295,
+      "step": 41595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011060121093198591,
+      "loss": 3.2016,
+      "step": 41600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011053742028221629,
+      "loss": 3.1608,
+      "step": 41605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011047364387906157,
+      "loss": 3.1431,
+      "step": 41610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001104098817273176,
+      "loss": 3.1741,
+      "step": 41615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011034613383177893,
+      "loss": 3.2561,
+      "step": 41620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000110282400197239,
+      "loss": 3.1301,
+      "step": 41625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011021868082849031,
+      "loss": 3.1035,
+      "step": 41630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011015497573032411,
+      "loss": 3.2929,
+      "step": 41635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011009128490753086,
+      "loss": 3.233,
+      "step": 41640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011002760836489986,
+      "loss": 3.0038,
+      "step": 41645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010996394610721889,
+      "loss": 3.2885,
+      "step": 41650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010990029813927538,
+      "loss": 3.1889,
+      "step": 41655
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010983666446585519,
+      "loss": 3.2143,
+      "step": 41660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010977304509174315,
+      "loss": 3.1514,
+      "step": 41665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010970944002172335,
+      "loss": 3.1891,
+      "step": 41670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010964584926057843,
+      "loss": 3.1459,
+      "step": 41675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010958227281309008,
+      "loss": 3.069,
+      "step": 41680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010951871068403893,
+      "loss": 3.3585,
+      "step": 41685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010945516287820446,
+      "loss": 3.2503,
+      "step": 41690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001093916294003653,
+      "loss": 3.1437,
+      "step": 41695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010932811025529875,
+      "loss": 3.0947,
+      "step": 41700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010926460544778115,
+      "loss": 3.2289,
+      "step": 41705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010920111498258771,
+      "loss": 3.1003,
+      "step": 41710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010913763886449249,
+      "loss": 3.1294,
+      "step": 41715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010907417709826878,
+      "loss": 3.126,
+      "step": 41720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010901072968868858,
+      "loss": 3.0671,
+      "step": 41725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010894729664052247,
+      "loss": 3.2908,
+      "step": 41730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010888387795854067,
+      "loss": 3.1179,
+      "step": 41735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010882047364751177,
+      "loss": 3.1891,
+      "step": 41740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010875708371220335,
+      "loss": 3.193,
+      "step": 41745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010869370815738237,
+      "loss": 3.2572,
+      "step": 41750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010863034698781387,
+      "loss": 3.1651,
+      "step": 41755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010856700020826264,
+      "loss": 3.1652,
+      "step": 41760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010850366782349193,
+      "loss": 3.1472,
+      "step": 41765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010844034983826389,
+      "loss": 3.2266,
+      "step": 41770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010837704625734002,
+      "loss": 3.2599,
+      "step": 41775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010831375708547998,
+      "loss": 3.2315,
+      "step": 41780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010825048232744318,
+      "loss": 3.2559,
+      "step": 41785
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010818722198798737,
+      "loss": 3.1243,
+      "step": 41790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010812397607186936,
+      "loss": 3.1431,
+      "step": 41795
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010806074458384519,
+      "loss": 3.2221,
+      "step": 41800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010799752752866928,
+      "loss": 3.2885,
+      "step": 41805
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010793432491109518,
+      "loss": 3.2099,
+      "step": 41810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010787113673587563,
+      "loss": 3.2519,
+      "step": 41815
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010780796300776196,
+      "loss": 3.1942,
+      "step": 41820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010774480373150449,
+      "loss": 3.0997,
+      "step": 41825
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001076816589118525,
+      "loss": 3.1281,
+      "step": 41830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010761852855355405,
+      "loss": 3.1984,
+      "step": 41835
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010755541266135639,
+      "loss": 3.1747,
+      "step": 41840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010749231124000543,
+      "loss": 3.116,
+      "step": 41845
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010742922429424612,
+      "loss": 3.093,
+      "step": 41850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010736615182882218,
+      "loss": 3.2103,
+      "step": 41855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010730309384847643,
+      "loss": 3.2,
+      "step": 41860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010724005035795035,
+      "loss": 3.0806,
+      "step": 41865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010717702136198471,
+      "loss": 3.2047,
+      "step": 41870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010711400686531885,
+      "loss": 3.1632,
+      "step": 41875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010705100687269118,
+      "loss": 3.0442,
+      "step": 41880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010698802138883891,
+      "loss": 3.2539,
+      "step": 41885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010692505041849818,
+      "loss": 3.1052,
+      "step": 41890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010686209396640425,
+      "loss": 3.2667,
+      "step": 41895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010679915203729107,
+      "loss": 3.2379,
+      "step": 41900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010673622463589149,
+      "loss": 3.2051,
+      "step": 41905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010667331176693735,
+      "loss": 3.1852,
+      "step": 41910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010661041343515931,
+      "loss": 3.1764,
+      "step": 41915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010654752964528715,
+      "loss": 3.191,
+      "step": 41920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010648466040204932,
+      "loss": 3.2177,
+      "step": 41925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010642180571017332,
+      "loss": 3.176,
+      "step": 41930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010635896557438544,
+      "loss": 3.1663,
+      "step": 41935
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010629613999941095,
+      "loss": 3.1828,
+      "step": 41940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010623332898997389,
+      "loss": 3.246,
+      "step": 41945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010617053255079758,
+      "loss": 3.1608,
+      "step": 41950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010610775068660385,
+      "loss": 3.1494,
+      "step": 41955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010604498340211361,
+      "loss": 3.173,
+      "step": 41960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001059822307020466,
+      "loss": 3.2291,
+      "step": 41965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001059194925911214,
+      "loss": 3.1392,
+      "step": 41970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010585676907405583,
+      "loss": 3.1625,
+      "step": 41975
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010579406015556625,
+      "loss": 3.271,
+      "step": 41980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010573136584036807,
+      "loss": 3.1903,
+      "step": 41985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010566868613317558,
+      "loss": 3.1708,
+      "step": 41990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001056060210387018,
+      "loss": 3.1984,
+      "step": 41995
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010554337056165913,
+      "loss": 3.1468,
+      "step": 42000
+    },
+    {
+      "epoch": 0.73,
+      "eval_loss": 3.204759359359741,
+      "eval_runtime": 149.9776,
+      "eval_samples_per_second": 12.275,
+      "eval_steps_per_second": 0.773,
+      "step": 42000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010548073470675853,
+      "loss": 3.0838,
+      "step": 42005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010541811347870956,
+      "loss": 3.1097,
+      "step": 42010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010535550688222135,
+      "loss": 3.0943,
+      "step": 42015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010529291492200146,
+      "loss": 3.212,
+      "step": 42020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010523033760275644,
+      "loss": 3.1342,
+      "step": 42025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010516777492919206,
+      "loss": 3.1234,
+      "step": 42030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010510522690601228,
+      "loss": 3.09,
+      "step": 42035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010504269353792073,
+      "loss": 3.1094,
+      "step": 42040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010498017482961946,
+      "loss": 3.1887,
+      "step": 42045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010491767078580953,
+      "loss": 3.1215,
+      "step": 42050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010485518141119113,
+      "loss": 3.1981,
+      "step": 42055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010479270671046284,
+      "loss": 3.1392,
+      "step": 42060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010473024668832265,
+      "loss": 3.1256,
+      "step": 42065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010466780134946715,
+      "loss": 3.1203,
+      "step": 42070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010460537069859185,
+      "loss": 3.0632,
+      "step": 42075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001045429547403915,
+      "loss": 3.1372,
+      "step": 42080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010448055347955913,
+      "loss": 3.2098,
+      "step": 42085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010441816692078702,
+      "loss": 3.1605,
+      "step": 42090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001043557950687665,
+      "loss": 3.1536,
+      "step": 42095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001042934379281876,
+      "loss": 3.2327,
+      "step": 42100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010423109550373913,
+      "loss": 3.122,
+      "step": 42105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010416876780010898,
+      "loss": 3.1947,
+      "step": 42110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001041064548219838,
+      "loss": 3.2346,
+      "step": 42115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010404415657404935,
+      "loss": 3.2128,
+      "step": 42120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010398187306099007,
+      "loss": 3.2819,
+      "step": 42125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010391960428748938,
+      "loss": 3.1558,
+      "step": 42130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010385735025822954,
+      "loss": 3.1389,
+      "step": 42135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010379511097789162,
+      "loss": 3.2378,
+      "step": 42140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010373288645115594,
+      "loss": 3.1614,
+      "step": 42145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010367067668270137,
+      "loss": 3.106,
+      "step": 42150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010360848167720561,
+      "loss": 3.1168,
+      "step": 42155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010354630143934578,
+      "loss": 3.2302,
+      "step": 42160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010348413597379714,
+      "loss": 3.1316,
+      "step": 42165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010342198528523423,
+      "loss": 3.2556,
+      "step": 42170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010335984937833069,
+      "loss": 3.1538,
+      "step": 42175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010329772825775871,
+      "loss": 3.0533,
+      "step": 42180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001032356219281895,
+      "loss": 3.0731,
+      "step": 42185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010317353039429312,
+      "loss": 3.1566,
+      "step": 42190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010311145366073845,
+      "loss": 3.0046,
+      "step": 42195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010304939173219347,
+      "loss": 3.1053,
+      "step": 42200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010298734461332493,
+      "loss": 3.1393,
+      "step": 42205
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010292531230879836,
+      "loss": 3.1764,
+      "step": 42210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010286329482327833,
+      "loss": 3.2309,
+      "step": 42215
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010280129216142815,
+      "loss": 3.2166,
+      "step": 42220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010273930432791024,
+      "loss": 3.0617,
+      "step": 42225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010267733132738571,
+      "loss": 3.2347,
+      "step": 42230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010261537316451457,
+      "loss": 3.1509,
+      "step": 42235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010255342984395583,
+      "loss": 3.1592,
+      "step": 42240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010249150137036725,
+      "loss": 3.2813,
+      "step": 42245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010242958774840544,
+      "loss": 3.0992,
+      "step": 42250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010236768898272622,
+      "loss": 3.1728,
+      "step": 42255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010230580507798392,
+      "loss": 3.1918,
+      "step": 42260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010224393603883195,
+      "loss": 3.1306,
+      "step": 42265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010218208186992248,
+      "loss": 3.1408,
+      "step": 42270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010212024257590652,
+      "loss": 3.1967,
+      "step": 42275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010205841816143437,
+      "loss": 3.2242,
+      "step": 42280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001019966086311547,
+      "loss": 3.2118,
+      "step": 42285
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010193481398971532,
+      "loss": 3.195,
+      "step": 42290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010187303424176285,
+      "loss": 3.2024,
+      "step": 42295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010181126939194269,
+      "loss": 3.2437,
+      "step": 42300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010174951944489953,
+      "loss": 3.1947,
+      "step": 42305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010168778440527656,
+      "loss": 3.2454,
+      "step": 42310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010162606427771568,
+      "loss": 3.2738,
+      "step": 42315
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001015643590668582,
+      "loss": 3.1613,
+      "step": 42320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010150266877734395,
+      "loss": 3.117,
+      "step": 42325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010144099341381165,
+      "loss": 3.143,
+      "step": 42330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010137933298089923,
+      "loss": 3.2231,
+      "step": 42335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010131768748324284,
+      "loss": 3.1593,
+      "step": 42340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010125605692547824,
+      "loss": 3.1333,
+      "step": 42345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010119444131223962,
+      "loss": 3.1118,
+      "step": 42350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010113284064816004,
+      "loss": 3.1835,
+      "step": 42355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010107125493787187,
+      "loss": 3.2863,
+      "step": 42360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010100968418600574,
+      "loss": 3.1489,
+      "step": 42365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010094812839719143,
+      "loss": 3.1761,
+      "step": 42370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010088658757605786,
+      "loss": 3.0487,
+      "step": 42375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010082506172723244,
+      "loss": 3.0552,
+      "step": 42380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010076355085534155,
+      "loss": 3.1523,
+      "step": 42385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010070205496501075,
+      "loss": 3.1273,
+      "step": 42390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010064057406086381,
+      "loss": 3.1326,
+      "step": 42395
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010057910814752414,
+      "loss": 3.0923,
+      "step": 42400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010051765722961349,
+      "loss": 3.09,
+      "step": 42405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010045622131175262,
+      "loss": 3.2578,
+      "step": 42410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010039480039856145,
+      "loss": 3.0941,
+      "step": 42415
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001003333944946581,
+      "loss": 3.2674,
+      "step": 42420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010027200360466032,
+      "loss": 3.1428,
+      "step": 42425
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010021062773318431,
+      "loss": 3.2387,
+      "step": 42430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010014926688484507,
+      "loss": 3.0526,
+      "step": 42435
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010008792106425693,
+      "loss": 3.151,
+      "step": 42440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010002659027603253,
+      "loss": 3.113,
+      "step": 42445
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.996527452478354e-05,
+      "loss": 3.2201,
+      "step": 42450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.990397381512088e-05,
+      "loss": 3.1862,
+      "step": 42455
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.984268815165389e-05,
+      "loss": 3.2709,
+      "step": 42460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.978141753899098e-05,
+      "loss": 3.2524,
+      "step": 42465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.972016198173938e-05,
+      "loss": 3.12,
+      "step": 42470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.965892148450506e-05,
+      "loss": 3.2258,
+      "step": 42475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.959769605189322e-05,
+      "loss": 3.163,
+      "step": 42480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.953648568850758e-05,
+      "loss": 3.2217,
+      "step": 42485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.947529039895088e-05,
+      "loss": 3.1551,
+      "step": 42490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.941411018782465e-05,
+      "loss": 3.0359,
+      "step": 42495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.935294505972927e-05,
+      "loss": 3.2666,
+      "step": 42500
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 3.201066255569458,
+      "eval_runtime": 149.9729,
+      "eval_samples_per_second": 12.276,
+      "eval_steps_per_second": 0.773,
+      "step": 42500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.929179501926418e-05,
+      "loss": 3.1589,
+      "step": 42505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.923066007102752e-05,
+      "loss": 3.2029,
+      "step": 42510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.91695402196163e-05,
+      "loss": 3.2643,
+      "step": 42515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.910843546962638e-05,
+      "loss": 3.1305,
+      "step": 42520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.904734582565261e-05,
+      "loss": 3.1963,
+      "step": 42525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.898627129228842e-05,
+      "loss": 3.2523,
+      "step": 42530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.89252118741266e-05,
+      "loss": 3.2724,
+      "step": 42535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.886416757575832e-05,
+      "loss": 3.1106,
+      "step": 42540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.880313840177383e-05,
+      "loss": 3.0961,
+      "step": 42545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.874212435676224e-05,
+      "loss": 3.1022,
+      "step": 42550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.868112544531136e-05,
+      "loss": 3.2171,
+      "step": 42555
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.862014167200818e-05,
+      "loss": 3.1618,
+      "step": 42560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.855917304143833e-05,
+      "loss": 3.1864,
+      "step": 42565
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.849821955818625e-05,
+      "loss": 3.1131,
+      "step": 42570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.843728122683545e-05,
+      "loss": 3.276,
+      "step": 42575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.837635805196792e-05,
+      "loss": 3.1902,
+      "step": 42580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.831545003816512e-05,
+      "loss": 3.0883,
+      "step": 42585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.825455719000695e-05,
+      "loss": 3.1129,
+      "step": 42590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.819367951207196e-05,
+      "loss": 3.077,
+      "step": 42595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.813281700893813e-05,
+      "loss": 3.1294,
+      "step": 42600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.807196968518187e-05,
+      "loss": 3.2189,
+      "step": 42605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.801113754537855e-05,
+      "loss": 3.2169,
+      "step": 42610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.795032059410263e-05,
+      "loss": 3.2404,
+      "step": 42615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.788951883592709e-05,
+      "loss": 3.1659,
+      "step": 42620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.782873227542393e-05,
+      "loss": 3.0816,
+      "step": 42625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.7767960917164e-05,
+      "loss": 3.1314,
+      "step": 42630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.770720476571688e-05,
+      "loss": 3.2118,
+      "step": 42635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.764646382565133e-05,
+      "loss": 3.1068,
+      "step": 42640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.758573810153467e-05,
+      "loss": 3.1281,
+      "step": 42645
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.752502759793312e-05,
+      "loss": 3.1041,
+      "step": 42650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.746433231941186e-05,
+      "loss": 3.3104,
+      "step": 42655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.740365227053469e-05,
+      "loss": 3.2378,
+      "step": 42660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.734298745586472e-05,
+      "loss": 3.2226,
+      "step": 42665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.728233787996356e-05,
+      "loss": 3.1264,
+      "step": 42670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.72217035473915e-05,
+      "loss": 3.165,
+      "step": 42675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.716108446270822e-05,
+      "loss": 3.1995,
+      "step": 42680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.710048063047184e-05,
+      "loss": 3.187,
+      "step": 42685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.703989205523936e-05,
+      "loss": 3.1576,
+      "step": 42690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.697931874156707e-05,
+      "loss": 3.2503,
+      "step": 42695
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.691876069400931e-05,
+      "loss": 3.1001,
+      "step": 42700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.685821791712011e-05,
+      "loss": 3.2139,
+      "step": 42705
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.679769041545181e-05,
+      "loss": 3.0835,
+      "step": 42710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.673717819355571e-05,
+      "loss": 3.2361,
+      "step": 42715
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.66766812559823e-05,
+      "loss": 3.1695,
+      "step": 42720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.661619960728026e-05,
+      "loss": 3.1172,
+      "step": 42725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.655573325199778e-05,
+      "loss": 3.2464,
+      "step": 42730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.649528219468151e-05,
+      "loss": 3.22,
+      "step": 42735
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.643484643987698e-05,
+      "loss": 3.144,
+      "step": 42740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.637442599212894e-05,
+      "loss": 3.16,
+      "step": 42745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.631402085598038e-05,
+      "loss": 3.1275,
+      "step": 42750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.625363103597352e-05,
+      "loss": 3.2621,
+      "step": 42755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.61932565366495e-05,
+      "loss": 3.1393,
+      "step": 42760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.613289736254807e-05,
+      "loss": 3.1324,
+      "step": 42765
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.607255351820802e-05,
+      "loss": 2.9853,
+      "step": 42770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.601222500816679e-05,
+      "loss": 3.1869,
+      "step": 42775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.595191183696073e-05,
+      "loss": 3.0711,
+      "step": 42780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.589161400912524e-05,
+      "loss": 3.069,
+      "step": 42785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.583133152919437e-05,
+      "loss": 3.1541,
+      "step": 42790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.577106440170101e-05,
+      "loss": 3.1853,
+      "step": 42795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.571081263117695e-05,
+      "loss": 3.1427,
+      "step": 42800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.565057622215274e-05,
+      "loss": 3.0839,
+      "step": 42805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.5590355179158e-05,
+      "loss": 3.1947,
+      "step": 42810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.553014950672097e-05,
+      "loss": 3.1567,
+      "step": 42815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.54699592093688e-05,
+      "loss": 3.2465,
+      "step": 42820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.540978429162751e-05,
+      "loss": 3.1557,
+      "step": 42825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.534962475802193e-05,
+      "loss": 3.0594,
+      "step": 42830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.528948061307565e-05,
+      "loss": 3.0756,
+      "step": 42835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.522935186131141e-05,
+      "loss": 3.2361,
+      "step": 42840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.516923850725044e-05,
+      "loss": 3.1698,
+      "step": 42845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.510914055541304e-05,
+      "loss": 3.2094,
+      "step": 42850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.504905801031819e-05,
+      "loss": 3.1183,
+      "step": 42855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.498899087648373e-05,
+      "loss": 3.207,
+      "step": 42860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.492893915842661e-05,
+      "loss": 3.1721,
+      "step": 42865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.486890286066229e-05,
+      "loss": 3.2029,
+      "step": 42870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.480888198770516e-05,
+      "loss": 3.1417,
+      "step": 42875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.474887654406857e-05,
+      "loss": 3.1568,
+      "step": 42880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.468888653426444e-05,
+      "loss": 3.1464,
+      "step": 42885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.462891196280393e-05,
+      "loss": 3.179,
+      "step": 42890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.456895283419674e-05,
+      "loss": 3.1786,
+      "step": 42895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.450900915295147e-05,
+      "loss": 3.0743,
+      "step": 42900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.444908092357558e-05,
+      "loss": 3.1999,
+      "step": 42905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.438916815057541e-05,
+      "loss": 3.1467,
+      "step": 42910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.432927083845592e-05,
+      "loss": 3.201,
+      "step": 42915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.426938899172132e-05,
+      "loss": 3.2788,
+      "step": 42920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.420952261487434e-05,
+      "loss": 3.2115,
+      "step": 42925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.41496717124166e-05,
+      "loss": 3.1825,
+      "step": 42930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.408983628884856e-05,
+      "loss": 3.1825,
+      "step": 42935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.403001634866948e-05,
+      "loss": 3.0352,
+      "step": 42940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.397021189637765e-05,
+      "loss": 3.0817,
+      "step": 42945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.391042293647012e-05,
+      "loss": 3.1235,
+      "step": 42950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.385064947344241e-05,
+      "loss": 3.1679,
+      "step": 42955
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.379089151178945e-05,
+      "loss": 3.1045,
+      "step": 42960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.373114905600464e-05,
+      "loss": 3.146,
+      "step": 42965
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.367142211058023e-05,
+      "loss": 3.2207,
+      "step": 42970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.361171068000762e-05,
+      "loss": 3.2523,
+      "step": 42975
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.355201476877647e-05,
+      "loss": 3.078,
+      "step": 42980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.349233438137589e-05,
+      "loss": 3.1585,
+      "step": 42985
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.343266952229341e-05,
+      "loss": 3.171,
+      "step": 42990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.337302019601545e-05,
+      "loss": 3.2222,
+      "step": 42995
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.331338640702764e-05,
+      "loss": 3.1117,
+      "step": 43000
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 3.1977977752685547,
+      "eval_runtime": 149.6717,
+      "eval_samples_per_second": 12.3,
+      "eval_steps_per_second": 0.775,
+      "step": 43000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.32537681598137e-05,
+      "loss": 3.2042,
+      "step": 43005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.319416545885693e-05,
+      "loss": 3.1313,
+      "step": 43010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.313457830863906e-05,
+      "loss": 3.169,
+      "step": 43015
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.307500671364064e-05,
+      "loss": 3.136,
+      "step": 43020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.301545067834142e-05,
+      "loss": 3.1929,
+      "step": 43025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.295591020721944e-05,
+      "loss": 3.1781,
+      "step": 43030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.289638530475186e-05,
+      "loss": 3.1836,
+      "step": 43035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.283687597541477e-05,
+      "loss": 3.1999,
+      "step": 43040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.277738222368294e-05,
+      "loss": 3.0913,
+      "step": 43045
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.271790405402996e-05,
+      "loss": 3.2435,
+      "step": 43050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.265844147092832e-05,
+      "loss": 3.295,
+      "step": 43055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.259899447884915e-05,
+      "loss": 3.1696,
+      "step": 43060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.253956308226277e-05,
+      "loss": 3.0823,
+      "step": 43065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.248014728563801e-05,
+      "loss": 3.1608,
+      "step": 43070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.242074709344258e-05,
+      "loss": 3.1251,
+      "step": 43075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.236136251014336e-05,
+      "loss": 3.1621,
+      "step": 43080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.230199354020533e-05,
+      "loss": 3.2057,
+      "step": 43085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.224264018809304e-05,
+      "loss": 3.1949,
+      "step": 43090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.218330245826948e-05,
+      "loss": 3.1971,
+      "step": 43095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.212398035519641e-05,
+      "loss": 3.0869,
+      "step": 43100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.206467388333491e-05,
+      "loss": 3.093,
+      "step": 43105
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.200538304714416e-05,
+      "loss": 3.1078,
+      "step": 43110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.194610785108253e-05,
+      "loss": 3.2347,
+      "step": 43115
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.188684829960746e-05,
+      "loss": 3.1788,
+      "step": 43120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.182760439717482e-05,
+      "loss": 3.1977,
+      "step": 43125
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.176837614823945e-05,
+      "loss": 3.1689,
+      "step": 43130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.170916355725505e-05,
+      "loss": 3.1488,
+      "step": 43135
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.164996662867398e-05,
+      "loss": 3.1792,
+      "step": 43140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.159078536694774e-05,
+      "loss": 3.1396,
+      "step": 43145
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.153161977652634e-05,
+      "loss": 3.1108,
+      "step": 43150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.147246986185874e-05,
+      "loss": 3.0777,
+      "step": 43155
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.141333562739275e-05,
+      "loss": 3.2488,
+      "step": 43160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.135421707757483e-05,
+      "loss": 3.2371,
+      "step": 43165
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.129511421685058e-05,
+      "loss": 3.1491,
+      "step": 43170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.123602704966415e-05,
+      "loss": 3.2448,
+      "step": 43175
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.117695558045858e-05,
+      "loss": 3.1201,
+      "step": 43180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.111789981367576e-05,
+      "loss": 3.1906,
+      "step": 43185
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.105885975375637e-05,
+      "loss": 3.0943,
+      "step": 43190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.099983540513981e-05,
+      "loss": 3.2098,
+      "step": 43195
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.094082677226462e-05,
+      "loss": 3.2461,
+      "step": 43200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.088183385956785e-05,
+      "loss": 3.2476,
+      "step": 43205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.082285667148545e-05,
+      "loss": 3.1644,
+      "step": 43210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.076389521245225e-05,
+      "loss": 3.2083,
+      "step": 43215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.070494948690165e-05,
+      "loss": 3.2944,
+      "step": 43220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.064601949926638e-05,
+      "loss": 3.3015,
+      "step": 43225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.058710525397748e-05,
+      "loss": 3.0546,
+      "step": 43230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.052820675546508e-05,
+      "loss": 3.1223,
+      "step": 43235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.046932400815798e-05,
+      "loss": 3.1651,
+      "step": 43240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.041045701648382e-05,
+      "loss": 2.9694,
+      "step": 43245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.035160578486924e-05,
+      "loss": 3.0904,
+      "step": 43250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.029277031773958e-05,
+      "loss": 3.2837,
+      "step": 43255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.023395061951864e-05,
+      "loss": 3.1519,
+      "step": 43260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.017514669462969e-05,
+      "loss": 3.2259,
+      "step": 43265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.011635854749439e-05,
+      "loss": 3.1521,
+      "step": 43270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.005758618253316e-05,
+      "loss": 3.1211,
+      "step": 43275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.999882960416572e-05,
+      "loss": 3.2336,
+      "step": 43280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.994008881680989e-05,
+      "loss": 3.1052,
+      "step": 43285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.988136382488292e-05,
+      "loss": 3.1502,
+      "step": 43290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.982265463280054e-05,
+      "loss": 3.1397,
+      "step": 43295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.976396124497731e-05,
+      "loss": 3.2139,
+      "step": 43300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.97052836658268e-05,
+      "loss": 3.1242,
+      "step": 43305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.964662189976128e-05,
+      "loss": 3.2127,
+      "step": 43310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.95879759511917e-05,
+      "loss": 3.1406,
+      "step": 43315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.9529345824528e-05,
+      "loss": 3.1651,
+      "step": 43320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.947073152417877e-05,
+      "loss": 3.1555,
+      "step": 43325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.941213305455166e-05,
+      "loss": 3.2542,
+      "step": 43330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.9353550420053e-05,
+      "loss": 3.2307,
+      "step": 43335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.929498362508762e-05,
+      "loss": 3.239,
+      "step": 43340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.92364326740597e-05,
+      "loss": 3.096,
+      "step": 43345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.91778975713719e-05,
+      "loss": 3.2674,
+      "step": 43350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.911937832142569e-05,
+      "loss": 3.0949,
+      "step": 43355
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.906087492862166e-05,
+      "loss": 3.157,
+      "step": 43360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.900238739735859e-05,
+      "loss": 3.1264,
+      "step": 43365
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.894391573203475e-05,
+      "loss": 3.2266,
+      "step": 43370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.888545993704678e-05,
+      "loss": 3.1276,
+      "step": 43375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.882702001679017e-05,
+      "loss": 3.1758,
+      "step": 43380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.876859597565965e-05,
+      "loss": 3.1014,
+      "step": 43385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.871018781804796e-05,
+      "loss": 3.261,
+      "step": 43390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.865179554834739e-05,
+      "loss": 3.098,
+      "step": 43395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.859341917094863e-05,
+      "loss": 3.3551,
+      "step": 43400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.853505869024127e-05,
+      "loss": 3.2433,
+      "step": 43405
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.847671411061389e-05,
+      "loss": 3.1784,
+      "step": 43410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.841838543645353e-05,
+      "loss": 3.1108,
+      "step": 43415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.83600726721461e-05,
+      "loss": 3.2022,
+      "step": 43420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.830177582207669e-05,
+      "loss": 3.0931,
+      "step": 43425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.824349489062876e-05,
+      "loss": 3.24,
+      "step": 43430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.818522988218483e-05,
+      "loss": 3.1067,
+      "step": 43435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.812698080112607e-05,
+      "loss": 3.0216,
+      "step": 43440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.806874765183241e-05,
+      "loss": 3.09,
+      "step": 43445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.80105304386829e-05,
+      "loss": 3.1525,
+      "step": 43450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.795232916605506e-05,
+      "loss": 3.2424,
+      "step": 43455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.789414383832535e-05,
+      "loss": 3.2136,
+      "step": 43460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.783597445986901e-05,
+      "loss": 3.1622,
+      "step": 43465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.777782103506007e-05,
+      "loss": 3.2465,
+      "step": 43470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.771968356827123e-05,
+      "loss": 3.1255,
+      "step": 43475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.766156206387442e-05,
+      "loss": 3.1675,
+      "step": 43480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.760345652623987e-05,
+      "loss": 3.1265,
+      "step": 43485
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.75453669597369e-05,
+      "loss": 3.0942,
+      "step": 43490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.748729336873353e-05,
+      "loss": 3.1477,
+      "step": 43495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.742923575759649e-05,
+      "loss": 3.1318,
+      "step": 43500
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 3.1929843425750732,
+      "eval_runtime": 150.078,
+      "eval_samples_per_second": 12.267,
+      "eval_steps_per_second": 0.773,
+      "step": 43500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.737119413069158e-05,
+      "loss": 3.221,
+      "step": 43505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.73131684923832e-05,
+      "loss": 3.0582,
+      "step": 43510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.725515884703453e-05,
+      "loss": 3.1917,
+      "step": 43515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.719716519900763e-05,
+      "loss": 3.1326,
+      "step": 43520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.713918755266321e-05,
+      "loss": 3.0365,
+      "step": 43525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.708122591236109e-05,
+      "loss": 3.181,
+      "step": 43530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.702328028245956e-05,
+      "loss": 3.2332,
+      "step": 43535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.69653506673159e-05,
+      "loss": 3.1744,
+      "step": 43540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.690743707128605e-05,
+      "loss": 3.0682,
+      "step": 43545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.68495394987249e-05,
+      "loss": 3.3009,
+      "step": 43550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.679165795398587e-05,
+      "loss": 3.126,
+      "step": 43555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.67337924414216e-05,
+      "loss": 3.2137,
+      "step": 43560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.667594296538315e-05,
+      "loss": 3.1615,
+      "step": 43565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.661810953022052e-05,
+      "loss": 3.1869,
+      "step": 43570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.656029214028249e-05,
+      "loss": 3.1582,
+      "step": 43575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.650249079991654e-05,
+      "loss": 3.0512,
+      "step": 43580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.644470551346921e-05,
+      "loss": 3.1223,
+      "step": 43585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.638693628528559e-05,
+      "loss": 3.2332,
+      "step": 43590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.63291831197096e-05,
+      "loss": 3.0745,
+      "step": 43595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.627144602108399e-05,
+      "loss": 3.1756,
+      "step": 43600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.621372499375021e-05,
+      "loss": 3.2514,
+      "step": 43605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.615602004204876e-05,
+      "loss": 3.1066,
+      "step": 43610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.609833117031879e-05,
+      "loss": 3.1709,
+      "step": 43615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.604065838289788e-05,
+      "loss": 3.0782,
+      "step": 43620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.598300168412301e-05,
+      "loss": 3.1607,
+      "step": 43625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.592536107832962e-05,
+      "loss": 3.1411,
+      "step": 43630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.586773656985185e-05,
+      "loss": 3.0961,
+      "step": 43635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.581012816302309e-05,
+      "loss": 3.202,
+      "step": 43640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.575253586217478e-05,
+      "loss": 3.2263,
+      "step": 43645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.569495967163786e-05,
+      "loss": 3.0951,
+      "step": 43650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.56373995957417e-05,
+      "loss": 3.1947,
+      "step": 43655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.557985563881439e-05,
+      "loss": 3.2083,
+      "step": 43660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.552232780518324e-05,
+      "loss": 3.169,
+      "step": 43665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.546481609917367e-05,
+      "loss": 3.1267,
+      "step": 43670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.540732052511058e-05,
+      "loss": 3.288,
+      "step": 43675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.534984108731717e-05,
+      "loss": 3.0646,
+      "step": 43680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.529237779011557e-05,
+      "loss": 3.2094,
+      "step": 43685
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.523493063782699e-05,
+      "loss": 3.1029,
+      "step": 43690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.517749963477087e-05,
+      "loss": 3.1713,
+      "step": 43695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.512008478526575e-05,
+      "loss": 3.1246,
+      "step": 43700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.50626860936291e-05,
+      "loss": 3.1438,
+      "step": 43705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.500530356417692e-05,
+      "loss": 3.2288,
+      "step": 43710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.494793720122407e-05,
+      "loss": 3.049,
+      "step": 43715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.489058700908424e-05,
+      "loss": 3.1557,
+      "step": 43720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.483325299206972e-05,
+      "loss": 3.1421,
+      "step": 43725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.4775935154492e-05,
+      "loss": 3.1228,
+      "step": 43730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.47186335006609e-05,
+      "loss": 3.2315,
+      "step": 43735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.466134803488532e-05,
+      "loss": 3.2098,
+      "step": 43740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.460407876147273e-05,
+      "loss": 3.0859,
+      "step": 43745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.454682568472946e-05,
+      "loss": 3.1922,
+      "step": 43750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.44895888089608e-05,
+      "loss": 3.1085,
+      "step": 43755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.44323681384706e-05,
+      "loss": 3.1564,
+      "step": 43760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.437516367756153e-05,
+      "loss": 3.2095,
+      "step": 43765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.431797543053512e-05,
+      "loss": 3.2012,
+      "step": 43770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.426080340169158e-05,
+      "loss": 3.1939,
+      "step": 43775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.42036475953299e-05,
+      "loss": 3.1871,
+      "step": 43780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.414650801574806e-05,
+      "loss": 3.1464,
+      "step": 43785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.408938466724257e-05,
+      "loss": 3.2299,
+      "step": 43790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.403227755410887e-05,
+      "loss": 3.1247,
+      "step": 43795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.397518668064109e-05,
+      "loss": 3.1662,
+      "step": 43800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.391811205113204e-05,
+      "loss": 3.1429,
+      "step": 43805
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.386105366987363e-05,
+      "loss": 3.1746,
+      "step": 43810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.380401154115631e-05,
+      "loss": 3.2565,
+      "step": 43815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.374698566926937e-05,
+      "loss": 3.0521,
+      "step": 43820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.368997605850078e-05,
+      "loss": 3.1862,
+      "step": 43825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.363298271313735e-05,
+      "loss": 3.1848,
+      "step": 43830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.357600563746485e-05,
+      "loss": 3.1177,
+      "step": 43835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.351904483576758e-05,
+      "loss": 3.1069,
+      "step": 43840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.346210031232866e-05,
+      "loss": 3.2112,
+      "step": 43845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.340517207143007e-05,
+      "loss": 3.0668,
+      "step": 43850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.334826011735252e-05,
+      "loss": 3.0612,
+      "step": 43855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.32913644543754e-05,
+      "loss": 3.2284,
+      "step": 43860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.323448508677713e-05,
+      "loss": 3.1964,
+      "step": 43865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.317762201883468e-05,
+      "loss": 3.1696,
+      "step": 43870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.312077525482387e-05,
+      "loss": 3.0291,
+      "step": 43875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.306394479901928e-05,
+      "loss": 3.0917,
+      "step": 43880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.300713065569415e-05,
+      "loss": 3.1189,
+      "step": 43885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.295033282912081e-05,
+      "loss": 3.2172,
+      "step": 43890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.289355132357011e-05,
+      "loss": 3.213,
+      "step": 43895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.283678614331165e-05,
+      "loss": 3.2169,
+      "step": 43900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.278003729261396e-05,
+      "loss": 3.0397,
+      "step": 43905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.272330477574414e-05,
+      "loss": 3.144,
+      "step": 43910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.266658859696835e-05,
+      "loss": 3.0974,
+      "step": 43915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.26098887605514e-05,
+      "loss": 3.171,
+      "step": 43920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.255320527075651e-05,
+      "loss": 3.1021,
+      "step": 43925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.249653813184628e-05,
+      "loss": 3.2077,
+      "step": 43930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.243988734808168e-05,
+      "loss": 3.1471,
+      "step": 43935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.23832529237225e-05,
+      "loss": 3.0669,
+      "step": 43940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.232663486302761e-05,
+      "loss": 3.1868,
+      "step": 43945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.227003317025403e-05,
+      "loss": 3.2153,
+      "step": 43950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.22134478496582e-05,
+      "loss": 3.091,
+      "step": 43955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.215687890549499e-05,
+      "loss": 3.1243,
+      "step": 43960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.210032634201792e-05,
+      "loss": 3.2038,
+      "step": 43965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.204379016347979e-05,
+      "loss": 3.2283,
+      "step": 43970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.198727037413156e-05,
+      "loss": 3.2619,
+      "step": 43975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.193076697822316e-05,
+      "loss": 3.1312,
+      "step": 43980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.187427998000364e-05,
+      "loss": 3.0785,
+      "step": 43985
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.181780938372037e-05,
+      "loss": 2.9717,
+      "step": 43990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.176135519361964e-05,
+      "loss": 3.0844,
+      "step": 43995
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.170491741394658e-05,
+      "loss": 3.2462,
+      "step": 44000
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 3.190524101257324,
+      "eval_runtime": 149.6752,
+      "eval_samples_per_second": 12.3,
+      "eval_steps_per_second": 0.775,
+      "step": 44000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.164849604894487e-05,
+      "loss": 3.2107,
+      "step": 44005
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.159209110285734e-05,
+      "loss": 3.1158,
+      "step": 44010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.153570257992521e-05,
+      "loss": 3.1124,
+      "step": 44015
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.147933048438856e-05,
+      "loss": 3.2516,
+      "step": 44020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.142297482048653e-05,
+      "loss": 3.1914,
+      "step": 44025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.136663559245643e-05,
+      "loss": 3.1275,
+      "step": 44030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.131031280453494e-05,
+      "loss": 3.1475,
+      "step": 44035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.125400646095717e-05,
+      "loss": 3.2376,
+      "step": 44040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.119771656595696e-05,
+      "loss": 3.1365,
+      "step": 44045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.114144312376731e-05,
+      "loss": 3.1429,
+      "step": 44050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.108518613861941e-05,
+      "loss": 3.1624,
+      "step": 44055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.102894561474349e-05,
+      "loss": 3.1822,
+      "step": 44060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.097272155636877e-05,
+      "loss": 3.0874,
+      "step": 44065
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.091651396772286e-05,
+      "loss": 3.232,
+      "step": 44070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.086032285303232e-05,
+      "loss": 3.2127,
+      "step": 44075
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.080414821652244e-05,
+      "loss": 3.2376,
+      "step": 44080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.074799006241716e-05,
+      "loss": 3.156,
+      "step": 44085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.069184839493947e-05,
+      "loss": 3.232,
+      "step": 44090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.063572321831084e-05,
+      "loss": 3.0889,
+      "step": 44095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.05796145367516e-05,
+      "loss": 3.2148,
+      "step": 44100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.052352235448084e-05,
+      "loss": 3.1608,
+      "step": 44105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.046744667571629e-05,
+      "loss": 3.0584,
+      "step": 44110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.041138750467478e-05,
+      "loss": 3.2204,
+      "step": 44115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.035534484557153e-05,
+      "loss": 3.0466,
+      "step": 44120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.029931870262073e-05,
+      "loss": 3.1303,
+      "step": 44125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.024330908003516e-05,
+      "loss": 3.1986,
+      "step": 44130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.018731598202655e-05,
+      "loss": 3.1084,
+      "step": 44135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.013133941280516e-05,
+      "loss": 3.2342,
+      "step": 44140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.007537937658032e-05,
+      "loss": 3.1171,
+      "step": 44145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.001943587755984e-05,
+      "loss": 3.1854,
+      "step": 44150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.996350891995045e-05,
+      "loss": 3.175,
+      "step": 44155
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.990759850795746e-05,
+      "loss": 3.2055,
+      "step": 44160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.985170464578506e-05,
+      "loss": 3.0755,
+      "step": 44165
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.97958273376363e-05,
+      "loss": 3.0841,
+      "step": 44170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.97399665877128e-05,
+      "loss": 3.1463,
+      "step": 44175
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.9684122400215e-05,
+      "loss": 3.1655,
+      "step": 44180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.962829477934207e-05,
+      "loss": 3.1207,
+      "step": 44185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.957248372929192e-05,
+      "loss": 3.1037,
+      "step": 44190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.95166892542614e-05,
+      "loss": 3.125,
+      "step": 44195
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.946091135844601e-05,
+      "loss": 3.1905,
+      "step": 44200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.940515004603963e-05,
+      "loss": 3.2434,
+      "step": 44205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.934940532123552e-05,
+      "loss": 3.1113,
+      "step": 44210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.929367718822534e-05,
+      "loss": 3.2207,
+      "step": 44215
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.923796565119941e-05,
+      "loss": 3.1306,
+      "step": 44220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.918227071434729e-05,
+      "loss": 3.2167,
+      "step": 44225
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.912659238185654e-05,
+      "loss": 3.0545,
+      "step": 44230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.907093065791418e-05,
+      "loss": 3.1117,
+      "step": 44235
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.90152855467056e-05,
+      "loss": 3.2648,
+      "step": 44240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.895965705241489e-05,
+      "loss": 3.1396,
+      "step": 44245
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.890404517922529e-05,
+      "loss": 3.1897,
+      "step": 44250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.884844993131836e-05,
+      "loss": 2.9301,
+      "step": 44255
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.879287131287463e-05,
+      "loss": 3.1565,
+      "step": 44260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.873730932807332e-05,
+      "loss": 3.163,
+      "step": 44265
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.868176398109228e-05,
+      "loss": 3.097,
+      "step": 44270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.862623527610847e-05,
+      "loss": 3.1801,
+      "step": 44275
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.857072321729732e-05,
+      "loss": 3.0713,
+      "step": 44280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.851522780883277e-05,
+      "loss": 2.9592,
+      "step": 44285
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.845974905488812e-05,
+      "loss": 3.1761,
+      "step": 44290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.840428695963497e-05,
+      "loss": 3.1655,
+      "step": 44295
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.834884152724367e-05,
+      "loss": 3.0895,
+      "step": 44300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.829341276188372e-05,
+      "loss": 3.0921,
+      "step": 44305
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.823800066772273e-05,
+      "loss": 3.2266,
+      "step": 44310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.81826052489277e-05,
+      "loss": 3.2352,
+      "step": 44315
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.812722650966392e-05,
+      "loss": 3.1349,
+      "step": 44320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.807186445409554e-05,
+      "loss": 3.2131,
+      "step": 44325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.801651908638578e-05,
+      "loss": 3.1752,
+      "step": 44330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.796119041069595e-05,
+      "loss": 3.2089,
+      "step": 44335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.790587843118677e-05,
+      "loss": 3.1466,
+      "step": 44340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.785058315201728e-05,
+      "loss": 3.1338,
+      "step": 44345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.77953045773454e-05,
+      "loss": 3.0996,
+      "step": 44350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.7740042711328e-05,
+      "loss": 3.1595,
+      "step": 44355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.768479755812024e-05,
+      "loss": 3.1891,
+      "step": 44360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.762956912187625e-05,
+      "loss": 3.1434,
+      "step": 44365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.757435740674913e-05,
+      "loss": 3.1736,
+      "step": 44370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.751916241689042e-05,
+      "loss": 3.1572,
+      "step": 44375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.746398415645048e-05,
+      "loss": 3.0956,
+      "step": 44380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.740882262957847e-05,
+      "loss": 3.0352,
+      "step": 44385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.735367784042214e-05,
+      "loss": 3.11,
+      "step": 44390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.729854979312828e-05,
+      "loss": 3.1855,
+      "step": 44395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.724343849184217e-05,
+      "loss": 3.2492,
+      "step": 44400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.718834394070786e-05,
+      "loss": 3.1713,
+      "step": 44405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.713326614386818e-05,
+      "loss": 3.157,
+      "step": 44410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.707820510546466e-05,
+      "loss": 3.1427,
+      "step": 44415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.702316082963773e-05,
+      "loss": 3.1726,
+      "step": 44420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.696813332052634e-05,
+      "loss": 3.2024,
+      "step": 44425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.691312258226834e-05,
+      "loss": 3.1871,
+      "step": 44430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.685812861900023e-05,
+      "loss": 3.1701,
+      "step": 44435
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.680315143485724e-05,
+      "loss": 3.1798,
+      "step": 44440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.67481910339733e-05,
+      "loss": 3.1527,
+      "step": 44445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.669324742048133e-05,
+      "loss": 3.2117,
+      "step": 44450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.663832059851276e-05,
+      "loss": 3.2039,
+      "step": 44455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.658341057219775e-05,
+      "loss": 3.1415,
+      "step": 44460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.652851734566529e-05,
+      "loss": 3.1103,
+      "step": 44465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.647364092304292e-05,
+      "loss": 3.1017,
+      "step": 44470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.641878130845732e-05,
+      "loss": 3.0508,
+      "step": 44475
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.636393850603352e-05,
+      "loss": 3.142,
+      "step": 44480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.630911251989544e-05,
+      "loss": 3.1147,
+      "step": 44485
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.62543033541657e-05,
+      "loss": 3.1511,
+      "step": 44490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.619951101296557e-05,
+      "loss": 3.2319,
+      "step": 44495
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.614473550041533e-05,
+      "loss": 3.0392,
+      "step": 44500
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 3.186605215072632,
+      "eval_runtime": 149.8734,
+      "eval_samples_per_second": 12.284,
+      "eval_steps_per_second": 0.774,
+      "step": 44500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.608997682063378e-05,
+      "loss": 3.1118,
+      "step": 44505
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.603523497773841e-05,
+      "loss": 3.1806,
+      "step": 44510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.598050997584561e-05,
+      "loss": 3.0252,
+      "step": 44515
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.592580181907033e-05,
+      "loss": 3.2698,
+      "step": 44520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.587111051152633e-05,
+      "loss": 3.1393,
+      "step": 44525
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.58164360573263e-05,
+      "loss": 3.1737,
+      "step": 44530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.576177846058128e-05,
+      "loss": 3.2284,
+      "step": 44535
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.570713772540138e-05,
+      "loss": 3.2084,
+      "step": 44540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.56525138558952e-05,
+      "loss": 3.0973,
+      "step": 44545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.559790685617017e-05,
+      "loss": 3.1736,
+      "step": 44550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.554331673033255e-05,
+      "loss": 3.1482,
+      "step": 44555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.54887434824873e-05,
+      "loss": 3.2329,
+      "step": 44560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.543418711673774e-05,
+      "loss": 3.0866,
+      "step": 44565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.537964763718653e-05,
+      "loss": 3.21,
+      "step": 44570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.532512504793467e-05,
+      "loss": 3.103,
+      "step": 44575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.527061935308184e-05,
+      "loss": 3.1107,
+      "step": 44580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.521613055672693e-05,
+      "loss": 3.1379,
+      "step": 44585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.516165866296681e-05,
+      "loss": 3.1808,
+      "step": 44590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.510720367589777e-05,
+      "loss": 3.1765,
+      "step": 44595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.505276559961444e-05,
+      "loss": 3.2069,
+      "step": 44600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.499834443821023e-05,
+      "loss": 3.1409,
+      "step": 44605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.494394019577759e-05,
+      "loss": 3.1709,
+      "step": 44610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.488955287640706e-05,
+      "loss": 3.1446,
+      "step": 44615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.483518248418858e-05,
+      "loss": 3.0126,
+      "step": 44620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.47808290232104e-05,
+      "loss": 3.1909,
+      "step": 44625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.47264924975596e-05,
+      "loss": 3.1479,
+      "step": 44630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.46721729113222e-05,
+      "loss": 3.084,
+      "step": 44635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.461787026858255e-05,
+      "loss": 3.1349,
+      "step": 44640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.456358457342386e-05,
+      "loss": 3.1268,
+      "step": 44645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.450931582992839e-05,
+      "loss": 3.1649,
+      "step": 44650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.445506404217674e-05,
+      "loss": 3.1787,
+      "step": 44655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.440082921424836e-05,
+      "loss": 3.0321,
+      "step": 44660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.434661135022149e-05,
+      "loss": 3.1882,
+      "step": 44665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.429241045417287e-05,
+      "loss": 3.1999,
+      "step": 44670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.423822653017831e-05,
+      "loss": 3.1162,
+      "step": 44675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.418405958231217e-05,
+      "loss": 3.1264,
+      "step": 44680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.412990961464742e-05,
+      "loss": 3.2139,
+      "step": 44685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.40757766312559e-05,
+      "loss": 3.0781,
+      "step": 44690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.402166063620805e-05,
+      "loss": 3.1487,
+      "step": 44695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.39675616335733e-05,
+      "loss": 3.0855,
+      "step": 44700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.391347962741948e-05,
+      "loss": 3.1442,
+      "step": 44705
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.385941462181327e-05,
+      "loss": 3.2047,
+      "step": 44710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.38053666208203e-05,
+      "loss": 3.184,
+      "step": 44715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.375133562850443e-05,
+      "loss": 3.2336,
+      "step": 44720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.369732164892854e-05,
+      "loss": 3.1874,
+      "step": 44725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.364332468615436e-05,
+      "loss": 3.2044,
+      "step": 44730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.35893447442421e-05,
+      "loss": 3.209,
+      "step": 44735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.353538182725077e-05,
+      "loss": 3.1973,
+      "step": 44740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.348143593923814e-05,
+      "loss": 3.2708,
+      "step": 44745
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.342750708426053e-05,
+      "loss": 3.1732,
+      "step": 44750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.337359526637332e-05,
+      "loss": 3.1316,
+      "step": 44755
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.331970048963027e-05,
+      "loss": 3.0108,
+      "step": 44760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.326582275808409e-05,
+      "loss": 3.1858,
+      "step": 44765
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.3211962075786e-05,
+      "loss": 3.1266,
+      "step": 44770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.3158118446786e-05,
+      "loss": 3.2553,
+      "step": 44775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.310429187513305e-05,
+      "loss": 3.0143,
+      "step": 44780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.305048236487452e-05,
+      "loss": 3.2031,
+      "step": 44785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.299668992005664e-05,
+      "loss": 3.1179,
+      "step": 44790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.294291454472429e-05,
+      "loss": 3.0976,
+      "step": 44795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.28891562429211e-05,
+      "loss": 3.1156,
+      "step": 44800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.283541501868935e-05,
+      "loss": 3.2027,
+      "step": 44805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.27816908760703e-05,
+      "loss": 3.1327,
+      "step": 44810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.272798381910361e-05,
+      "loss": 3.0802,
+      "step": 44815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.267429385182777e-05,
+      "loss": 3.1307,
+      "step": 44820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.262062097828005e-05,
+      "loss": 3.065,
+      "step": 44825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.256696520249623e-05,
+      "loss": 3.0924,
+      "step": 44830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.251332652851111e-05,
+      "loss": 3.1293,
+      "step": 44835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.245970496035803e-05,
+      "loss": 3.1039,
+      "step": 44840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.240610050206896e-05,
+      "loss": 3.1689,
+      "step": 44845
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.235251315767475e-05,
+      "loss": 3.2031,
+      "step": 44850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.22989429312048e-05,
+      "loss": 3.0999,
+      "step": 44855
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.224538982668746e-05,
+      "loss": 3.1289,
+      "step": 44860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.21918538481497e-05,
+      "loss": 3.145,
+      "step": 44865
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.213833499961684e-05,
+      "loss": 3.2122,
+      "step": 44870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.208483328511349e-05,
+      "loss": 3.1842,
+      "step": 44875
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.203134870866264e-05,
+      "loss": 3.1976,
+      "step": 44880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.197788127428595e-05,
+      "loss": 3.2405,
+      "step": 44885
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.192443098600421e-05,
+      "loss": 3.0957,
+      "step": 44890
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.187099784783617e-05,
+      "loss": 3.1705,
+      "step": 44895
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.181758186380007e-05,
+      "loss": 2.9637,
+      "step": 44900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.176418303791237e-05,
+      "loss": 3.2228,
+      "step": 44905
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.171080137418832e-05,
+      "loss": 3.1422,
+      "step": 44910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.165743687664224e-05,
+      "loss": 3.225,
+      "step": 44915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.160408954928651e-05,
+      "loss": 3.1538,
+      "step": 44920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.155075939613279e-05,
+      "loss": 3.0714,
+      "step": 44925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.14974464211912e-05,
+      "loss": 3.1305,
+      "step": 44930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.144415062847047e-05,
+      "loss": 3.1041,
+      "step": 44935
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.139087202197842e-05,
+      "loss": 3.122,
+      "step": 44940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.133761060572123e-05,
+      "loss": 3.1977,
+      "step": 44945
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.128436638370367e-05,
+      "loss": 3.0151,
+      "step": 44950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.123113935992975e-05,
+      "loss": 3.1652,
+      "step": 44955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.117792953840168e-05,
+      "loss": 3.2035,
+      "step": 44960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.112473692312054e-05,
+      "loss": 3.1325,
+      "step": 44965
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.10715615180864e-05,
+      "loss": 3.0583,
+      "step": 44970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.101840332729739e-05,
+      "loss": 3.0872,
+      "step": 44975
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.096526235475102e-05,
+      "loss": 3.2089,
+      "step": 44980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.091213860444318e-05,
+      "loss": 2.9796,
+      "step": 44985
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.085903208036837e-05,
+      "loss": 3.0638,
+      "step": 44990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.080594278652017e-05,
+      "loss": 3.1517,
+      "step": 44995
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.07528707268903e-05,
+      "loss": 3.1721,
+      "step": 45000
+    },
+    {
+      "epoch": 0.78,
+      "eval_loss": 3.183523654937744,
+      "eval_runtime": 150.1695,
+      "eval_samples_per_second": 12.259,
+      "eval_steps_per_second": 0.772,
+      "step": 45000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.069981590546978e-05,
+      "loss": 3.0841,
+      "step": 45005
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.064677832624794e-05,
+      "loss": 3.2613,
+      "step": 45010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.059375799321287e-05,
+      "loss": 3.0345,
+      "step": 45015
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.054075491035168e-05,
+      "loss": 3.196,
+      "step": 45020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.04877690816497e-05,
+      "loss": 3.1549,
+      "step": 45025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.043480051109114e-05,
+      "loss": 3.0979,
+      "step": 45030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.038184920265916e-05,
+      "loss": 3.1891,
+      "step": 45035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.032891516033538e-05,
+      "loss": 3.119,
+      "step": 45040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.027599838810011e-05,
+      "loss": 3.169,
+      "step": 45045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.022309888993245e-05,
+      "loss": 3.1685,
+      "step": 45050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.017021666981009e-05,
+      "loss": 3.2098,
+      "step": 45055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.011735173170964e-05,
+      "loss": 3.1718,
+      "step": 45060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.006450407960622e-05,
+      "loss": 3.1111,
+      "step": 45065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.001167371747369e-05,
+      "loss": 3.1789,
+      "step": 45070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.995886064928464e-05,
+      "loss": 3.0517,
+      "step": 45075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.990606487901033e-05,
+      "loss": 3.1516,
+      "step": 45080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.985328641062064e-05,
+      "loss": 3.1516,
+      "step": 45085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.98005252480844e-05,
+      "loss": 3.2177,
+      "step": 45090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.97477813953689e-05,
+      "loss": 3.0959,
+      "step": 45095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.969505485644023e-05,
+      "loss": 3.1638,
+      "step": 45100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.964234563526314e-05,
+      "loss": 3.2965,
+      "step": 45105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.958965373580101e-05,
+      "loss": 3.2153,
+      "step": 45110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.953697916201618e-05,
+      "loss": 3.1298,
+      "step": 45115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.948432191786936e-05,
+      "loss": 3.0617,
+      "step": 45120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.94316820073202e-05,
+      "loss": 3.1779,
+      "step": 45125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.93790594343269e-05,
+      "loss": 3.1907,
+      "step": 45130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.932645420284636e-05,
+      "loss": 3.1058,
+      "step": 45135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.927386631683435e-05,
+      "loss": 3.0789,
+      "step": 45140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.92212957802452e-05,
+      "loss": 3.0602,
+      "step": 45145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.916874259703172e-05,
+      "loss": 3.197,
+      "step": 45150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.911620677114592e-05,
+      "loss": 3.1129,
+      "step": 45155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.90636883065381e-05,
+      "loss": 3.2127,
+      "step": 45160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.901118720715731e-05,
+      "loss": 3.1139,
+      "step": 45165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.895870347695151e-05,
+      "loss": 3.2134,
+      "step": 45170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.890623711986715e-05,
+      "loss": 3.1768,
+      "step": 45175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.88537881398494e-05,
+      "loss": 3.1486,
+      "step": 45180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.88013565408422e-05,
+      "loss": 3.2026,
+      "step": 45185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.874894232678804e-05,
+      "loss": 3.1052,
+      "step": 45190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.869654550162833e-05,
+      "loss": 3.1453,
+      "step": 45195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.864416606930299e-05,
+      "loss": 3.1664,
+      "step": 45200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.859180403375072e-05,
+      "loss": 3.1615,
+      "step": 45205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.853945939890878e-05,
+      "loss": 3.0932,
+      "step": 45210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.848713216871323e-05,
+      "loss": 3.1201,
+      "step": 45215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.843482234709893e-05,
+      "loss": 3.0702,
+      "step": 45220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.838252993799932e-05,
+      "loss": 3.0548,
+      "step": 45225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.833025494534629e-05,
+      "loss": 3.1788,
+      "step": 45230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.827799737307088e-05,
+      "loss": 3.2091,
+      "step": 45235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.822575722510249e-05,
+      "loss": 3.29,
+      "step": 45240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.817353450536926e-05,
+      "loss": 3.0955,
+      "step": 45245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.812132921779832e-05,
+      "loss": 3.219,
+      "step": 45250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.806914136631487e-05,
+      "loss": 3.2833,
+      "step": 45255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.801697095484346e-05,
+      "loss": 3.1134,
+      "step": 45260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.796481798730692e-05,
+      "loss": 3.1767,
+      "step": 45265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.791268246762687e-05,
+      "loss": 3.1696,
+      "step": 45270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.78605643997238e-05,
+      "loss": 3.1449,
+      "step": 45275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.780846378751643e-05,
+      "loss": 3.1968,
+      "step": 45280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.775638063492272e-05,
+      "loss": 3.0654,
+      "step": 45285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.770431494585892e-05,
+      "loss": 3.1028,
+      "step": 45290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.765226672424007e-05,
+      "loss": 3.1161,
+      "step": 45295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.76002359739802e-05,
+      "loss": 3.2315,
+      "step": 45300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.754822269899145e-05,
+      "loss": 3.0886,
+      "step": 45305
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.749622690318497e-05,
+      "loss": 3.1833,
+      "step": 45310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.744424859047076e-05,
+      "loss": 3.2618,
+      "step": 45315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.739228776475721e-05,
+      "loss": 3.1447,
+      "step": 45320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.734034442995154e-05,
+      "loss": 3.0796,
+      "step": 45325
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.72884185899596e-05,
+      "loss": 3.0618,
+      "step": 45330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.723651024868586e-05,
+      "loss": 3.1823,
+      "step": 45335
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.718461941003375e-05,
+      "loss": 3.0862,
+      "step": 45340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.713274607790507e-05,
+      "loss": 3.2958,
+      "step": 45345
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.708089025620049e-05,
+      "loss": 3.1402,
+      "step": 45350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.702905194881928e-05,
+      "loss": 3.3028,
+      "step": 45355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.69772311596593e-05,
+      "loss": 3.1589,
+      "step": 45360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.692542789261742e-05,
+      "loss": 3.0914,
+      "step": 45365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.687364215158884e-05,
+      "loss": 3.1536,
+      "step": 45370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.682187394046763e-05,
+      "loss": 3.1351,
+      "step": 45375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.677012326314649e-05,
+      "loss": 3.1095,
+      "step": 45380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.671839012351682e-05,
+      "loss": 3.1706,
+      "step": 45385
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.666667452546854e-05,
+      "loss": 3.1585,
+      "step": 45390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.661497647289062e-05,
+      "loss": 3.1016,
+      "step": 45395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.656329596967038e-05,
+      "loss": 3.1136,
+      "step": 45400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.651163301969397e-05,
+      "loss": 3.1383,
+      "step": 45405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.645998762684613e-05,
+      "loss": 3.1012,
+      "step": 45410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.640835979501025e-05,
+      "loss": 3.1399,
+      "step": 45415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.635674952806867e-05,
+      "loss": 3.158,
+      "step": 45420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.630515682990212e-05,
+      "loss": 3.2877,
+      "step": 45425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.62535817043901e-05,
+      "loss": 3.0882,
+      "step": 45430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.620202415541081e-05,
+      "loss": 3.2474,
+      "step": 45435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.615048418684106e-05,
+      "loss": 3.1571,
+      "step": 45440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.609896180255649e-05,
+      "loss": 3.1484,
+      "step": 45445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.604745700643126e-05,
+      "loss": 3.1594,
+      "step": 45450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.599596980233828e-05,
+      "loss": 3.2426,
+      "step": 45455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.594450019414913e-05,
+      "loss": 3.169,
+      "step": 45460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.589304818573405e-05,
+      "loss": 3.1357,
+      "step": 45465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.584161378096187e-05,
+      "loss": 3.0433,
+      "step": 45470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.57901969837004e-05,
+      "loss": 3.1571,
+      "step": 45475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.573879779781578e-05,
+      "loss": 3.1311,
+      "step": 45480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.568741622717303e-05,
+      "loss": 3.191,
+      "step": 45485
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.563605227563572e-05,
+      "loss": 3.1198,
+      "step": 45490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.558470594706609e-05,
+      "loss": 3.1676,
+      "step": 45495
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.553337724532533e-05,
+      "loss": 3.0357,
+      "step": 45500
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 3.180738925933838,
+      "eval_runtime": 151.371,
+      "eval_samples_per_second": 12.162,
+      "eval_steps_per_second": 0.766,
+      "step": 45500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.548206617427295e-05,
+      "loss": 3.1704,
+      "step": 45505
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.543077273776732e-05,
+      "loss": 3.1313,
+      "step": 45510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.537949693966543e-05,
+      "loss": 3.0594,
+      "step": 45515
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.53282387838229e-05,
+      "loss": 3.0647,
+      "step": 45520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.52769982740942e-05,
+      "loss": 3.1225,
+      "step": 45525
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.522577541433235e-05,
+      "loss": 3.1439,
+      "step": 45530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.517457020838887e-05,
+      "loss": 3.1825,
+      "step": 45535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.51233826601143e-05,
+      "loss": 3.2263,
+      "step": 45540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.507221277335767e-05,
+      "loss": 3.167,
+      "step": 45545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.502106055196654e-05,
+      "loss": 3.0891,
+      "step": 45550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.496992599978757e-05,
+      "loss": 3.0816,
+      "step": 45555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.491880912066549e-05,
+      "loss": 3.1269,
+      "step": 45560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.486770991844429e-05,
+      "loss": 3.1808,
+      "step": 45565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.481662839696626e-05,
+      "loss": 3.2526,
+      "step": 45570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.476556456007242e-05,
+      "loss": 3.0343,
+      "step": 45575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.47145184116027e-05,
+      "loss": 3.268,
+      "step": 45580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.466348995539531e-05,
+      "loss": 3.0679,
+      "step": 45585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.46124791952873e-05,
+      "loss": 3.1898,
+      "step": 45590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.456148613511462e-05,
+      "loss": 3.0314,
+      "step": 45595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.451051077871154e-05,
+      "loss": 3.1434,
+      "step": 45600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.445955312991122e-05,
+      "loss": 3.2544,
+      "step": 45605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.440861319254536e-05,
+      "loss": 3.0762,
+      "step": 45610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.435769097044431e-05,
+      "loss": 2.9781,
+      "step": 45615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.430678646743735e-05,
+      "loss": 3.1502,
+      "step": 45620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.425589968735217e-05,
+      "loss": 3.0586,
+      "step": 45625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.420503063401511e-05,
+      "loss": 3.0531,
+      "step": 45630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.415417931125134e-05,
+      "loss": 3.0447,
+      "step": 45635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.41033457228845e-05,
+      "loss": 3.1198,
+      "step": 45640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.405252987273723e-05,
+      "loss": 3.1565,
+      "step": 45645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.400173176463047e-05,
+      "loss": 3.1355,
+      "step": 45650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.395095140238393e-05,
+      "loss": 3.0957,
+      "step": 45655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.390018878981627e-05,
+      "loss": 3.0195,
+      "step": 45660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.384944393074434e-05,
+      "loss": 3.1708,
+      "step": 45665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.379871682898392e-05,
+      "loss": 3.0612,
+      "step": 45670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.374800748834953e-05,
+      "loss": 3.1129,
+      "step": 45675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.369731591265423e-05,
+      "loss": 3.1009,
+      "step": 45680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.364664210570973e-05,
+      "loss": 3.0948,
+      "step": 45685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.359598607132644e-05,
+      "loss": 3.0709,
+      "step": 45690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.35453478133134e-05,
+      "loss": 3.043,
+      "step": 45695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.349472733547844e-05,
+      "loss": 3.1692,
+      "step": 45700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.344412464162795e-05,
+      "loss": 3.247,
+      "step": 45705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.33935397355669e-05,
+      "loss": 3.1907,
+      "step": 45710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.334297262109912e-05,
+      "loss": 3.3283,
+      "step": 45715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.329242330202686e-05,
+      "loss": 3.2082,
+      "step": 45720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.324189178215133e-05,
+      "loss": 3.1577,
+      "step": 45725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.319137806527219e-05,
+      "loss": 3.096,
+      "step": 45730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.314088215518776e-05,
+      "loss": 3.1203,
+      "step": 45735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.309040405569515e-05,
+      "loss": 3.082,
+      "step": 45740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.303994377058999e-05,
+      "loss": 3.1651,
+      "step": 45745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.298950130366659e-05,
+      "loss": 3.2569,
+      "step": 45750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.293907665871808e-05,
+      "loss": 3.0798,
+      "step": 45755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.288866983953612e-05,
+      "loss": 3.2079,
+      "step": 45760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.283828084991102e-05,
+      "loss": 3.2477,
+      "step": 45765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.278790969363175e-05,
+      "loss": 3.2101,
+      "step": 45770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.27375563744859e-05,
+      "loss": 3.2387,
+      "step": 45775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.268722089625992e-05,
+      "loss": 3.1574,
+      "step": 45780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.263690326273874e-05,
+      "loss": 3.1126,
+      "step": 45785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.258660347770597e-05,
+      "loss": 3.2076,
+      "step": 45790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.253632154494388e-05,
+      "loss": 3.1809,
+      "step": 45795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.248605746823339e-05,
+      "loss": 3.131,
+      "step": 45800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.243581125135417e-05,
+      "loss": 3.1605,
+      "step": 45805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.238558289808456e-05,
+      "loss": 3.1935,
+      "step": 45810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.233537241220117e-05,
+      "loss": 3.1822,
+      "step": 45815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.228517979747987e-05,
+      "loss": 3.1973,
+      "step": 45820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.223500505769476e-05,
+      "loss": 3.0712,
+      "step": 45825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.218484819661864e-05,
+      "loss": 3.2074,
+      "step": 45830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.213470921802335e-05,
+      "loss": 3.1901,
+      "step": 45835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.208458812567873e-05,
+      "loss": 3.1254,
+      "step": 45840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.203448492335385e-05,
+      "loss": 3.2218,
+      "step": 45845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.198439961481615e-05,
+      "loss": 3.194,
+      "step": 45850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.193433220383174e-05,
+      "loss": 3.1396,
+      "step": 45855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.188428269416561e-05,
+      "loss": 3.2048,
+      "step": 45860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.183425108958094e-05,
+      "loss": 3.085,
+      "step": 45865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.178423739384006e-05,
+      "loss": 3.1692,
+      "step": 45870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.173424161070372e-05,
+      "loss": 3.1756,
+      "step": 45875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.168426374393125e-05,
+      "loss": 3.1431,
+      "step": 45880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.163430379728086e-05,
+      "loss": 3.2415,
+      "step": 45885
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.158436177450931e-05,
+      "loss": 3.27,
+      "step": 45890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.153443767937175e-05,
+      "loss": 3.0112,
+      "step": 45895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.148453151562243e-05,
+      "loss": 3.1826,
+      "step": 45900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.143464328701394e-05,
+      "loss": 3.2231,
+      "step": 45905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.13847729972976e-05,
+      "loss": 3.1354,
+      "step": 45910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.133492065022358e-05,
+      "loss": 3.0341,
+      "step": 45915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.128508624954025e-05,
+      "loss": 3.1803,
+      "step": 45920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.12352697989951e-05,
+      "loss": 3.0783,
+      "step": 45925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.118547130233401e-05,
+      "loss": 3.1693,
+      "step": 45930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.113569076330152e-05,
+      "loss": 3.1872,
+      "step": 45935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.108592818564105e-05,
+      "loss": 3.1872,
+      "step": 45940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.103618357309423e-05,
+      "loss": 3.1362,
+      "step": 45945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.0986456929401795e-05,
+      "loss": 3.0724,
+      "step": 45950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.09367482583029e-05,
+      "loss": 3.2822,
+      "step": 45955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.0887057563535266e-05,
+      "loss": 3.1595,
+      "step": 45960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.0837384848835634e-05,
+      "loss": 3.0659,
+      "step": 45965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.0787730117938885e-05,
+      "loss": 3.192,
+      "step": 45970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.073809337457879e-05,
+      "loss": 3.1099,
+      "step": 45975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.068847462248797e-05,
+      "loss": 3.1255,
+      "step": 45980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.063887386539741e-05,
+      "loss": 3.2569,
+      "step": 45985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.058929110703685e-05,
+      "loss": 3.1493,
+      "step": 45990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.0539726351134585e-05,
+      "loss": 3.1143,
+      "step": 45995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.049017960141767e-05,
+      "loss": 3.1716,
+      "step": 46000
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 3.1786859035491943,
+      "eval_runtime": 150.5699,
+      "eval_samples_per_second": 12.227,
+      "eval_steps_per_second": 0.77,
+      "step": 46000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.04406508616118e-05,
+      "loss": 3.0864,
+      "step": 46005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.0391140135441295e-05,
+      "loss": 3.2294,
+      "step": 46010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.0341647426629056e-05,
+      "loss": 3.2553,
+      "step": 46015
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.029217273889672e-05,
+      "loss": 3.1414,
+      "step": 46020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.0242716075964424e-05,
+      "loss": 3.1521,
+      "step": 46025
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.0193277441551244e-05,
+      "loss": 3.2215,
+      "step": 46030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.014385683937457e-05,
+      "loss": 3.1964,
+      "step": 46035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.009445427315066e-05,
+      "loss": 2.9809,
+      "step": 46040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.0045069746594275e-05,
+      "loss": 3.1355,
+      "step": 46045
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.999570326341888e-05,
+      "loss": 3.1077,
+      "step": 46050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.994635482733654e-05,
+      "loss": 3.167,
+      "step": 46055
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.989702444205812e-05,
+      "loss": 3.2413,
+      "step": 46060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.9847712111292946e-05,
+      "loss": 3.1135,
+      "step": 46065
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.9798417838749094e-05,
+      "loss": 3.2011,
+      "step": 46070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.974914162813319e-05,
+      "loss": 3.1403,
+      "step": 46075
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.969988348315048e-05,
+      "loss": 3.1994,
+      "step": 46080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.965064340750513e-05,
+      "loss": 3.1759,
+      "step": 46085
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.960142140489967e-05,
+      "loss": 3.1331,
+      "step": 46090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.955221747903512e-05,
+      "loss": 3.2708,
+      "step": 46095
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.9503031633611654e-05,
+      "loss": 3.103,
+      "step": 46100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.94538638723276e-05,
+      "loss": 3.1251,
+      "step": 46105
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.9404714198880275e-05,
+      "loss": 3.1968,
+      "step": 46110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.9355582616965425e-05,
+      "loss": 3.0474,
+      "step": 46115
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.9306469130277456e-05,
+      "loss": 3.168,
+      "step": 46120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.925737374250949e-05,
+      "loss": 3.1837,
+      "step": 46125
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.920829645735322e-05,
+      "loss": 3.1468,
+      "step": 46130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.9159237278498947e-05,
+      "loss": 3.0686,
+      "step": 46135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.911019620963582e-05,
+      "loss": 3.1152,
+      "step": 46140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.906117325445141e-05,
+      "loss": 3.2782,
+      "step": 46145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.901216841663194e-05,
+      "loss": 3.089,
+      "step": 46150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.89631816998624e-05,
+      "loss": 3.1593,
+      "step": 46155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.891421310782621e-05,
+      "loss": 3.2204,
+      "step": 46160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.886526264420573e-05,
+      "loss": 3.1318,
+      "step": 46165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.881633031268179e-05,
+      "loss": 3.1237,
+      "step": 46170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.876741611693361e-05,
+      "loss": 3.0531,
+      "step": 46175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.8718520060639494e-05,
+      "loss": 3.1773,
+      "step": 46180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.866964214747615e-05,
+      "loss": 3.1155,
+      "step": 46185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.8620782381118826e-05,
+      "loss": 3.1278,
+      "step": 46190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.8571940765241755e-05,
+      "loss": 3.0726,
+      "step": 46195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.85231173035173e-05,
+      "loss": 3.2045,
+      "step": 46200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.8474311999616954e-05,
+      "loss": 3.1445,
+      "step": 46205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.842552485721055e-05,
+      "loss": 3.2156,
+      "step": 46210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.8376755879966554e-05,
+      "loss": 3.1789,
+      "step": 46215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.8328005071552364e-05,
+      "loss": 3.1581,
+      "step": 46220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.827927243563348e-05,
+      "loss": 3.2274,
+      "step": 46225
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.823055797587458e-05,
+      "loss": 3.0204,
+      "step": 46230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.8181861695938695e-05,
+      "loss": 3.1658,
+      "step": 46235
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.813318359948742e-05,
+      "loss": 3.0678,
+      "step": 46240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.8084523690181336e-05,
+      "loss": 3.1534,
+      "step": 46245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.803588197167922e-05,
+      "loss": 3.1554,
+      "step": 46250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.798725844763864e-05,
+      "loss": 3.1348,
+      "step": 46255
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.793865312171604e-05,
+      "loss": 3.1477,
+      "step": 46260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.7890065997566126e-05,
+      "loss": 3.2401,
+      "step": 46265
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.784149707884248e-05,
+      "loss": 3.2408,
+      "step": 46270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.779294636919721e-05,
+      "loss": 3.1458,
+      "step": 46275
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.774441387228104e-05,
+      "loss": 3.0876,
+      "step": 46280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.769589959174343e-05,
+      "loss": 3.081,
+      "step": 46285
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.7647403531232415e-05,
+      "loss": 3.1654,
+      "step": 46290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.7598925694394605e-05,
+      "loss": 3.1191,
+      "step": 46295
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.755046608487528e-05,
+      "loss": 3.1663,
+      "step": 46300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.750202470631833e-05,
+      "loss": 3.1126,
+      "step": 46305
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.74536015623664e-05,
+      "loss": 3.1349,
+      "step": 46310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.740519665666059e-05,
+      "loss": 3.1206,
+      "step": 46315
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.7356809992840715e-05,
+      "loss": 3.078,
+      "step": 46320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.730844157454523e-05,
+      "loss": 3.0895,
+      "step": 46325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.7260091405411124e-05,
+      "loss": 3.2058,
+      "step": 46330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.721175948907406e-05,
+      "loss": 3.2039,
+      "step": 46335
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.71634458291685e-05,
+      "loss": 3.1625,
+      "step": 46340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.71151504293273e-05,
+      "loss": 3.1704,
+      "step": 46345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.706687329318202e-05,
+      "loss": 3.112,
+      "step": 46350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.701861442436285e-05,
+      "loss": 3.1327,
+      "step": 46355
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.6970373826498536e-05,
+      "loss": 3.1424,
+      "step": 46360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.692215150321667e-05,
+      "loss": 3.1835,
+      "step": 46365
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.6873947458143255e-05,
+      "loss": 3.0898,
+      "step": 46370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.682576169490302e-05,
+      "loss": 3.0496,
+      "step": 46375
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.6777594217119224e-05,
+      "loss": 3.1112,
+      "step": 46380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.672944502841379e-05,
+      "loss": 3.201,
+      "step": 46385
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.668131413240744e-05,
+      "loss": 3.0926,
+      "step": 46390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.6633201532719256e-05,
+      "loss": 3.1919,
+      "step": 46395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.6585107232967095e-05,
+      "loss": 3.2124,
+      "step": 46400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.653703123676738e-05,
+      "loss": 3.2094,
+      "step": 46405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.6488973547735184e-05,
+      "loss": 3.0477,
+      "step": 46410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.6440934169484106e-05,
+      "loss": 3.1221,
+      "step": 46415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.639291310562667e-05,
+      "loss": 3.0849,
+      "step": 46420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.634491035977371e-05,
+      "loss": 3.053,
+      "step": 46425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.6296925935534746e-05,
+      "loss": 3.1511,
+      "step": 46430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.624895983651801e-05,
+      "loss": 3.1053,
+      "step": 46435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.620101206633021e-05,
+      "loss": 3.2083,
+      "step": 46440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.6153082628576937e-05,
+      "loss": 3.1903,
+      "step": 46445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.610517152686212e-05,
+      "loss": 3.1463,
+      "step": 46450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.605727876478852e-05,
+      "loss": 3.1161,
+      "step": 46455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.600940434595733e-05,
+      "loss": 3.1253,
+      "step": 46460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.596154827396842e-05,
+      "loss": 3.2649,
+      "step": 46465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.591371055242049e-05,
+      "loss": 3.1203,
+      "step": 46470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.5865891184910706e-05,
+      "loss": 3.1973,
+      "step": 46475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.581809017503457e-05,
+      "loss": 3.196,
+      "step": 46480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.5770307526386703e-05,
+      "loss": 3.0937,
+      "step": 46485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.572254324256004e-05,
+      "loss": 3.1002,
+      "step": 46490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.567479732714616e-05,
+      "loss": 3.2265,
+      "step": 46495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.5627069783735534e-05,
+      "loss": 3.0773,
+      "step": 46500
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 3.175149440765381,
+      "eval_runtime": 150.553,
+      "eval_samples_per_second": 12.228,
+      "eval_steps_per_second": 0.77,
+      "step": 46500
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 57783,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 1.3814529059671327e+18,
+  "trial_name": null,
+  "trial_params": null
+}