diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4584 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998749194192412,
+  "eval_steps": 500,
+  "global_step": 6495,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0015394533016462528,
+      "grad_norm": 4.0,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 1.4213,
+      "step": 10
+    },
+    {
+      "epoch": 0.0030789066032925055,
+      "grad_norm": 4.125,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 1.4854,
+      "step": 20
+    },
+    {
+      "epoch": 0.004618359904938758,
+      "grad_norm": 3.421875,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 1.4679,
+      "step": 30
+    },
+    {
+      "epoch": 0.006157813206585011,
+      "grad_norm": 3.328125,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.4725,
+      "step": 40
+    },
+    {
+      "epoch": 0.007697266508231265,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.4414,
+      "step": 50
+    },
+    {
+      "epoch": 0.009236719809877517,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.3931,
+      "step": 60
+    },
+    {
+      "epoch": 0.01077617311152377,
+      "grad_norm": 2.015625,
+      "learning_rate": 5.3846153846153855e-06,
+      "loss": 1.3508,
+      "step": 70
+    },
+    {
+      "epoch": 0.012315626413170022,
+      "grad_norm": 1.65625,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.3672,
+      "step": 80
+    },
+    {
+      "epoch": 0.013855079714816277,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 1.3791,
+      "step": 90
+    },
+    {
+      "epoch": 0.01539453301646253,
+      "grad_norm": 1.640625,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.3597,
+      "step": 100
+    },
+    {
+      "epoch": 0.01693398631810878,
+      "grad_norm": 1.703125,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 1.3254,
+      "step": 110
+    },
+    {
+      "epoch": 0.018473439619755033,
+      "grad_norm": 1.71875,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.3207,
+      "step": 120
+    },
+    {
+      "epoch": 0.020012892921401286,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1e-05,
+      "loss": 1.3245,
+      "step": 130
+    },
+    {
+      "epoch": 0.02155234622304754,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.0769230769230771e-05,
+      "loss": 1.2972,
+      "step": 140
+    },
+    {
+      "epoch": 0.02309179952469379,
+      "grad_norm": 1.6796875,
+      "learning_rate": 1.153846153846154e-05,
+      "loss": 1.2657,
+      "step": 150
+    },
+    {
+      "epoch": 0.024631252826340044,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 1.2835,
+      "step": 160
+    },
+    {
+      "epoch": 0.0261707061279863,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 1.2358,
+      "step": 170
+    },
+    {
+      "epoch": 0.027710159429632553,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 1.2516,
+      "step": 180
+    },
+    {
+      "epoch": 0.029249612731278806,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.4615384615384617e-05,
+      "loss": 1.2113,
+      "step": 190
+    },
+    {
+      "epoch": 0.03078906603292506,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.2447,
+      "step": 200
+    },
+    {
+      "epoch": 0.03232851933457131,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 1.2566,
+      "step": 210
+    },
+    {
+      "epoch": 0.03386797263621756,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 1.2361,
+      "step": 220
+    },
+    {
+      "epoch": 0.035407425937863814,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 1.2331,
+      "step": 230
+    },
+    {
+      "epoch": 0.036946879239510066,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 1.2507,
+      "step": 240
+    },
+    {
+      "epoch": 0.03848633254115632,
+      "grad_norm": 1.1796875,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 1.2425,
+      "step": 250
+    },
+    {
+      "epoch": 0.04002578584280257,
+      "grad_norm": 1.0703125,
+      "learning_rate": 2e-05,
+      "loss": 1.1911,
+      "step": 260
+    },
+    {
+      "epoch": 0.041565239144448825,
+      "grad_norm": 1.078125,
+      "learning_rate": 2.0769230769230772e-05,
+      "loss": 1.2153,
+      "step": 270
+    },
+    {
+      "epoch": 0.04310469244609508,
+      "grad_norm": 1.03125,
+      "learning_rate": 2.1538461538461542e-05,
+      "loss": 1.2405,
+      "step": 280
+    },
+    {
+      "epoch": 0.04464414574774133,
+      "grad_norm": 1.0859375,
+      "learning_rate": 2.230769230769231e-05,
+      "loss": 1.2489,
+      "step": 290
+    },
+    {
+      "epoch": 0.04618359904938758,
+      "grad_norm": 1.1015625,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 1.248,
+      "step": 300
+    },
+    {
+      "epoch": 0.047723052351033836,
+      "grad_norm": 1.109375,
+      "learning_rate": 2.384615384615385e-05,
+      "loss": 1.2451,
+      "step": 310
+    },
+    {
+      "epoch": 0.04926250565268009,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.461538461538462e-05,
+      "loss": 1.2182,
+      "step": 320
+    },
+    {
+      "epoch": 0.05080195895432634,
+      "grad_norm": 1.1171875,
+      "learning_rate": 2.5384615384615383e-05,
+      "loss": 1.2031,
+      "step": 330
+    },
+    {
+      "epoch": 0.0523414122559726,
+      "grad_norm": 1.0859375,
+      "learning_rate": 2.6153846153846157e-05,
+      "loss": 1.2058,
+      "step": 340
+    },
+    {
+      "epoch": 0.053880865557618854,
+      "grad_norm": 1.0703125,
+      "learning_rate": 2.6923076923076923e-05,
+      "loss": 1.239,
+      "step": 350
+    },
+    {
+      "epoch": 0.05542031885926511,
+      "grad_norm": 1.0859375,
+      "learning_rate": 2.7692307692307694e-05,
+      "loss": 1.2285,
+      "step": 360
+    },
+    {
+      "epoch": 0.05695977216091136,
+      "grad_norm": 1.0859375,
+      "learning_rate": 2.846153846153846e-05,
+      "loss": 1.232,
+      "step": 370
+    },
+    {
+      "epoch": 0.05849922546255761,
+      "grad_norm": 1.140625,
+      "learning_rate": 2.9230769230769234e-05,
+      "loss": 1.253,
+      "step": 380
+    },
+    {
+      "epoch": 0.060038678764203865,
+      "grad_norm": 1.0546875,
+      "learning_rate": 3e-05,
+      "loss": 1.1876,
+      "step": 390
+    },
+    {
+      "epoch": 0.06157813206585012,
+      "grad_norm": 1.0703125,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 1.1652,
+      "step": 400
+    },
+    {
+      "epoch": 0.06311758536749637,
+      "grad_norm": 1.2109375,
+      "learning_rate": 3.153846153846154e-05,
+      "loss": 1.1771,
+      "step": 410
+    },
+    {
+      "epoch": 0.06465703866914262,
+      "grad_norm": 1.1796875,
+      "learning_rate": 3.230769230769231e-05,
+      "loss": 1.2,
+      "step": 420
+    },
+    {
+      "epoch": 0.06619649197078888,
+      "grad_norm": 1.265625,
+      "learning_rate": 3.307692307692308e-05,
+      "loss": 1.2124,
+      "step": 430
+    },
+    {
+      "epoch": 0.06773594527243512,
+      "grad_norm": 1.0546875,
+      "learning_rate": 3.384615384615385e-05,
+      "loss": 1.2581,
+      "step": 440
+    },
+    {
+      "epoch": 0.06927539857408138,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.461538461538462e-05,
+      "loss": 1.2447,
+      "step": 450
+    },
+    {
+      "epoch": 0.07081485187572763,
+      "grad_norm": 1.15625,
+      "learning_rate": 3.538461538461539e-05,
+      "loss": 1.208,
+      "step": 460
+    },
+    {
+      "epoch": 0.07235430517737389,
+      "grad_norm": 1.1796875,
+      "learning_rate": 3.615384615384615e-05,
+      "loss": 1.2736,
+      "step": 470
+    },
+    {
+      "epoch": 0.07389375847902013,
+      "grad_norm": 1.1171875,
+      "learning_rate": 3.692307692307693e-05,
+      "loss": 1.2272,
+      "step": 480
+    },
+    {
+      "epoch": 0.07543321178066639,
+      "grad_norm": 0.96484375,
+      "learning_rate": 3.769230769230769e-05,
+      "loss": 1.2181,
+      "step": 490
+    },
+    {
+      "epoch": 0.07697266508231264,
+      "grad_norm": 1.0703125,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 1.2161,
+      "step": 500
+    },
+    {
+      "epoch": 0.0785121183839589,
+      "grad_norm": 1.1875,
+      "learning_rate": 3.923076923076923e-05,
+      "loss": 1.1964,
+      "step": 510
+    },
+    {
+      "epoch": 0.08005157168560514,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4e-05,
+      "loss": 1.227,
+      "step": 520
+    },
+    {
+      "epoch": 0.0815910249872514,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.0769230769230773e-05,
+      "loss": 1.2121,
+      "step": 530
+    },
+    {
+      "epoch": 0.08313047828889765,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.1538461538461544e-05,
+      "loss": 1.1781,
+      "step": 540
+    },
+    {
+      "epoch": 0.08466993159054391,
+      "grad_norm": 0.953125,
+      "learning_rate": 4.230769230769231e-05,
+      "loss": 1.2137,
+      "step": 550
+    },
+    {
+      "epoch": 0.08620938489219016,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.3076923076923084e-05,
+      "loss": 1.2057,
+      "step": 560
+    },
+    {
+      "epoch": 0.08774883819383641,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.384615384615385e-05,
+      "loss": 1.2393,
+      "step": 570
+    },
+    {
+      "epoch": 0.08928829149548266,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.461538461538462e-05,
+      "loss": 1.2319,
+      "step": 580
+    },
+    {
+      "epoch": 0.09082774479712892,
+      "grad_norm": 1.21875,
+      "learning_rate": 4.538461538461539e-05,
+      "loss": 1.1786,
+      "step": 590
+    },
+    {
+      "epoch": 0.09236719809877517,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 1.2016,
+      "step": 600
+    },
+    {
+      "epoch": 0.09390665140042143,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.692307692307693e-05,
+      "loss": 1.1856,
+      "step": 610
+    },
+    {
+      "epoch": 0.09544610470206767,
+      "grad_norm": 1.234375,
+      "learning_rate": 4.76923076923077e-05,
+      "loss": 1.2402,
+      "step": 620
+    },
+    {
+      "epoch": 0.09698555800371393,
+      "grad_norm": 1.1484375,
+      "learning_rate": 4.846153846153846e-05,
+      "loss": 1.2123,
+      "step": 630
+    },
+    {
+      "epoch": 0.09852501130536018,
+      "grad_norm": 1.1875,
+      "learning_rate": 4.923076923076924e-05,
+      "loss": 1.1946,
+      "step": 640
+    },
+    {
+      "epoch": 0.10006446460700644,
+      "grad_norm": 1.046875,
+      "learning_rate": 5e-05,
+      "loss": 1.1618,
+      "step": 650
+    },
+    {
+      "epoch": 0.10160391790865268,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.999963888986145e-05,
+      "loss": 1.2054,
+      "step": 660
+    },
+    {
+      "epoch": 0.10314337121029894,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.999855556987784e-05,
+      "loss": 1.2076,
+      "step": 670
+    },
+    {
+      "epoch": 0.1046828245119452,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.9996750071345e-05,
+      "loss": 1.2417,
+      "step": 680
+    },
+    {
+      "epoch": 0.10622227781359145,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.999422244642164e-05,
+      "loss": 1.1945,
+      "step": 690
+    },
+    {
+      "epoch": 0.10776173111523771,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.9990972768127816e-05,
+      "loss": 1.2154,
+      "step": 700
+    },
+    {
+      "epoch": 0.10930118441688395,
+      "grad_norm": 0.97265625,
+      "learning_rate": 4.998700113034289e-05,
+      "loss": 1.1973,
+      "step": 710
+    },
+    {
+      "epoch": 0.11084063771853021,
+      "grad_norm": 1.125,
+      "learning_rate": 4.9982307647802765e-05,
+      "loss": 1.2,
+      "step": 720
+    },
+    {
+      "epoch": 0.11238009102017646,
+      "grad_norm": 1.125,
+      "learning_rate": 4.997689245609655e-05,
+      "loss": 1.1884,
+      "step": 730
+    },
+    {
+      "epoch": 0.11391954432182272,
+      "grad_norm": 0.94921875,
+      "learning_rate": 4.99707557116627e-05,
+      "loss": 1.1956,
+      "step": 740
+    },
+    {
+      "epoch": 0.11545899762346896,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.996389759178448e-05,
+      "loss": 1.2056,
+      "step": 750
+    },
+    {
+      "epoch": 0.11699845092511522,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.99563182945848e-05,
+      "loss": 1.1736,
+      "step": 760
+    },
+    {
+      "epoch": 0.11853790422676147,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.994801803902056e-05,
+      "loss": 1.2311,
+      "step": 770
+    },
+    {
+      "epoch": 0.12007735752840773,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.993899706487626e-05,
+      "loss": 1.2538,
+      "step": 780
+    },
+    {
+      "epoch": 0.12161681083005398,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.992925563275714e-05,
+      "loss": 1.1951,
+      "step": 790
+    },
+    {
+      "epoch": 0.12315626413170024,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.991879402408157e-05,
+      "loss": 1.2059,
+      "step": 800
+    },
+    {
+      "epoch": 0.12469571743334648,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.9907612541073e-05,
+      "loss": 1.2064,
+      "step": 810
+    },
+    {
+      "epoch": 0.12623517073499274,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.989571150675118e-05,
+      "loss": 1.1775,
+      "step": 820
+    },
+    {
+      "epoch": 0.127774624036639,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.988309126492282e-05,
+      "loss": 1.164,
+      "step": 830
+    },
+    {
+      "epoch": 0.12931407733828523,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.986975218017174e-05,
+      "loss": 1.2446,
+      "step": 840
+    },
+    {
+      "epoch": 0.1308535306399315,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.985569463784821e-05,
+      "loss": 1.1994,
+      "step": 850
+    },
+    {
+      "epoch": 0.13239298394157775,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.984091904405793e-05,
+      "loss": 1.2029,
+      "step": 860
+    },
+    {
+      "epoch": 0.133932437243224,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.982542582565023e-05,
+      "loss": 1.1986,
+      "step": 870
+    },
+    {
+      "epoch": 0.13547189054487024,
+      "grad_norm": 1.15625,
+      "learning_rate": 4.980921543020577e-05,
+      "loss": 1.2325,
+      "step": 880
+    },
+    {
+      "epoch": 0.1370113438465165,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.97922883260236e-05,
+      "loss": 1.2313,
+      "step": 890
+    },
+    {
+      "epoch": 0.13855079714816276,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.977464500210765e-05,
+      "loss": 1.2085,
+      "step": 900
+    },
+    {
+      "epoch": 0.14009025044980902,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.9756285968152535e-05,
+      "loss": 1.2607,
+      "step": 910
+    },
+    {
+      "epoch": 0.14162970375145525,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.9737211754528966e-05,
+      "loss": 1.2213,
+      "step": 920
+    },
+    {
+      "epoch": 0.14316915705310151,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.971742291226827e-05,
+      "loss": 1.2146,
+      "step": 930
+    },
+    {
+      "epoch": 0.14470861035474777,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.9696920013046574e-05,
+      "loss": 1.2348,
+      "step": 940
+    },
+    {
+      "epoch": 0.14624806365639403,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.967570364916826e-05,
+      "loss": 1.2555,
+      "step": 950
+    },
+    {
+      "epoch": 0.14778751695804027,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.965377443354886e-05,
+      "loss": 1.2129,
+      "step": 960
+    },
+    {
+      "epoch": 0.14932697025968653,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.963113299969734e-05,
+      "loss": 1.22,
+      "step": 970
+    },
+    {
+      "epoch": 0.15086642356133279,
+      "grad_norm": 0.97265625,
+      "learning_rate": 4.9607780001697815e-05,
+      "loss": 1.237,
+      "step": 980
+    },
+    {
+      "epoch": 0.15240587686297905,
+      "grad_norm": 1.1484375,
+      "learning_rate": 4.9583716114190614e-05,
+      "loss": 1.2248,
+      "step": 990
+    },
+    {
+      "epoch": 0.15394533016462528,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.955894203235284e-05,
+      "loss": 1.2451,
+      "step": 1000
+    },
+    {
+      "epoch": 0.15548478346627154,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.953345847187828e-05,
+      "loss": 1.2024,
+      "step": 1010
+    },
+    {
+      "epoch": 0.1570242367679178,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.9507266168956686e-05,
+      "loss": 1.1975,
+      "step": 1020
+    },
+    {
+      "epoch": 0.15856369006956406,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.948036588025254e-05,
+      "loss": 1.2345,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1601031433712103,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.9452758382883226e-05,
+      "loss": 1.2226,
+      "step": 1040
+    },
+    {
+      "epoch": 0.16164259667285655,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.94244444743965e-05,
+      "loss": 1.2036,
+      "step": 1050
+    },
+    {
+      "epoch": 0.1631820499745028,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.9395424972747514e-05,
+      "loss": 1.1978,
+      "step": 1060
+    },
+    {
+      "epoch": 0.16472150327614907,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.936570071627518e-05,
+      "loss": 1.2293,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1662609565777953,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.933527256367792e-05,
+      "loss": 1.2682,
+      "step": 1080
+    },
+    {
+      "epoch": 0.16780040987944156,
+      "grad_norm": 1.0,
+      "learning_rate": 4.930414139398889e-05,
+      "loss": 1.205,
+      "step": 1090
+    },
+    {
+      "epoch": 0.16933986318108782,
+      "grad_norm": 1.0,
+      "learning_rate": 4.9272308106550566e-05,
+      "loss": 1.229,
+      "step": 1100
+    },
+    {
+      "epoch": 0.17087931648273408,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.9239773620988783e-05,
+      "loss": 1.2287,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1724187697843803,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.920653887718615e-05,
+      "loss": 1.2351,
+      "step": 1120
+    },
+    {
+      "epoch": 0.17395822308602657,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.9172604835254875e-05,
+      "loss": 1.2307,
+      "step": 1130
+    },
+    {
+      "epoch": 0.17549767638767283,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.913797247550912e-05,
+      "loss": 1.231,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1770371296893191,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.910264279843656e-05,
+      "loss": 1.1713,
+      "step": 1150
+    },
+    {
+      "epoch": 0.17857658299096532,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.9066616824669576e-05,
+      "loss": 1.19,
+      "step": 1160
+    },
+    {
+      "epoch": 0.18011603629261158,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.902989559495571e-05,
+      "loss": 1.2307,
+      "step": 1170
+    },
+    {
+      "epoch": 0.18165548959425784,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.899248017012763e-05,
+      "loss": 1.2423,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1831949428959041,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.895437163107248e-05,
+      "loss": 1.2439,
+      "step": 1190
+    },
+    {
+      "epoch": 0.18473439619755033,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.891557107870064e-05,
+      "loss": 1.2317,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1862738494991966,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.887607963391394e-05,
+      "loss": 1.2153,
+      "step": 1210
+    },
+    {
+      "epoch": 0.18781330280084285,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.883589843757327e-05,
+      "loss": 1.2074,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1893527561024891,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.8795028650465616e-05,
+      "loss": 1.2099,
+      "step": 1230
+    },
+    {
+      "epoch": 0.19089220940413534,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.875347145327054e-05,
+      "loss": 1.2024,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1924316627057816,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.871122804652606e-05,
+      "loss": 1.1947,
+      "step": 1250
+    },
+    {
+      "epoch": 0.19397111600742786,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.8668299650593966e-05,
+      "loss": 1.2361,
+      "step": 1260
+    },
+    {
+      "epoch": 0.19551056930907412,
+      "grad_norm": 1.0,
+      "learning_rate": 4.862468750562459e-05,
+      "loss": 1.2304,
+      "step": 1270
+    },
+    {
+      "epoch": 0.19705002261072035,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.8580392871520946e-05,
+      "loss": 1.2222,
+      "step": 1280
+    },
+    {
+      "epoch": 0.19858947591236661,
+      "grad_norm": 1.125,
+      "learning_rate": 4.853541702790235e-05,
+      "loss": 1.2122,
+      "step": 1290
+    },
+    {
+      "epoch": 0.20012892921401287,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.848976127406745e-05,
+      "loss": 1.2163,
+      "step": 1300
+    },
+    {
+      "epoch": 0.20166838251565913,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.8443426928956695e-05,
+      "loss": 1.2212,
+      "step": 1310
+    },
+    {
+      "epoch": 0.20320783581730537,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.839641533111422e-05,
+      "loss": 1.2074,
+      "step": 1320
+    },
+    {
+      "epoch": 0.20474728911895163,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.8348727838649206e-05,
+      "loss": 1.1857,
+      "step": 1330
+    },
+    {
+      "epoch": 0.20628674242059789,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.83003658291966e-05,
+      "loss": 1.233,
+      "step": 1340
+    },
+    {
+      "epoch": 0.20782619572224414,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.825133069987737e-05,
+      "loss": 1.1558,
+      "step": 1350
+    },
+    {
+      "epoch": 0.2093656490238904,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.820162386725811e-05,
+      "loss": 1.267,
+      "step": 1360
+    },
+    {
+      "epoch": 0.21090510232553664,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.815124676731009e-05,
+      "loss": 1.1837,
+      "step": 1370
+    },
+    {
+      "epoch": 0.2124445556271829,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.810020085536785e-05,
+      "loss": 1.2251,
+      "step": 1380
+    },
+    {
+      "epoch": 0.21398400892882916,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.80484876060871e-05,
+      "loss": 1.222,
+      "step": 1390
+    },
+    {
+      "epoch": 0.21552346223047542,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.799610851340213e-05,
+      "loss": 1.201,
+      "step": 1400
+    },
+    {
+      "epoch": 0.21706291553212165,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.7943065090482644e-05,
+      "loss": 1.1822,
+      "step": 1410
+    },
+    {
+      "epoch": 0.2186023688337679,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.7889358869690056e-05,
+      "loss": 1.1993,
+      "step": 1420
+    },
+    {
+      "epoch": 0.22014182213541417,
+      "grad_norm": 1.0,
+      "learning_rate": 4.7834991402533266e-05,
+      "loss": 1.2284,
+      "step": 1430
+    },
+    {
+      "epoch": 0.22168127543706043,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.777996425962373e-05,
+      "loss": 1.2241,
+      "step": 1440
+    },
+    {
+      "epoch": 0.22322072873870666,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.7724279030630205e-05,
+      "loss": 1.2111,
+      "step": 1450
+    },
+    {
+      "epoch": 0.22476018204035292,
+      "grad_norm": 0.953125,
+      "learning_rate": 4.766793732423274e-05,
+      "loss": 1.1791,
+      "step": 1460
+    },
+    {
+      "epoch": 0.22629963534199918,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.761094076807624e-05,
+      "loss": 1.2766,
+      "step": 1470
+    },
+    {
+      "epoch": 0.22783908864364544,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.755329100872346e-05,
+      "loss": 1.1749,
+      "step": 1480
+    },
+    {
+      "epoch": 0.22937854194529167,
+      "grad_norm": 0.9765625,
+      "learning_rate": 4.749498971160742e-05,
+      "loss": 1.1916,
+      "step": 1490
+    },
+    {
+      "epoch": 0.23091799524693793,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.7436038560983255e-05,
+      "loss": 1.2191,
+      "step": 1500
+    },
+    {
+      "epoch": 0.2324574485485842,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.7376439259879625e-05,
+      "loss": 1.1889,
+      "step": 1510
+    },
+    {
+      "epoch": 0.23399690185023045,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.731619353004948e-05,
+      "loss": 1.1923,
+      "step": 1520
+    },
+    {
+      "epoch": 0.23553635515187668,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.7255303111920337e-05,
+      "loss": 1.1966,
+      "step": 1530
+    },
+    {
+      "epoch": 0.23707580845352294,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.719376976454397e-05,
+      "loss": 1.2466,
+      "step": 1540
+    },
+    {
+      "epoch": 0.2386152617551692,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.7131595265545636e-05,
+      "loss": 1.1839,
+      "step": 1550
+    },
+    {
+      "epoch": 0.24015471505681546,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.7068781411072686e-05,
+      "loss": 1.1991,
+      "step": 1560
+    },
+    {
+      "epoch": 0.2416941683584617,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.70053300157427e-05,
+      "loss": 1.2379,
+      "step": 1570
+    },
+    {
+      "epoch": 0.24323362166010795,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.694124291259104e-05,
+      "loss": 1.2105,
+      "step": 1580
+    },
+    {
+      "epoch": 0.2447730749617542,
+      "grad_norm": 0.9765625,
+      "learning_rate": 4.6876521953017926e-05,
+      "loss": 1.1746,
+      "step": 1590
+    },
+    {
+      "epoch": 0.24631252826340047,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.681116900673495e-05,
+      "loss": 1.2292,
+      "step": 1600
+    },
+    {
+      "epoch": 0.2478519815650467,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.6745185961711e-05,
+      "loss": 1.2369,
+      "step": 1610
+    },
+    {
+      "epoch": 0.24939143486669296,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.667857472411782e-05,
+      "loss": 1.2159,
+      "step": 1620
+    },
+    {
+      "epoch": 0.2509308881683392,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.661133721827486e-05,
+      "loss": 1.1516,
+      "step": 1630
+    },
+    {
+      "epoch": 0.2524703414699855,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.6543475386593736e-05,
+      "loss": 1.1843,
+      "step": 1640
+    },
+    {
+      "epoch": 0.2540097947716317,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.647499118952207e-05,
+      "loss": 1.2279,
+      "step": 1650
+    },
+    {
+      "epoch": 0.255549248073278,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.64058866054869e-05,
+      "loss": 1.176,
+      "step": 1660
+    },
+    {
+      "epoch": 0.25708870137492423,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.6336163630837495e-05,
+      "loss": 1.2189,
+      "step": 1670
+    },
+    {
+      "epoch": 0.25862815467657047,
+      "grad_norm": 0.9765625,
+      "learning_rate": 4.62658242797877e-05,
+      "loss": 1.2056,
+      "step": 1680
+    },
+    {
+      "epoch": 0.26016760797821675,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.619487058435774e-05,
+      "loss": 1.1876,
+      "step": 1690
+    },
+    {
+      "epoch": 0.261707061279863,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.612330459431552e-05,
+      "loss": 1.1865,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2632465145815092,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.60511283771174e-05,
+      "loss": 1.1943,
+      "step": 1710
+    },
+    {
+      "epoch": 0.2647859678831555,
+      "grad_norm": 0.94921875,
+      "learning_rate": 4.597834401784848e-05,
+      "loss": 1.1785,
+      "step": 1720
+    },
+    {
+      "epoch": 0.26632542118480174,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.590495361916239e-05,
+      "loss": 1.1922,
+      "step": 1730
+    },
+    {
+      "epoch": 0.267864874486448,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.5830959301220464e-05,
+      "loss": 1.1962,
+      "step": 1740
+    },
+    {
+      "epoch": 0.26940432778809426,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.575636320163058e-05,
+      "loss": 1.2296,
+      "step": 1750
+    },
+    {
+      "epoch": 0.2709437810897405,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.5681167475385385e-05,
+      "loss": 1.2038,
+      "step": 1760
+    },
+    {
+      "epoch": 0.2724832343913868,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.560537429479998e-05,
+      "loss": 1.1718,
+      "step": 1770
+    },
+    {
+      "epoch": 0.274022687693033,
+      "grad_norm": 0.9296875,
+      "learning_rate": 4.552898584944927e-05,
+      "loss": 1.2036,
+      "step": 1780
+    },
+    {
+      "epoch": 0.27556214099467924,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.54520043461046e-05,
+      "loss": 1.2393,
+      "step": 1790
+    },
+    {
+      "epoch": 0.2771015942963255,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.53744320086701e-05,
+      "loss": 1.1675,
+      "step": 1800
+    },
+    {
+      "epoch": 0.27864104759797176,
+      "grad_norm": 1.125,
+      "learning_rate": 4.529627107811834e-05,
+      "loss": 1.2104,
+      "step": 1810
+    },
+    {
+      "epoch": 0.28018050089961805,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.52175238124257e-05,
+      "loss": 1.189,
+      "step": 1820
+    },
+    {
+      "epoch": 0.2817199542012643,
+      "grad_norm": 1.21875,
+      "learning_rate": 4.513819248650706e-05,
+      "loss": 1.1706,
+      "step": 1830
+    },
+    {
+      "epoch": 0.2832594075029105,
+      "grad_norm": 1.1484375,
+      "learning_rate": 4.5058279392150096e-05,
+      "loss": 1.2233,
+      "step": 1840
+    },
+    {
+      "epoch": 0.2847988608045568,
+      "grad_norm": 0.9296875,
+      "learning_rate": 4.49777868379491e-05,
+      "loss": 1.1989,
+      "step": 1850
+    },
+    {
+      "epoch": 0.28633831410620303,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.489671714923827e-05,
+      "loss": 1.1789,
+      "step": 1860
+    },
+    {
+      "epoch": 0.28787776740784926,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.4815072668024515e-05,
+      "loss": 1.1997,
+      "step": 1870
+    },
+    {
+      "epoch": 0.28941722070949555,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.4732855752919845e-05,
+      "loss": 1.2101,
+      "step": 1880
+    },
+    {
+      "epoch": 0.2909566740111418,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.465006877907317e-05,
+      "loss": 1.2279,
+      "step": 1890
+    },
+    {
+      "epoch": 0.29249612731278807,
+      "grad_norm": 0.97265625,
+      "learning_rate": 4.4566714138101756e-05,
+      "loss": 1.1862,
+      "step": 1900
+    },
+    {
+      "epoch": 0.2940355806144343,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.448279423802207e-05,
+      "loss": 1.2302,
+      "step": 1910
+    },
+    {
+      "epoch": 0.29557503391608053,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.439831150318024e-05,
+      "loss": 1.2218,
+      "step": 1920
+    },
+    {
+      "epoch": 0.2971144872177268,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.431326837418206e-05,
+      "loss": 1.2114,
+      "step": 1930
+    },
+    {
+      "epoch": 0.29865394051937305,
+      "grad_norm": 0.9140625,
+      "learning_rate": 4.42276673078224e-05,
+      "loss": 1.2029,
+      "step": 1940
+    },
+    {
+      "epoch": 0.3001933938210193,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.41415107770143e-05,
+      "loss": 1.2153,
+      "step": 1950
+    },
+    {
+      "epoch": 0.30173284712266557,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.4054801270717496e-05,
+      "loss": 1.1641,
+      "step": 1960
+    },
+    {
+      "epoch": 0.3032723004243118,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3967541293866535e-05,
+      "loss": 1.1981,
+      "step": 1970
+    },
+    {
+      "epoch": 0.3048117537259581,
+      "grad_norm": 1.0,
+      "learning_rate": 4.3879733367298405e-05,
+      "loss": 1.1873,
+      "step": 1980
+    },
+    {
+      "epoch": 0.3063512070276043,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3791380027679716e-05,
+      "loss": 1.2043,
+      "step": 1990
+    },
+    {
+      "epoch": 0.30789066032925055,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.3702483827433395e-05,
+      "loss": 1.2042,
+      "step": 2000
+    },
+    {
+      "epoch": 0.30943011363089684,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.361304733466498e-05,
+      "loss": 1.2064,
+      "step": 2010
+    },
+    {
+      "epoch": 0.3109695669325431,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.3523073133088415e-05,
+      "loss": 1.1705,
+      "step": 2020
+    },
+    {
+      "epoch": 0.31250902023418936,
+      "grad_norm": 0.97265625,
+      "learning_rate": 4.3432563821951414e-05,
+      "loss": 1.1561,
+      "step": 2030
+    },
+    {
+      "epoch": 0.3140484735358356,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.334152201596037e-05,
+      "loss": 1.1948,
+      "step": 2040
+    },
+    {
+      "epoch": 0.3155879268374818,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.3249950345204806e-05,
+      "loss": 1.1804,
+      "step": 2050
+    },
+    {
+      "epoch": 0.3171273801391281,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.315785145508142e-05,
+      "loss": 1.194,
+      "step": 2060
+    },
+    {
+      "epoch": 0.31866683344077434,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.306522800621767e-05,
+      "loss": 1.203,
+      "step": 2070
+    },
+    {
+      "epoch": 0.3202062867424206,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.297208267439485e-05,
+      "loss": 1.1987,
+      "step": 2080
+    },
+    {
+      "epoch": 0.32174574004406686,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.2878418150470865e-05,
+      "loss": 1.1579,
+      "step": 2090
+    },
+    {
+      "epoch": 0.3232851933457131,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.278423714030245e-05,
+      "loss": 1.2011,
+      "step": 2100
+    },
+    {
+      "epoch": 0.3248246466473594,
+      "grad_norm": 0.94921875,
+      "learning_rate": 4.268954236466701e-05,
+      "loss": 1.1842,
+      "step": 2110
+    },
+    {
+      "epoch": 0.3263640999490056,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.259433655918404e-05,
+      "loss": 1.2112,
+      "step": 2120
+    },
+    {
+      "epoch": 0.32790355325065185,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.249862247423606e-05,
+      "loss": 1.1673,
+      "step": 2130
+    },
+    {
+      "epoch": 0.32944300655229813,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.240240287488918e-05,
+      "loss": 1.2076,
+      "step": 2140
+    },
+    {
+      "epoch": 0.33098245985394437,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.230568054081325e-05,
+      "loss": 1.2097,
+      "step": 2150
+    },
+    {
+      "epoch": 0.3325219131555906,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.22084582662015e-05,
+      "loss": 1.2014,
+      "step": 2160
+    },
+    {
+      "epoch": 0.3340613664572369,
+      "grad_norm": 1.1484375,
+      "learning_rate": 4.211073885968984e-05,
+      "loss": 1.1854,
+      "step": 2170
+    },
+    {
+      "epoch": 0.3356008197588831,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.201252514427576e-05,
+      "loss": 1.2184,
+      "step": 2180
+    },
+    {
+      "epoch": 0.3371402730605294,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.191381995723672e-05,
+      "loss": 1.1831,
+      "step": 2190
+    },
+    {
+      "epoch": 0.33867972636217564,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.181462615004823e-05,
+      "loss": 1.2157,
+      "step": 2200
+    },
+    {
+      "epoch": 0.34021917966382187,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.171494658830144e-05,
+      "loss": 1.1835,
+      "step": 2210
+    },
+    {
+      "epoch": 0.34175863296546816,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.161478415162038e-05,
+      "loss": 1.2038,
+      "step": 2220
+    },
+    {
+      "epoch": 0.3432980862671144,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.151414173357876e-05,
+      "loss": 1.214,
+      "step": 2230
+    },
+    {
+      "epoch": 0.3448375395687606,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.1413022241616384e-05,
+      "loss": 1.2467,
+      "step": 2240
+    },
+    {
+      "epoch": 0.3463769928704069,
+      "grad_norm": 0.9765625,
+      "learning_rate": 4.1311428596955145e-05,
+      "loss": 1.1849,
+      "step": 2250
+    },
+    {
+      "epoch": 0.34791644617205314,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.1209363734514674e-05,
+      "loss": 1.2335,
+      "step": 2260
+    },
+    {
+      "epoch": 0.3494558994736994,
+      "grad_norm": 0.9765625,
+      "learning_rate": 4.1106830602827466e-05,
+      "loss": 1.2058,
+      "step": 2270
+    },
+    {
+      "epoch": 0.35099535277534566,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.100383216395382e-05,
+      "loss": 1.2037,
+      "step": 2280
+    },
+    {
+      "epoch": 0.3525348060769919,
+      "grad_norm": 0.96484375,
+      "learning_rate": 4.0900371393396165e-05,
+      "loss": 1.2077,
+      "step": 2290
+    },
+    {
+      "epoch": 0.3540742593786382,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.079645128001316e-05,
+      "loss": 1.2013,
+      "step": 2300
+    },
+    {
+      "epoch": 0.3556137126802844,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.0692074825933335e-05,
+      "loss": 1.2023,
+      "step": 2310
+    },
+    {
+      "epoch": 0.35715316598193064,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.058724504646834e-05,
+      "loss": 1.2467,
+      "step": 2320
+    },
+    {
+      "epoch": 0.35869261928357693,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.048196497002588e-05,
+      "loss": 1.1872,
+      "step": 2330
+    },
+    {
+      "epoch": 0.36023207258522316,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.037623763802218e-05,
+      "loss": 1.186,
+      "step": 2340
+    },
+    {
+      "epoch": 0.36177152588686945,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.027006610479419e-05,
+      "loss": 1.1928,
+      "step": 2350
+    },
+    {
+      "epoch": 0.3633109791885157,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.016345343751125e-05,
+      "loss": 1.1928,
+      "step": 2360
+    },
+    {
+      "epoch": 0.3648504324901619,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.005640271608656e-05,
+      "loss": 1.2408,
+      "step": 2370
+    },
+    {
+      "epoch": 0.3663898857918082,
+      "grad_norm": 1.15625,
+      "learning_rate": 3.994891703308822e-05,
+      "loss": 1.1955,
+      "step": 2380
+    },
+    {
+      "epoch": 0.36792933909345443,
+      "grad_norm": 0.9609375,
+      "learning_rate": 3.984099949364979e-05,
+      "loss": 1.1629,
+      "step": 2390
+    },
+    {
+      "epoch": 0.36946879239510066,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.973265321538069e-05,
+      "loss": 1.2194,
+      "step": 2400
+    },
+    {
+      "epoch": 0.37100824569674695,
+      "grad_norm": 1.046875,
+      "learning_rate": 3.962388132827608e-05,
+      "loss": 1.1751,
+      "step": 2410
+    },
+    {
+      "epoch": 0.3725476989983932,
+      "grad_norm": 0.94140625,
+      "learning_rate": 3.9514686974626465e-05,
+      "loss": 1.1944,
+      "step": 2420
+    },
+    {
+      "epoch": 0.37408715230003947,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.9405073308926895e-05,
+      "loss": 1.1654,
+      "step": 2430
+    },
+    {
+      "epoch": 0.3756266056016857,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.929504349778585e-05,
+      "loss": 1.2007,
+      "step": 2440
+    },
+    {
+      "epoch": 0.37716605890333194,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.9184600719833756e-05,
+      "loss": 1.2444,
+      "step": 2450
+    },
+    {
+      "epoch": 0.3787055122049782,
+      "grad_norm": 1.0,
+      "learning_rate": 3.907374816563118e-05,
+      "loss": 1.1839,
+      "step": 2460
+    },
+    {
+      "epoch": 0.38024496550662445,
+      "grad_norm": 1.0390625,
+      "learning_rate": 3.8962489037576586e-05,
+      "loss": 1.2032,
+      "step": 2470
+    },
+    {
+      "epoch": 0.3817844188082707,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.8850826549813926e-05,
+      "loss": 1.1615,
+      "step": 2480
+    },
+    {
+      "epoch": 0.383323872109917,
+      "grad_norm": 0.91015625,
+      "learning_rate": 3.873876392813971e-05,
+      "loss": 1.1735,
+      "step": 2490
+    },
+    {
+      "epoch": 0.3848633254115632,
+      "grad_norm": 1.078125,
+      "learning_rate": 3.8626304409909856e-05,
+      "loss": 1.2034,
+      "step": 2500
+    },
+    {
+      "epoch": 0.3864027787132095,
+      "grad_norm": 1.0390625,
+      "learning_rate": 3.8513451243946115e-05,
+      "loss": 1.2183,
+      "step": 2510
+    },
+    {
+      "epoch": 0.3879422320148557,
+      "grad_norm": 1.0078125,
+      "learning_rate": 3.84002076904423e-05,
+      "loss": 1.18,
+      "step": 2520
+    },
+    {
+      "epoch": 0.38948168531650196,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.828657702087003e-05,
+      "loss": 1.2019,
+      "step": 2530
+    },
+    {
+      "epoch": 0.39102113861814825,
+      "grad_norm": 0.984375,
+      "learning_rate": 3.817256251788425e-05,
+      "loss": 1.1767,
+      "step": 2540
+    },
+    {
+      "epoch": 0.3925605919197945,
+      "grad_norm": 1.0546875,
+      "learning_rate": 3.8058167475228395e-05,
+      "loss": 1.1719,
+      "step": 2550
+    },
+    {
+      "epoch": 0.3941000452214407,
+      "grad_norm": 1.109375,
+      "learning_rate": 3.794339519763924e-05,
+      "loss": 1.2031,
+      "step": 2560
+    },
+    {
+      "epoch": 0.395639498523087,
+      "grad_norm": 0.984375,
+      "learning_rate": 3.7828249000751446e-05,
+      "loss": 1.2315,
+      "step": 2570
+    },
+    {
+      "epoch": 0.39717895182473323,
+      "grad_norm": 1.046875,
+      "learning_rate": 3.7712732211001724e-05,
+      "loss": 1.1983,
+      "step": 2580
+    },
+    {
+      "epoch": 0.3987184051263795,
+      "grad_norm": 1.0,
+      "learning_rate": 3.759684816553279e-05,
+      "loss": 1.1737,
+      "step": 2590
+    },
+    {
+      "epoch": 0.40025785842802575,
+      "grad_norm": 1.109375,
+      "learning_rate": 3.7480600212096964e-05,
+      "loss": 1.2017,
+      "step": 2600
+    },
+    {
+      "epoch": 0.401797311729672,
+      "grad_norm": 0.97265625,
+      "learning_rate": 3.7363991708959386e-05,
+      "loss": 1.1808,
+      "step": 2610
+    },
+    {
+      "epoch": 0.40333676503131827,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.724702602480108e-05,
+      "loss": 1.192,
+      "step": 2620
+    },
+    {
+      "epoch": 0.4048762183329645,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.71297065386216e-05,
+      "loss": 1.2062,
+      "step": 2630
+    },
+    {
+      "epoch": 0.40641567163461073,
+      "grad_norm": 1.0546875,
+      "learning_rate": 3.7012036639641425e-05,
+      "loss": 1.22,
+      "step": 2640
+    },
+    {
+      "epoch": 0.407955124936257,
+      "grad_norm": 1.03125,
+      "learning_rate": 3.689401972720404e-05,
+      "loss": 1.2256,
+      "step": 2650
+    },
+    {
+      "epoch": 0.40949457823790325,
+      "grad_norm": 1.125,
+      "learning_rate": 3.6775659210677716e-05,
+      "loss": 1.1889,
+      "step": 2660
+    },
+    {
+      "epoch": 0.41103403153954954,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.6656958509357066e-05,
+      "loss": 1.1933,
+      "step": 2670
+    },
+    {
+      "epoch": 0.41257348484119577,
+      "grad_norm": 1.03125,
+      "learning_rate": 3.653792105236422e-05,
+      "loss": 1.2477,
+      "step": 2680
+    },
+    {
+      "epoch": 0.414112938142842,
+      "grad_norm": 1.0703125,
+      "learning_rate": 3.6418550278549794e-05,
+      "loss": 1.1728,
+      "step": 2690
+    },
+    {
+      "epoch": 0.4156523914444883,
+      "grad_norm": 1.046875,
+      "learning_rate": 3.6298849636393525e-05,
+      "loss": 1.1445,
+      "step": 2700
+    },
+    {
+      "epoch": 0.4171918447461345,
+      "grad_norm": 0.98046875,
+      "learning_rate": 3.6178822583904636e-05,
+      "loss": 1.1985,
+      "step": 2710
+    },
+    {
+      "epoch": 0.4187312980477808,
+      "grad_norm": 1.0078125,
+      "learning_rate": 3.605847258852197e-05,
+      "loss": 1.1977,
+      "step": 2720
+    },
+    {
+      "epoch": 0.42027075134942704,
+      "grad_norm": 1.0546875,
+      "learning_rate": 3.593780312701383e-05,
+      "loss": 1.1851,
+      "step": 2730
+    },
+    {
+      "epoch": 0.4218102046510733,
+      "grad_norm": 0.99609375,
+      "learning_rate": 3.5816817685377474e-05,
+      "loss": 1.1372,
+      "step": 2740
+    },
+    {
+      "epoch": 0.42334965795271956,
+      "grad_norm": 1.125,
+      "learning_rate": 3.569551975873847e-05,
+      "loss": 1.1996,
+      "step": 2750
+    },
+    {
+      "epoch": 0.4248891112543658,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.5573912851249716e-05,
+      "loss": 1.1584,
+      "step": 2760
+    },
+    {
+      "epoch": 0.426428564556012,
+      "grad_norm": 0.9921875,
+      "learning_rate": 3.545200047599018e-05,
+      "loss": 1.1875,
+      "step": 2770
+    },
+    {
+      "epoch": 0.4279680178576583,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.5329786154863444e-05,
+      "loss": 1.1846,
+      "step": 2780
+    },
+    {
+      "epoch": 0.42950747115930454,
+      "grad_norm": 1.03125,
+      "learning_rate": 3.5207273418495944e-05,
+      "loss": 1.2036,
+      "step": 2790
+    },
+    {
+      "epoch": 0.43104692446095083,
+      "grad_norm": 1.125,
+      "learning_rate": 3.508446580613497e-05,
+      "loss": 1.2134,
+      "step": 2800
+    },
+    {
+      "epoch": 0.43258637776259706,
+      "grad_norm": 1.0390625,
+      "learning_rate": 3.496136686554644e-05,
+      "loss": 1.2302,
+      "step": 2810
+    },
+    {
+      "epoch": 0.4341258310642433,
+      "grad_norm": 1.0546875,
+      "learning_rate": 3.483798015291239e-05,
+      "loss": 1.1964,
+      "step": 2820
+    },
+    {
+      "epoch": 0.4356652843658896,
+      "grad_norm": 0.91015625,
+      "learning_rate": 3.471430923272826e-05,
+      "loss": 1.2308,
+      "step": 2830
+    },
+    {
+      "epoch": 0.4372047376675358,
+      "grad_norm": 0.984375,
+      "learning_rate": 3.459035767769989e-05,
+      "loss": 1.1798,
+      "step": 2840
+    },
+    {
+      "epoch": 0.43874419096918205,
+      "grad_norm": 1.0546875,
+      "learning_rate": 3.4466129068640336e-05,
+      "loss": 1.1828,
+      "step": 2850
+    },
+    {
+      "epoch": 0.44028364427082833,
+      "grad_norm": 0.94921875,
+      "learning_rate": 3.434162699436642e-05,
+      "loss": 1.1762,
+      "step": 2860
+    },
+    {
+      "epoch": 0.44182309757247457,
+      "grad_norm": 1.1484375,
+      "learning_rate": 3.421685505159505e-05,
+      "loss": 1.2141,
+      "step": 2870
+    },
+    {
+      "epoch": 0.44336255087412085,
+      "grad_norm": 0.98828125,
+      "learning_rate": 3.409181684483931e-05,
+      "loss": 1.2184,
+      "step": 2880
+    },
+    {
+      "epoch": 0.4449020041757671,
+      "grad_norm": 1.0625,
+      "learning_rate": 3.396651598630432e-05,
+      "loss": 1.1672,
+      "step": 2890
+    },
+    {
+      "epoch": 0.4464414574774133,
+      "grad_norm": 0.95703125,
+      "learning_rate": 3.384095609578293e-05,
+      "loss": 1.2005,
+      "step": 2900
+    },
+    {
+      "epoch": 0.4479809107790596,
+      "grad_norm": 1.0390625,
+      "learning_rate": 3.3715140800551074e-05,
+      "loss": 1.1804,
+      "step": 2910
+    },
+    {
+      "epoch": 0.44952036408070584,
+      "grad_norm": 0.98046875,
+      "learning_rate": 3.3589073735263067e-05,
+      "loss": 1.1754,
+      "step": 2920
+    },
+    {
+      "epoch": 0.45105981738235207,
+      "grad_norm": 1.046875,
+      "learning_rate": 3.3462758541846535e-05,
+      "loss": 1.204,
+      "step": 2930
+    },
+    {
+      "epoch": 0.45259927068399836,
+      "grad_norm": 1.0390625,
+      "learning_rate": 3.3336198869397235e-05,
+      "loss": 1.1911,
+      "step": 2940
+    },
+    {
+      "epoch": 0.4541387239856446,
+      "grad_norm": 0.99609375,
+      "learning_rate": 3.3209398374073645e-05,
+      "loss": 1.1929,
+      "step": 2950
+    },
+    {
+      "epoch": 0.4556781772872909,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.3082360718991304e-05,
+      "loss": 1.181,
+      "step": 2960
+    },
+    {
+      "epoch": 0.4572176305889371,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.295508957411704e-05,
+      "loss": 1.1937,
+      "step": 2970
+    },
+    {
+      "epoch": 0.45875708389058334,
+      "grad_norm": 0.94140625,
+      "learning_rate": 3.2827588616162914e-05,
+      "loss": 1.1827,
+      "step": 2980
+    },
+    {
+      "epoch": 0.4602965371922296,
+      "grad_norm": 0.9921875,
+      "learning_rate": 3.269986152848001e-05,
+      "loss": 1.1699,
+      "step": 2990
+    },
+    {
+      "epoch": 0.46183599049387586,
+      "grad_norm": 1.046875,
+      "learning_rate": 3.257191200095204e-05,
+      "loss": 1.1883,
+      "step": 3000
+    },
+    {
+      "epoch": 0.4633754437955221,
+      "grad_norm": 0.95703125,
+      "learning_rate": 3.244374372988873e-05,
+      "loss": 1.1331,
+      "step": 3010
+    },
+    {
+      "epoch": 0.4649148970971684,
+      "grad_norm": 0.9921875,
+      "learning_rate": 3.231536041791905e-05,
+      "loss": 1.1812,
+      "step": 3020
+    },
+    {
+      "epoch": 0.4664543503988146,
+      "grad_norm": 1.125,
+      "learning_rate": 3.2186765773884245e-05,
+      "loss": 1.2116,
+      "step": 3030
+    },
+    {
+      "epoch": 0.4679938037004609,
+      "grad_norm": 1.0390625,
+      "learning_rate": 3.205796351273068e-05,
+      "loss": 1.2497,
+      "step": 3040
+    },
+    {
+      "epoch": 0.46953325700210713,
+      "grad_norm": 1.0390625,
+      "learning_rate": 3.192895735540257e-05,
+      "loss": 1.2368,
+      "step": 3050
+    },
+    {
+      "epoch": 0.47107271030375336,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.1799751028734406e-05,
+      "loss": 1.1682,
+      "step": 3060
+    },
+    {
+      "epoch": 0.47261216360539965,
+      "grad_norm": 1.0546875,
+      "learning_rate": 3.1670348265343356e-05,
+      "loss": 1.1793,
+      "step": 3070
+    },
+    {
+      "epoch": 0.4741516169070459,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.1540752803521403e-05,
+      "loss": 1.2154,
+      "step": 3080
+    },
+    {
+      "epoch": 0.4756910702086921,
+      "grad_norm": 1.03125,
+      "learning_rate": 3.141096838712737e-05,
+      "loss": 1.1915,
+      "step": 3090
+    },
+    {
+      "epoch": 0.4772305235103384,
+      "grad_norm": 1.0703125,
+      "learning_rate": 3.1280998765478727e-05,
+      "loss": 1.2009,
+      "step": 3100
+    },
+    {
+      "epoch": 0.47876997681198463,
+      "grad_norm": 1.0859375,
+      "learning_rate": 3.1150847693243337e-05,
+      "loss": 1.2399,
+      "step": 3110
+    },
+    {
+      "epoch": 0.4803094301136309,
+      "grad_norm": 1.0703125,
+      "learning_rate": 3.1020518930330914e-05,
+      "loss": 1.1628,
+      "step": 3120
+    },
+    {
+      "epoch": 0.48184888341527715,
+      "grad_norm": 0.98046875,
+      "learning_rate": 3.089001624178449e-05,
+      "loss": 1.1774,
+      "step": 3130
+    },
+    {
+      "epoch": 0.4833883367169234,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.075934339767158e-05,
+      "loss": 1.1452,
+      "step": 3140
+    },
+    {
+      "epoch": 0.48492779001856967,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.0628504172975286e-05,
+      "loss": 1.1671,
+      "step": 3150
+    },
+    {
+      "epoch": 0.4864672433202159,
+      "grad_norm": 0.96875,
+      "learning_rate": 3.049750234748525e-05,
+      "loss": 1.1931,
+      "step": 3160
+    },
+    {
+      "epoch": 0.48800669662186213,
+      "grad_norm": 1.0,
+      "learning_rate": 3.0366341705688468e-05,
+      "loss": 1.2028,
+      "step": 3170
+    },
+    {
+      "epoch": 0.4895461499235084,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.023502603665993e-05,
+      "loss": 1.18,
+      "step": 3180
+    },
+    {
+      "epoch": 0.49108560322515465,
+      "grad_norm": 1.046875,
+      "learning_rate": 3.0103559133953208e-05,
+      "loss": 1.1694,
+      "step": 3190
+    },
+    {
+      "epoch": 0.49262505652680094,
+      "grad_norm": 1.0390625,
+      "learning_rate": 2.997194479549081e-05,
+      "loss": 1.2034,
+      "step": 3200
+    },
+    {
+      "epoch": 0.4941645098284472,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.9840186823454485e-05,
+      "loss": 1.2064,
+      "step": 3210
+    },
+    {
+      "epoch": 0.4957039631300934,
+      "grad_norm": 1.0,
+      "learning_rate": 2.9708289024175406e-05,
+      "loss": 1.2035,
+      "step": 3220
+    },
+    {
+      "epoch": 0.4972434164317397,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.9576255208024184e-05,
+      "loss": 1.2081,
+      "step": 3230
+    },
+    {
+      "epoch": 0.4987828697333859,
+      "grad_norm": 1.0546875,
+      "learning_rate": 2.9444089189300783e-05,
+      "loss": 1.1803,
+      "step": 3240
+    },
+    {
+      "epoch": 0.5003223230350322,
+      "grad_norm": 1.1328125,
+      "learning_rate": 2.931179478612436e-05,
+      "loss": 1.2337,
+      "step": 3250
+    },
+    {
+      "epoch": 0.5018617763366784,
+      "grad_norm": 0.91796875,
+      "learning_rate": 2.9179375820322917e-05,
+      "loss": 1.197,
+      "step": 3260
+    },
+    {
+      "epoch": 0.5034012296383247,
+      "grad_norm": 1.0234375,
+      "learning_rate": 2.904683611732295e-05,
+      "loss": 1.1522,
+      "step": 3270
+    },
+    {
+      "epoch": 0.504940682939971,
+      "grad_norm": 0.94921875,
+      "learning_rate": 2.8914179506038907e-05,
+      "loss": 1.2044,
+      "step": 3280
+    },
+    {
+      "epoch": 0.5064801362416171,
+      "grad_norm": 1.0,
+      "learning_rate": 2.8781409818762562e-05,
+      "loss": 1.1569,
+      "step": 3290
+    },
+    {
+      "epoch": 0.5080195895432634,
+      "grad_norm": 0.99609375,
+      "learning_rate": 2.864853089105234e-05,
+      "loss": 1.2165,
+      "step": 3300
+    },
+    {
+      "epoch": 0.5095590428449097,
+      "grad_norm": 1.0859375,
+      "learning_rate": 2.8515546561622462e-05,
+      "loss": 1.1779,
+      "step": 3310
+    },
+    {
+      "epoch": 0.511098496146556,
+      "grad_norm": 0.89453125,
+      "learning_rate": 2.838246067223211e-05,
+      "loss": 1.1369,
+      "step": 3320
+    },
+    {
+      "epoch": 0.5126379494482022,
+      "grad_norm": 1.1171875,
+      "learning_rate": 2.82492770675744e-05,
+      "loss": 1.1904,
+      "step": 3330
+    },
+    {
+      "epoch": 0.5141774027498485,
+      "grad_norm": 1.0859375,
+      "learning_rate": 2.8115999595165322e-05,
+      "loss": 1.1825,
+      "step": 3340
+    },
+    {
+      "epoch": 0.5157168560514948,
+      "grad_norm": 1.0703125,
+      "learning_rate": 2.7982632105232605e-05,
+      "loss": 1.1797,
+      "step": 3350
+    },
+    {
+      "epoch": 0.5172563093531409,
+      "grad_norm": 1.0625,
+      "learning_rate": 2.7849178450604463e-05,
+      "loss": 1.1895,
+      "step": 3360
+    },
+    {
+      "epoch": 0.5187957626547872,
+      "grad_norm": 1.0859375,
+      "learning_rate": 2.771564248659832e-05,
+      "loss": 1.179,
+      "step": 3370
+    },
+    {
+      "epoch": 0.5203352159564335,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.7582028070909415e-05,
+      "loss": 1.1896,
+      "step": 3380
+    },
+    {
+      "epoch": 0.5218746692580797,
+      "grad_norm": 0.984375,
+      "learning_rate": 2.7448339063499346e-05,
+      "loss": 1.1803,
+      "step": 3390
+    },
+    {
+      "epoch": 0.523414122559726,
+      "grad_norm": 0.99609375,
+      "learning_rate": 2.7314579326484608e-05,
+      "loss": 1.1481,
+      "step": 3400
+    },
+    {
+      "epoch": 0.5249535758613723,
+      "grad_norm": 1.1171875,
+      "learning_rate": 2.718075272402497e-05,
+      "loss": 1.2032,
+      "step": 3410
+    },
+    {
+      "epoch": 0.5264930291630184,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.7046863122211864e-05,
+      "loss": 1.1887,
+      "step": 3420
+    },
+    {
+      "epoch": 0.5280324824646647,
+      "grad_norm": 1.03125,
+      "learning_rate": 2.69129143889567e-05,
+      "loss": 1.164,
+      "step": 3430
+    },
+    {
+      "epoch": 0.529571935766311,
+      "grad_norm": 0.9453125,
+      "learning_rate": 2.6778910393879136e-05,
+      "loss": 1.1291,
+      "step": 3440
+    },
+    {
+      "epoch": 0.5311113890679573,
+      "grad_norm": 0.97265625,
+      "learning_rate": 2.6644855008195267e-05,
+      "loss": 1.1708,
+      "step": 3450
+    },
+    {
+      "epoch": 0.5326508423696035,
+      "grad_norm": 0.9296875,
+      "learning_rate": 2.6510752104605807e-05,
+      "loss": 1.1655,
+      "step": 3460
+    },
+    {
+      "epoch": 0.5341902956712498,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.6376605557184203e-05,
+      "loss": 1.1694,
+      "step": 3470
+    },
+    {
+      "epoch": 0.535729748972896,
+      "grad_norm": 0.9921875,
+      "learning_rate": 2.624241924126472e-05,
+      "loss": 1.1584,
+      "step": 3480
+    },
+    {
+      "epoch": 0.5372692022745422,
+      "grad_norm": 0.98046875,
+      "learning_rate": 2.610819703333049e-05,
+      "loss": 1.2091,
+      "step": 3490
+    },
+    {
+      "epoch": 0.5388086555761885,
+      "grad_norm": 0.97265625,
+      "learning_rate": 2.5973942810901524e-05,
+      "loss": 1.1921,
+      "step": 3500
+    },
+    {
+      "epoch": 0.5403481088778348,
+      "grad_norm": 1.0625,
+      "learning_rate": 2.5839660452422682e-05,
+      "loss": 1.189,
+      "step": 3510
+    },
+    {
+      "epoch": 0.541887562179481,
+      "grad_norm": 1.0390625,
+      "learning_rate": 2.570535383715165e-05,
+      "loss": 1.2385,
+      "step": 3520
+    },
+    {
+      "epoch": 0.5434270154811273,
+      "grad_norm": 1.0234375,
+      "learning_rate": 2.5571026845046876e-05,
+      "loss": 1.2173,
+      "step": 3530
+    },
+    {
+      "epoch": 0.5449664687827735,
+      "grad_norm": 1.078125,
+      "learning_rate": 2.5436683356655454e-05,
+      "loss": 1.17,
+      "step": 3540
+    },
+    {
+      "epoch": 0.5465059220844197,
+      "grad_norm": 1.03125,
+      "learning_rate": 2.5302327253001034e-05,
+      "loss": 1.1493,
+      "step": 3550
+    },
+    {
+      "epoch": 0.548045375386066,
+      "grad_norm": 0.984375,
+      "learning_rate": 2.516796241547172e-05,
+      "loss": 1.1651,
+      "step": 3560
+    },
+    {
+      "epoch": 0.5495848286877123,
+      "grad_norm": 1.015625,
+      "learning_rate": 2.503359272570791e-05,
+      "loss": 1.1835,
+      "step": 3570
+    },
+    {
+      "epoch": 0.5511242819893585,
+      "grad_norm": 1.078125,
+      "learning_rate": 2.4899222065490196e-05,
+      "loss": 1.2211,
+      "step": 3580
+    },
+    {
+      "epoch": 0.5526637352910048,
+      "grad_norm": 0.9140625,
+      "learning_rate": 2.47648543166272e-05,
+      "loss": 1.1825,
+      "step": 3590
+    },
+    {
+      "epoch": 0.554203188592651,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.4630493360843413e-05,
+      "loss": 1.192,
+      "step": 3600
+    },
+    {
+      "epoch": 0.5557426418942973,
+      "grad_norm": 0.97265625,
+      "learning_rate": 2.4496143079667133e-05,
+      "loss": 1.1657,
+      "step": 3610
+    },
+    {
+      "epoch": 0.5572820951959435,
+      "grad_norm": 1.046875,
+      "learning_rate": 2.436180735431823e-05,
+      "loss": 1.2041,
+      "step": 3620
+    },
+    {
+      "epoch": 0.5588215484975898,
+      "grad_norm": 1.046875,
+      "learning_rate": 2.4227490065596106e-05,
+      "loss": 1.1627,
+      "step": 3630
+    },
+    {
+      "epoch": 0.5603610017992361,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.409319509376753e-05,
+      "loss": 1.1779,
+      "step": 3640
+    },
+    {
+      "epoch": 0.5619004551008823,
+      "grad_norm": 0.95703125,
+      "learning_rate": 2.3958926318454593e-05,
+      "loss": 1.1837,
+      "step": 3650
+    },
+    {
+      "epoch": 0.5634399084025286,
+      "grad_norm": 0.953125,
+      "learning_rate": 2.3824687618522567e-05,
+      "loss": 1.174,
+      "step": 3660
+    },
+    {
+      "epoch": 0.5649793617041748,
+      "grad_norm": 0.9921875,
+      "learning_rate": 2.3690482871967883e-05,
+      "loss": 1.1829,
+      "step": 3670
+    },
+    {
+      "epoch": 0.566518815005821,
+      "grad_norm": 1.015625,
+      "learning_rate": 2.3556315955806134e-05,
+      "loss": 1.184,
+      "step": 3680
+    },
+    {
+      "epoch": 0.5680582683074673,
+      "grad_norm": 1.03125,
+      "learning_rate": 2.3422190745959996e-05,
+      "loss": 1.1252,
+      "step": 3690
+    },
+    {
+      "epoch": 0.5695977216091136,
+      "grad_norm": 1.046875,
+      "learning_rate": 2.3288111117147328e-05,
+      "loss": 1.1936,
+      "step": 3700
+    },
+    {
+      "epoch": 0.5711371749107598,
+      "grad_norm": 0.984375,
+      "learning_rate": 2.3154080942769186e-05,
+      "loss": 1.1825,
+      "step": 3710
+    },
+    {
+      "epoch": 0.5726766282124061,
+      "grad_norm": 1.03125,
+      "learning_rate": 2.3020104094797977e-05,
+      "loss": 1.1756,
+      "step": 3720
+    },
+    {
+      "epoch": 0.5742160815140523,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.2886184443665525e-05,
+      "loss": 1.1846,
+      "step": 3730
+    },
+    {
+      "epoch": 0.5757555348156985,
+      "grad_norm": 1.03125,
+      "learning_rate": 2.2752325858151357e-05,
+      "loss": 1.2064,
+      "step": 3740
+    },
+    {
+      "epoch": 0.5772949881173448,
+      "grad_norm": 0.91015625,
+      "learning_rate": 2.2618532205270853e-05,
+      "loss": 1.1361,
+      "step": 3750
+    },
+    {
+      "epoch": 0.5788344414189911,
+      "grad_norm": 1.0703125,
+      "learning_rate": 2.2484807350163564e-05,
+      "loss": 1.2123,
+      "step": 3760
+    },
+    {
+      "epoch": 0.5803738947206374,
+      "grad_norm": 1.0,
+      "learning_rate": 2.2351155155981585e-05,
+      "loss": 1.2297,
+      "step": 3770
+    },
+    {
+      "epoch": 0.5819133480222836,
+      "grad_norm": 1.0390625,
+      "learning_rate": 2.2217579483777885e-05,
+      "loss": 1.1668,
+      "step": 3780
+    },
+    {
+      "epoch": 0.5834528013239298,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.2084084192394847e-05,
+      "loss": 1.1967,
+      "step": 3790
+    },
+    {
+      "epoch": 0.5849922546255761,
+      "grad_norm": 1.046875,
+      "learning_rate": 2.19506731383527e-05,
+      "loss": 1.1691,
+      "step": 3800
+    },
+    {
+      "epoch": 0.5865317079272223,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.1817350175738204e-05,
+      "loss": 1.1515,
+      "step": 3810
+    },
+    {
+      "epoch": 0.5880711612288686,
+      "grad_norm": 0.94140625,
+      "learning_rate": 2.168411915609322e-05,
+      "loss": 1.1564,
+      "step": 3820
+    },
+    {
+      "epoch": 0.5896106145305149,
+      "grad_norm": 1.0703125,
+      "learning_rate": 2.1550983928303517e-05,
+      "loss": 1.1901,
+      "step": 3830
+    },
+    {
+      "epoch": 0.5911500678321611,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.141794833848753e-05,
+      "loss": 1.1631,
+      "step": 3840
+    },
+    {
+      "epoch": 0.5926895211338074,
+      "grad_norm": 0.94921875,
+      "learning_rate": 2.1285016229885295e-05,
+      "loss": 1.2279,
+      "step": 3850
+    },
+    {
+      "epoch": 0.5942289744354536,
+      "grad_norm": 1.1015625,
+      "learning_rate": 2.1152191442747377e-05,
+      "loss": 1.1402,
+      "step": 3860
+    },
+    {
+      "epoch": 0.5957684277370998,
+      "grad_norm": 0.94921875,
+      "learning_rate": 2.1019477814223944e-05,
+      "loss": 1.1584,
+      "step": 3870
+    },
+    {
+      "epoch": 0.5973078810387461,
+      "grad_norm": 1.03125,
+      "learning_rate": 2.088687917825396e-05,
+      "loss": 1.1524,
+      "step": 3880
+    },
+    {
+      "epoch": 0.5988473343403924,
+      "grad_norm": 0.984375,
+      "learning_rate": 2.0754399365454354e-05,
+      "loss": 1.2215,
+      "step": 3890
+    },
+    {
+      "epoch": 0.6003867876420386,
+      "grad_norm": 1.0546875,
+      "learning_rate": 2.0622042203009425e-05,
+      "loss": 1.171,
+      "step": 3900
+    },
+    {
+      "epoch": 0.6019262409436849,
+      "grad_norm": 1.0390625,
+      "learning_rate": 2.0489811514560218e-05,
+      "loss": 1.1716,
+      "step": 3910
+    },
+    {
+      "epoch": 0.6034656942453311,
+      "grad_norm": 1.1015625,
+      "learning_rate": 2.0357711120094134e-05,
+      "loss": 1.1459,
+      "step": 3920
+    },
+    {
+      "epoch": 0.6050051475469774,
+      "grad_norm": 1.0390625,
+      "learning_rate": 2.0225744835834486e-05,
+      "loss": 1.1889,
+      "step": 3930
+    },
+    {
+      "epoch": 0.6065446008486236,
+      "grad_norm": 0.99609375,
+      "learning_rate": 2.0093916474130353e-05,
+      "loss": 1.1453,
+      "step": 3940
+    },
+    {
+      "epoch": 0.6080840541502699,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.9962229843346364e-05,
+      "loss": 1.194,
+      "step": 3950
+    },
+    {
+      "epoch": 0.6096235074519162,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.98306887477527e-05,
+      "loss": 1.2177,
+      "step": 3960
+    },
+    {
+      "epoch": 0.6111629607535624,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.9699296987415253e-05,
+      "loss": 1.2156,
+      "step": 3970
+    },
+    {
+      "epoch": 0.6127024140552086,
+      "grad_norm": 0.9921875,
+      "learning_rate": 1.956805835808574e-05,
+      "loss": 1.1471,
+      "step": 3980
+    },
+    {
+      "epoch": 0.6142418673568549,
+      "grad_norm": 0.953125,
+      "learning_rate": 1.9436976651092144e-05,
+      "loss": 1.2147,
+      "step": 3990
+    },
+    {
+      "epoch": 0.6157813206585011,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.9306055653229127e-05,
+      "loss": 1.2103,
+      "step": 4000
+    },
+    {
+      "epoch": 0.6173207739601474,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.9175299146648674e-05,
+      "loss": 1.1906,
+      "step": 4010
+    },
+    {
+      "epoch": 0.6188602272617937,
+      "grad_norm": 0.92578125,
+      "learning_rate": 1.904471090875078e-05,
+      "loss": 1.1573,
+      "step": 4020
+    },
+    {
+      "epoch": 0.6203996805634399,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.8914294712074408e-05,
+      "loss": 1.2109,
+      "step": 4030
+    },
+    {
+      "epoch": 0.6219391338650861,
+      "grad_norm": 0.93359375,
+      "learning_rate": 1.8784054324188406e-05,
+      "loss": 1.2044,
+      "step": 4040
+    },
+    {
+      "epoch": 0.6234785871667324,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.8653993507582744e-05,
+      "loss": 1.2037,
+      "step": 4050
+    },
+    {
+      "epoch": 0.6250180404683787,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.8524116019559785e-05,
+      "loss": 1.1889,
+      "step": 4060
+    },
+    {
+      "epoch": 0.6265574937700249,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.8394425612125727e-05,
+      "loss": 1.1949,
+      "step": 4070
+    },
+    {
+      "epoch": 0.6280969470716712,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.8264926031882272e-05,
+      "loss": 1.2426,
+      "step": 4080
+    },
+    {
+      "epoch": 0.6296364003733175,
+      "grad_norm": 0.98828125,
+      "learning_rate": 1.813562101991831e-05,
+      "loss": 1.2207,
+      "step": 4090
+    },
+    {
+      "epoch": 0.6311758536749636,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.800651431170192e-05,
+      "loss": 1.1677,
+      "step": 4100
+    },
+    {
+      "epoch": 0.6327153069766099,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.78776096369724e-05,
+      "loss": 1.1935,
+      "step": 4110
+    },
+    {
+      "epoch": 0.6342547602782562,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.7748910719632537e-05,
+      "loss": 1.1756,
+      "step": 4120
+    },
+    {
+      "epoch": 0.6357942135799024,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.762042127764105e-05,
+      "loss": 1.2495,
+      "step": 4130
+    },
+    {
+      "epoch": 0.6373336668815487,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.749214502290516e-05,
+      "loss": 1.2229,
+      "step": 4140
+    },
+    {
+      "epoch": 0.638873120183195,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.7364085661173347e-05,
+      "loss": 1.216,
+      "step": 4150
+    },
+    {
+      "epoch": 0.6404125734848412,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.7236246891928314e-05,
+      "loss": 1.2154,
+      "step": 4160
+    },
+    {
+      "epoch": 0.6419520267864874,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.7108632408280138e-05,
+      "loss": 1.2008,
+      "step": 4170
+    },
+    {
+      "epoch": 0.6434914800881337,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.69812458968595e-05,
+      "loss": 1.1913,
+      "step": 4180
+    },
+    {
+      "epoch": 0.6450309333897799,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.6854091037711283e-05,
+      "loss": 1.2233,
+      "step": 4190
+    },
+    {
+      "epoch": 0.6465703866914262,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.672717150418818e-05,
+      "loss": 1.1729,
+      "step": 4200
+    },
+    {
+      "epoch": 0.6481098399930725,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.6600490962844632e-05,
+      "loss": 1.1807,
+      "step": 4210
+    },
+    {
+      "epoch": 0.6496492932947188,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.647405307333085e-05,
+      "loss": 1.2413,
+      "step": 4220
+    },
+    {
+      "epoch": 0.6511887465963649,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.634786148828715e-05,
+      "loss": 1.2019,
+      "step": 4230
+    },
+    {
+      "epoch": 0.6527281998980112,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.6221919853238385e-05,
+      "loss": 1.204,
+      "step": 4240
+    },
+    {
+      "epoch": 0.6542676531996575,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.6096231806488667e-05,
+      "loss": 1.2111,
+      "step": 4250
+    },
+    {
+      "epoch": 0.6558071065013037,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.5970800979016225e-05,
+      "loss": 1.2256,
+      "step": 4260
+    },
+    {
+      "epoch": 0.65734655980295,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.584563099436854e-05,
+      "loss": 1.2066,
+      "step": 4270
+    },
+    {
+      "epoch": 0.6588860131045963,
+      "grad_norm": 1.0,
+      "learning_rate": 1.5720725468557652e-05,
+      "loss": 1.2242,
+      "step": 4280
+    },
+    {
+      "epoch": 0.6604254664062424,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.5596088009955695e-05,
+      "loss": 1.1615,
+      "step": 4290
+    },
+    {
+      "epoch": 0.6619649197078887,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.5471722219190683e-05,
+      "loss": 1.2038,
+      "step": 4300
+    },
+    {
+      "epoch": 0.663504373009535,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.5347631689042417e-05,
+      "loss": 1.2119,
+      "step": 4310
+    },
+    {
+      "epoch": 0.6650438263111812,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.5223820004338817e-05,
+      "loss": 1.1574,
+      "step": 4320
+    },
+    {
+      "epoch": 0.6665832796128275,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.5100290741852224e-05,
+      "loss": 1.1853,
+      "step": 4330
+    },
+    {
+      "epoch": 0.6681227329144738,
+      "grad_norm": 0.9921875,
+      "learning_rate": 1.4977047470196188e-05,
+      "loss": 1.1764,
+      "step": 4340
+    },
+    {
+      "epoch": 0.66966218621612,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.4854093749722287e-05,
+      "loss": 1.1857,
+      "step": 4350
+    },
+    {
+      "epoch": 0.6712016395177662,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.4731433132417316e-05,
+      "loss": 1.1496,
+      "step": 4360
+    },
+    {
+      "epoch": 0.6727410928194125,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.4609069161800698e-05,
+      "loss": 1.1764,
+      "step": 4370
+    },
+    {
+      "epoch": 0.6742805461210588,
+      "grad_norm": 1.0,
+      "learning_rate": 1.4487005372822044e-05,
+      "loss": 1.1927,
+      "step": 4380
+    },
+    {
+      "epoch": 0.675819999422705,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.4365245291759117e-05,
+      "loss": 1.2108,
+      "step": 4390
+    },
+    {
+      "epoch": 0.6773594527243513,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.424379243611586e-05,
+      "loss": 1.2053,
+      "step": 4400
+    },
+    {
+      "epoch": 0.6788989060259976,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.4122650314520919e-05,
+      "loss": 1.1623,
+      "step": 4410
+    },
+    {
+      "epoch": 0.6804383593276437,
+      "grad_norm": 0.91796875,
+      "learning_rate": 1.4001822426626115e-05,
+      "loss": 1.1799,
+      "step": 4420
+    },
+    {
+      "epoch": 0.68197781262929,
+      "grad_norm": 1.125,
+      "learning_rate": 1.3881312263005519e-05,
+      "loss": 1.172,
+      "step": 4430
+    },
+    {
+      "epoch": 0.6835172659309363,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.3761123305054436e-05,
+      "loss": 1.1841,
+      "step": 4440
+    },
+    {
+      "epoch": 0.6850567192325825,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.3641259024889013e-05,
+      "loss": 1.1281,
+      "step": 4450
+    },
+    {
+      "epoch": 0.6865961725342288,
+      "grad_norm": 0.9921875,
+      "learning_rate": 1.3521722885245763e-05,
+      "loss": 1.196,
+      "step": 4460
+    },
+    {
+      "epoch": 0.6881356258358751,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.3402518339381653e-05,
+      "loss": 1.2035,
+      "step": 4470
+    },
+    {
+      "epoch": 0.6896750791375212,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.3283648830974282e-05,
+      "loss": 1.1809,
+      "step": 4480
+    },
+    {
+      "epoch": 0.6912145324391675,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.3165117794022432e-05,
+      "loss": 1.1252,
+      "step": 4490
+    },
+    {
+      "epoch": 0.6927539857408138,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.3046928652746832e-05,
+      "loss": 1.1516,
+      "step": 4500
+    },
+    {
+      "epoch": 0.69429343904246,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.2929084821491257e-05,
+      "loss": 1.2268,
+      "step": 4510
+    },
+    {
+      "epoch": 0.6958328923441063,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.2811589704623894e-05,
+      "loss": 1.1888,
+      "step": 4520
+    },
+    {
+      "epoch": 0.6973723456457526,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.2694446696438952e-05,
+      "loss": 1.2122,
+      "step": 4530
+    },
+    {
+      "epoch": 0.6989117989473989,
+      "grad_norm": 0.9921875,
+      "learning_rate": 1.25776591810587e-05,
+      "loss": 1.1822,
+      "step": 4540
+    },
+    {
+      "epoch": 0.700451252249045,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.2461230532335578e-05,
+      "loss": 1.194,
+      "step": 4550
+    },
+    {
+      "epoch": 0.7019907055506913,
+      "grad_norm": 0.96875,
+      "learning_rate": 1.2345164113754834e-05,
+      "loss": 1.1884,
+      "step": 4560
+    },
+    {
+      "epoch": 0.7035301588523376,
+      "grad_norm": 0.98828125,
+      "learning_rate": 1.2229463278337308e-05,
+      "loss": 1.1804,
+      "step": 4570
+    },
+    {
+      "epoch": 0.7050696121539838,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.2114131368542575e-05,
+      "loss": 1.1986,
+      "step": 4580
+    },
+    {
+      "epoch": 0.7066090654556301,
+      "grad_norm": 0.94140625,
+      "learning_rate": 1.1999171716172392e-05,
+      "loss": 1.1736,
+      "step": 4590
+    },
+    {
+      "epoch": 0.7081485187572764,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.1884587642274434e-05,
+      "loss": 1.2186,
+      "step": 4600
+    },
+    {
+      "epoch": 0.7096879720589225,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.177038245704637e-05,
+      "loss": 1.1758,
+      "step": 4610
+    },
+    {
+      "epoch": 0.7112274253605688,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.1656559459740218e-05,
+      "loss": 1.1773,
+      "step": 4620
+    },
+    {
+      "epoch": 0.7127668786622151,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.1543121938567045e-05,
+      "loss": 1.2007,
+      "step": 4630
+    },
+    {
+      "epoch": 0.7143063319638613,
+      "grad_norm": 0.984375,
+      "learning_rate": 1.1430073170601968e-05,
+      "loss": 1.1602,
+      "step": 4640
+    },
+    {
+      "epoch": 0.7158457852655076,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.1317416421689503e-05,
+      "loss": 1.2571,
+      "step": 4650
+    },
+    {
+      "epoch": 0.7173852385671539,
+      "grad_norm": 0.96875,
+      "learning_rate": 1.1205154946349161e-05,
+      "loss": 1.1542,
+      "step": 4660
+    },
+    {
+      "epoch": 0.7189246918688001,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.1093291987681511e-05,
+      "loss": 1.1819,
+      "step": 4670
+    },
+    {
+      "epoch": 0.7204641451704463,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.0981830777274427e-05,
+      "loss": 1.2414,
+      "step": 4680
+    },
+    {
+      "epoch": 0.7220035984720926,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.0870774535109764e-05,
+      "loss": 1.1888,
+      "step": 4690
+    },
+    {
+      "epoch": 0.7235430517737389,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.0760126469470322e-05,
+      "loss": 1.184,
+      "step": 4700
+    },
+    {
+      "epoch": 0.7250825050753851,
+      "grad_norm": 0.8984375,
+      "learning_rate": 1.0649889776847161e-05,
+      "loss": 1.181,
+      "step": 4710
+    },
+    {
+      "epoch": 0.7266219583770314,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.0540067641847273e-05,
+      "loss": 1.2008,
+      "step": 4720
+    },
+    {
+      "epoch": 0.7281614116786777,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.0430663237101571e-05,
+      "loss": 1.1711,
+      "step": 4730
+    },
+    {
+      "epoch": 0.7297008649803238,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.0321679723173231e-05,
+      "loss": 1.1349,
+      "step": 4740
+    },
+    {
+      "epoch": 0.7312403182819701,
+      "grad_norm": 0.984375,
+      "learning_rate": 1.0213120248466407e-05,
+      "loss": 1.1574,
+      "step": 4750
+    },
+    {
+      "epoch": 0.7327797715836164,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.0104987949135227e-05,
+      "loss": 1.148,
+      "step": 4760
+    },
+    {
+      "epoch": 0.7343192248852626,
+      "grad_norm": 0.95703125,
+      "learning_rate": 9.9972859489933e-06,
+      "loss": 1.1624,
+      "step": 4770
+    },
+    {
+      "epoch": 0.7358586781869089,
+      "grad_norm": 0.9765625,
+      "learning_rate": 9.890017359423325e-06,
+      "loss": 1.1756,
+      "step": 4780
+    },
+    {
+      "epoch": 0.7373981314885552,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.783185279287355e-06,
+      "loss": 1.1482,
+      "step": 4790
+    },
+    {
+      "epoch": 0.7389375847902013,
+      "grad_norm": 0.9609375,
+      "learning_rate": 9.676792794837142e-06,
+      "loss": 1.2482,
+      "step": 4800
+    },
+    {
+      "epoch": 0.7404770380918476,
+      "grad_norm": 0.953125,
+      "learning_rate": 9.57084297962508e-06,
+      "loss": 1.179,
+      "step": 4810
+    },
+    {
+      "epoch": 0.7420164913934939,
+      "grad_norm": 1.0,
+      "learning_rate": 9.465338894415365e-06,
+      "loss": 1.1759,
+      "step": 4820
+    },
+    {
+      "epoch": 0.7435559446951402,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.360283587095586e-06,
+      "loss": 1.1629,
+      "step": 4830
+    },
+    {
+      "epoch": 0.7450953979967864,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.25568009258867e-06,
+      "loss": 1.2064,
+      "step": 4840
+    },
+    {
+      "epoch": 0.7466348512984327,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.151531432765203e-06,
+      "loss": 1.1496,
+      "step": 4850
+    },
+    {
+      "epoch": 0.7481743046000789,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.047840616356154e-06,
+      "loss": 1.1868,
+      "step": 4860
+    },
+    {
+      "epoch": 0.7497137579017251,
+      "grad_norm": 1.0078125,
+      "learning_rate": 8.944610638865908e-06,
+      "loss": 1.2105,
+      "step": 4870
+    },
+    {
+      "epoch": 0.7512532112033714,
+      "grad_norm": 0.984375,
+      "learning_rate": 8.841844482485812e-06,
+      "loss": 1.169,
+      "step": 4880
+    },
+    {
+      "epoch": 0.7527926645050177,
+      "grad_norm": 1.0,
+      "learning_rate": 8.739545116007911e-06,
+      "loss": 1.1861,
+      "step": 4890
+    },
+    {
+      "epoch": 0.7543321178066639,
+      "grad_norm": 1.0703125,
+      "learning_rate": 8.637715494739313e-06,
+      "loss": 1.1611,
+      "step": 4900
+    },
+    {
+      "epoch": 0.7558715711083102,
+      "grad_norm": 0.9921875,
+      "learning_rate": 8.536358560416675e-06,
+      "loss": 1.1522,
+      "step": 4910
+    },
+    {
+      "epoch": 0.7574110244099564,
+      "grad_norm": 0.96875,
+      "learning_rate": 8.435477241121353e-06,
+      "loss": 1.1924,
+      "step": 4920
+    },
+    {
+      "epoch": 0.7589504777116026,
+      "grad_norm": 1.1015625,
+      "learning_rate": 8.335074451194705e-06,
+      "loss": 1.2074,
+      "step": 4930
+    },
+    {
+      "epoch": 0.7604899310132489,
+      "grad_norm": 1.0390625,
+      "learning_rate": 8.235153091153966e-06,
+      "loss": 1.1953,
+      "step": 4940
+    },
+    {
+      "epoch": 0.7620293843148952,
+      "grad_norm": 1.0703125,
+      "learning_rate": 8.13571604760843e-06,
+      "loss": 1.2009,
+      "step": 4950
+    },
+    {
+      "epoch": 0.7635688376165414,
+      "grad_norm": 1.015625,
+      "learning_rate": 8.03676619317606e-06,
+      "loss": 1.2033,
+      "step": 4960
+    },
+    {
+      "epoch": 0.7651082909181877,
+      "grad_norm": 1.046875,
+      "learning_rate": 7.938306386400521e-06,
+      "loss": 1.2157,
+      "step": 4970
+    },
+    {
+      "epoch": 0.766647744219834,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.840339471668542e-06,
+      "loss": 1.182,
+      "step": 4980
+    },
+    {
+      "epoch": 0.7681871975214802,
+      "grad_norm": 1.0625,
+      "learning_rate": 7.742868279127848e-06,
+      "loss": 1.1804,
+      "step": 4990
+    },
+    {
+      "epoch": 0.7697266508231264,
+      "grad_norm": 0.97265625,
+      "learning_rate": 7.645895624605275e-06,
+      "loss": 1.1501,
+      "step": 5000
+    },
+    {
+      "epoch": 0.7712661041247727,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.549424309525554e-06,
+      "loss": 1.1431,
+      "step": 5010
+    },
+    {
+      "epoch": 0.772805557426419,
+      "grad_norm": 1.0703125,
+      "learning_rate": 7.453457120830246e-06,
+      "loss": 1.2298,
+      "step": 5020
+    },
+    {
+      "epoch": 0.7743450107280652,
+      "grad_norm": 0.94140625,
+      "learning_rate": 7.357996830897368e-06,
+      "loss": 1.1941,
+      "step": 5030
+    },
+    {
+      "epoch": 0.7758844640297115,
+      "grad_norm": 1.03125,
+      "learning_rate": 7.263046197461173e-06,
+      "loss": 1.2013,
+      "step": 5040
+    },
+    {
+      "epoch": 0.7774239173313577,
+      "grad_norm": 1.046875,
+      "learning_rate": 7.1686079635326e-06,
+      "loss": 1.2135,
+      "step": 5050
+    },
+    {
+      "epoch": 0.7789633706330039,
+      "grad_norm": 0.94921875,
+      "learning_rate": 7.074684857319927e-06,
+      "loss": 1.1943,
+      "step": 5060
+    },
+    {
+      "epoch": 0.7805028239346502,
+      "grad_norm": 0.9296875,
+      "learning_rate": 6.981279592150031e-06,
+      "loss": 1.132,
+      "step": 5070
+    },
+    {
+      "epoch": 0.7820422772362965,
+      "grad_norm": 0.94140625,
+      "learning_rate": 6.888394866389977e-06,
+      "loss": 1.1556,
+      "step": 5080
+    },
+    {
+      "epoch": 0.7835817305379427,
+      "grad_norm": 1.0703125,
+      "learning_rate": 6.796033363369056e-06,
+      "loss": 1.1626,
+      "step": 5090
+    },
+    {
+      "epoch": 0.785121183839589,
+      "grad_norm": 0.97265625,
+      "learning_rate": 6.70419775130128e-06,
+      "loss": 1.2153,
+      "step": 5100
+    },
+    {
+      "epoch": 0.7866606371412352,
+      "grad_norm": 1.015625,
+      "learning_rate": 6.612890683208297e-06,
+      "loss": 1.1796,
+      "step": 5110
+    },
+    {
+      "epoch": 0.7882000904428814,
+      "grad_norm": 0.94140625,
+      "learning_rate": 6.522114796842757e-06,
+      "loss": 1.2185,
+      "step": 5120
+    },
+    {
+      "epoch": 0.7897395437445277,
+      "grad_norm": 0.94921875,
+      "learning_rate": 6.431872714612072e-06,
+      "loss": 1.1647,
+      "step": 5130
+    },
+    {
+      "epoch": 0.791278997046174,
+      "grad_norm": 1.0078125,
+      "learning_rate": 6.342167043502731e-06,
+      "loss": 1.1601,
+      "step": 5140
+    },
+    {
+      "epoch": 0.7928184503478203,
+      "grad_norm": 1.0078125,
+      "learning_rate": 6.253000375004903e-06,
+      "loss": 1.1668,
+      "step": 5150
+    },
+    {
+      "epoch": 0.7943579036494665,
+      "grad_norm": 0.95703125,
+      "learning_rate": 6.164375285037633e-06,
+      "loss": 1.2012,
+      "step": 5160
+    },
+    {
+      "epoch": 0.7958973569511127,
+      "grad_norm": 1.0390625,
+      "learning_rate": 6.0762943338744025e-06,
+      "loss": 1.2203,
+      "step": 5170
+    },
+    {
+      "epoch": 0.797436810252759,
+      "grad_norm": 0.98046875,
+      "learning_rate": 5.988760066069171e-06,
+      "loss": 1.2116,
+      "step": 5180
+    },
+    {
+      "epoch": 0.7989762635544052,
+      "grad_norm": 0.9609375,
+      "learning_rate": 5.901775010382868e-06,
+      "loss": 1.1751,
+      "step": 5190
+    },
+    {
+      "epoch": 0.8005157168560515,
+      "grad_norm": 0.98046875,
+      "learning_rate": 5.815341679710326e-06,
+      "loss": 1.152,
+      "step": 5200
+    },
+    {
+      "epoch": 0.8020551701576978,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.7294625710077154e-06,
+      "loss": 1.1982,
+      "step": 5210
+    },
+    {
+      "epoch": 0.803594623459344,
+      "grad_norm": 1.0,
+      "learning_rate": 5.644140165220377e-06,
+      "loss": 1.1842,
+      "step": 5220
+    },
+    {
+      "epoch": 0.8051340767609902,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.559376927211177e-06,
+      "loss": 1.0996,
+      "step": 5230
+    },
+    {
+      "epoch": 0.8066735300626365,
+      "grad_norm": 0.9140625,
+      "learning_rate": 5.475175305689281e-06,
+      "loss": 1.1882,
+      "step": 5240
+    },
+    {
+      "epoch": 0.8082129833642827,
+      "grad_norm": 1.0,
+      "learning_rate": 5.391537733139435e-06,
+      "loss": 1.1556,
+      "step": 5250
+    },
+    {
+      "epoch": 0.809752436665929,
+      "grad_norm": 0.98046875,
+      "learning_rate": 5.30846662575166e-06,
+      "loss": 1.1498,
+      "step": 5260
+    },
+    {
+      "epoch": 0.8112918899675753,
+      "grad_norm": 0.984375,
+      "learning_rate": 5.2259643833514896e-06,
+      "loss": 1.1543,
+      "step": 5270
+    },
+    {
+      "epoch": 0.8128313432692215,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.144033389330616e-06,
+      "loss": 1.1875,
+      "step": 5280
+    },
+    {
+      "epoch": 0.8143707965708677,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.062676010578049e-06,
+      "loss": 1.1899,
+      "step": 5290
+    },
+    {
+      "epoch": 0.815910249872514,
+      "grad_norm": 0.9453125,
+      "learning_rate": 4.9818945974117346e-06,
+      "loss": 1.1771,
+      "step": 5300
+    },
+    {
+      "epoch": 0.8174497031741603,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.901691483510659e-06,
+      "loss": 1.2201,
+      "step": 5310
+    },
+    {
+      "epoch": 0.8189891564758065,
+      "grad_norm": 0.8984375,
+      "learning_rate": 4.822068985847422e-06,
+      "loss": 1.2144,
+      "step": 5320
+    },
+    {
+      "epoch": 0.8205286097774528,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.743029404621321e-06,
+      "loss": 1.2004,
+      "step": 5330
+    },
+    {
+      "epoch": 0.8220680630790991,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.664575023191886e-06,
+      "loss": 1.2271,
+      "step": 5340
+    },
+    {
+      "epoch": 0.8236075163807453,
+      "grad_norm": 0.9375,
+      "learning_rate": 4.586708108012927e-06,
+      "loss": 1.1739,
+      "step": 5350
+    },
+    {
+      "epoch": 0.8251469696823915,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.509430908567025e-06,
+      "loss": 1.1795,
+      "step": 5360
+    },
+    {
+      "epoch": 0.8266864229840378,
+      "grad_norm": 0.9296875,
+      "learning_rate": 4.4327456573006245e-06,
+      "loss": 1.1966,
+      "step": 5370
+    },
+    {
+      "epoch": 0.828225876285684,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.35665456955944e-06,
+      "loss": 1.2244,
+      "step": 5380
+    },
+    {
+      "epoch": 0.8297653295873303,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.28115984352454e-06,
+      "loss": 1.1845,
+      "step": 5390
+    },
+    {
+      "epoch": 0.8313047828889766,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.206263660148799e-06,
+      "loss": 1.2527,
+      "step": 5400
+    },
+    {
+      "epoch": 0.8328442361906228,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.131968183093912e-06,
+      "loss": 1.1524,
+      "step": 5410
+    },
+    {
+      "epoch": 0.834383689492269,
+      "grad_norm": 0.94921875,
+      "learning_rate": 4.0582755586678815e-06,
+      "loss": 1.2096,
+      "step": 5420
+    },
+    {
+      "epoch": 0.8359231427939153,
+      "grad_norm": 1.0,
+      "learning_rate": 3.985187915763011e-06,
+      "loss": 1.1735,
+      "step": 5430
+    },
+    {
+      "epoch": 0.8374625960955616,
+      "grad_norm": 1.0390625,
+      "learning_rate": 3.912707365794408e-06,
+      "loss": 1.1806,
+      "step": 5440
+    },
+    {
+      "epoch": 0.8390020493972078,
+      "grad_norm": 1.0703125,
+      "learning_rate": 3.8408360026389895e-06,
+      "loss": 1.1441,
+      "step": 5450
+    },
+    {
+      "epoch": 0.8405415026988541,
+      "grad_norm": 0.96484375,
+      "learning_rate": 3.7695759025749926e-06,
+      "loss": 1.1526,
+      "step": 5460
+    },
+    {
+      "epoch": 0.8420809560005004,
+      "grad_norm": 0.9140625,
+      "learning_rate": 3.6989291242219697e-06,
+      "loss": 1.2,
+      "step": 5470
+    },
+    {
+      "epoch": 0.8436204093021465,
+      "grad_norm": 0.98828125,
+      "learning_rate": 3.6288977084813767e-06,
+      "loss": 1.1826,
+      "step": 5480
+    },
+    {
+      "epoch": 0.8451598626037928,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.5594836784775295e-06,
+      "loss": 1.1116,
+      "step": 5490
+    },
+    {
+      "epoch": 0.8466993159054391,
+      "grad_norm": 0.98046875,
+      "learning_rate": 3.490689039499251e-06,
+      "loss": 1.1989,
+      "step": 5500
+    },
+    {
+      "epoch": 0.8482387692070853,
+      "grad_norm": 1.0390625,
+      "learning_rate": 3.4225157789418482e-06,
+      "loss": 1.2241,
+      "step": 5510
+    },
+    {
+      "epoch": 0.8497782225087316,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.3549658662497846e-06,
+      "loss": 1.2107,
+      "step": 5520
+    },
+    {
+      "epoch": 0.8513176758103779,
+      "grad_norm": 1.046875,
+      "learning_rate": 3.2880412528597167e-06,
+      "loss": 1.1589,
+      "step": 5530
+    },
+    {
+      "epoch": 0.852857129112024,
+      "grad_norm": 1.0390625,
+      "learning_rate": 3.221743872144159e-06,
+      "loss": 1.1815,
+      "step": 5540
+    },
+    {
+      "epoch": 0.8543965824136703,
+      "grad_norm": 1.0,
+      "learning_rate": 3.1560756393556183e-06,
+      "loss": 1.1994,
+      "step": 5550
+    },
+    {
+      "epoch": 0.8559360357153166,
+      "grad_norm": 1.0,
+      "learning_rate": 3.091038451571268e-06,
+      "loss": 1.2053,
+      "step": 5560
+    },
+    {
+      "epoch": 0.8574754890169628,
+      "grad_norm": 1.0,
+      "learning_rate": 3.0266341876381444e-06,
+      "loss": 1.1789,
+      "step": 5570
+    },
+    {
+      "epoch": 0.8590149423186091,
+      "grad_norm": 1.015625,
+      "learning_rate": 2.9628647081188425e-06,
+      "loss": 1.2091,
+      "step": 5580
+    },
+    {
+      "epoch": 0.8605543956202554,
+      "grad_norm": 0.98828125,
+      "learning_rate": 2.8997318552378333e-06,
+      "loss": 1.1924,
+      "step": 5590
+    },
+    {
+      "epoch": 0.8620938489219017,
+      "grad_norm": 0.98046875,
+      "learning_rate": 2.8372374528281535e-06,
+      "loss": 1.1808,
+      "step": 5600
+    },
+    {
+      "epoch": 0.8636333022235478,
+      "grad_norm": 0.95703125,
+      "learning_rate": 2.7753833062788114e-06,
+      "loss": 1.1326,
+      "step": 5610
+    },
+    {
+      "epoch": 0.8651727555251941,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.714171202482538e-06,
+      "loss": 1.1976,
+      "step": 5620
+    },
+    {
+      "epoch": 0.8667122088268404,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.6536029097842587e-06,
+      "loss": 1.1789,
+      "step": 5630
+    },
+    {
+      "epoch": 0.8682516621284866,
+      "grad_norm": 1.0390625,
+      "learning_rate": 2.593680177929919e-06,
+      "loss": 1.2061,
+      "step": 5640
+    },
+    {
+      "epoch": 0.8697911154301329,
+      "grad_norm": 0.96484375,
+      "learning_rate": 2.5344047380160164e-06,
+      "loss": 1.1649,
+      "step": 5650
+    },
+    {
+      "epoch": 0.8713305687317792,
+      "grad_norm": 0.99609375,
+      "learning_rate": 2.475778302439524e-06,
+      "loss": 1.1574,
+      "step": 5660
+    },
+    {
+      "epoch": 0.8728700220334253,
+      "grad_norm": 0.94921875,
+      "learning_rate": 2.4178025648484678e-06,
+      "loss": 1.1924,
+      "step": 5670
+    },
+    {
+      "epoch": 0.8744094753350716,
+      "grad_norm": 1.03125,
+      "learning_rate": 2.360479200092977e-06,
+      "loss": 1.1812,
+      "step": 5680
+    },
+    {
+      "epoch": 0.8759489286367179,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.303809864176909e-06,
+      "loss": 1.1606,
+      "step": 5690
+    },
+    {
+      "epoch": 0.8774883819383641,
+      "grad_norm": 0.94921875,
+      "learning_rate": 2.2477961942100085e-06,
+      "loss": 1.1486,
+      "step": 5700
+    },
+    {
+      "epoch": 0.8790278352400104,
+      "grad_norm": 1.078125,
+      "learning_rate": 2.192439808360591e-06,
+      "loss": 1.1824,
+      "step": 5710
+    },
+    {
+      "epoch": 0.8805672885416567,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.1377423058088457e-06,
+      "loss": 1.2095,
+      "step": 5720
+    },
+    {
+      "epoch": 0.8821067418433028,
+      "grad_norm": 1.0234375,
+      "learning_rate": 2.0837052667005753e-06,
+      "loss": 1.155,
+      "step": 5730
+    },
+    {
+      "epoch": 0.8836461951449491,
+      "grad_norm": 1.0390625,
+      "learning_rate": 2.0303302521016087e-06,
+      "loss": 1.1822,
+      "step": 5740
+    },
+    {
+      "epoch": 0.8851856484465954,
+      "grad_norm": 1.0,
+      "learning_rate": 1.9776188039526495e-06,
+      "loss": 1.162,
+      "step": 5750
+    },
+    {
+      "epoch": 0.8867251017482417,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.9255724450247674e-06,
+      "loss": 1.2288,
+      "step": 5760
+    },
+    {
+      "epoch": 0.8882645550498879,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.8741926788753917e-06,
+      "loss": 1.2347,
+      "step": 5770
+    },
+    {
+      "epoch": 0.8898040083515342,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.823480989804885e-06,
+      "loss": 1.1959,
+      "step": 5780
+    },
+    {
+      "epoch": 0.8913434616531805,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.773438842813649e-06,
+      "loss": 1.2208,
+      "step": 5790
+    },
+    {
+      "epoch": 0.8928829149548266,
+      "grad_norm": 0.9140625,
+      "learning_rate": 1.7240676835598156e-06,
+      "loss": 1.1785,
+      "step": 5800
+    },
+    {
+      "epoch": 0.8944223682564729,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.6753689383174786e-06,
+      "loss": 1.1836,
+      "step": 5810
+    },
+    {
+      "epoch": 0.8959618215581192,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.6273440139354862e-06,
+      "loss": 1.1512,
+      "step": 5820
+    },
+    {
+      "epoch": 0.8975012748597654,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.579994297796808e-06,
+      "loss": 1.1279,
+      "step": 5830
+    },
+    {
+      "epoch": 0.8990407281614117,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.5333211577784474e-06,
+      "loss": 1.1724,
+      "step": 5840
+    },
+    {
+      "epoch": 0.900580181463058,
+      "grad_norm": 0.89453125,
+      "learning_rate": 1.487325942211934e-06,
+      "loss": 1.1603,
+      "step": 5850
+    },
+    {
+      "epoch": 0.9021196347647041,
+      "grad_norm": 1.0,
+      "learning_rate": 1.4420099798443553e-06,
+      "loss": 1.168,
+      "step": 5860
+    },
+    {
+      "epoch": 0.9036590880663504,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.397374579799987e-06,
+      "loss": 1.1559,
+      "step": 5870
+    },
+    {
+      "epoch": 0.9051985413679967,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.3534210315424706e-06,
+      "loss": 1.166,
+      "step": 5880
+    },
+    {
+      "epoch": 0.9067379946696429,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3101506048375594e-06,
+      "loss": 1.1659,
+      "step": 5890
+    },
+    {
+      "epoch": 0.9082774479712892,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.267564549716435e-06,
+      "loss": 1.1608,
+      "step": 5900
+    },
+    {
+      "epoch": 0.9098169012729355,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.2256640964395993e-06,
+      "loss": 1.1687,
+      "step": 5910
+    },
+    {
+      "epoch": 0.9113563545745818,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.1844504554613278e-06,
+      "loss": 1.2032,
+      "step": 5920
+    },
+    {
+      "epoch": 0.9128958078762279,
+      "grad_norm": 0.98828125,
+      "learning_rate": 1.143924817394712e-06,
+      "loss": 1.1827,
+      "step": 5930
+    },
+    {
+      "epoch": 0.9144352611778742,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.104088352977256e-06,
+      "loss": 1.1681,
+      "step": 5940
+    },
+    {
+      "epoch": 0.9159747144795205,
+      "grad_norm": 0.96875,
+      "learning_rate": 1.064942213037054e-06,
+      "loss": 1.1983,
+      "step": 5950
+    },
+    {
+      "epoch": 0.9175141677811667,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.0264875284595381e-06,
+      "loss": 1.2013,
+      "step": 5960
+    },
+    {
+      "epoch": 0.919053621082813,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.88725410154842e-07,
+      "loss": 1.1782,
+      "step": 5970
+    },
+    {
+      "epoch": 0.9205930743844593,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.516569490256533e-07,
+      "loss": 1.2126,
+      "step": 5980
+    },
+    {
+      "epoch": 0.9221325276861054,
+      "grad_norm": 0.9453125,
+      "learning_rate": 9.152832159357527e-07,
+      "loss": 1.1777,
+      "step": 5990
+    },
+    {
+      "epoch": 0.9236719809877517,
+      "grad_norm": 0.9921875,
+      "learning_rate": 8.796052616790368e-07,
+      "loss": 1.1426,
+      "step": 6000
+    },
+    {
+      "epoch": 0.925211434289398,
+      "grad_norm": 0.953125,
+      "learning_rate": 8.446241169491914e-07,
+      "loss": 1.1745,
+      "step": 6010
+    },
+    {
+      "epoch": 0.9267508875910442,
+      "grad_norm": 0.9453125,
+      "learning_rate": 8.103407923098938e-07,
+      "loss": 1.1548,
+      "step": 6020
+    },
+    {
+      "epoch": 0.9282903408926905,
+      "grad_norm": 1.125,
+      "learning_rate": 7.767562781656356e-07,
+      "loss": 1.2066,
+      "step": 6030
+    },
+    {
+      "epoch": 0.9298297941943368,
+      "grad_norm": 0.9375,
+      "learning_rate": 7.438715447331018e-07,
+      "loss": 1.1717,
+      "step": 6040
+    },
+    {
+      "epoch": 0.931369247495983,
+      "grad_norm": 1.015625,
+      "learning_rate": 7.116875420131425e-07,
+      "loss": 1.1907,
+      "step": 6050
+    },
+    {
+      "epoch": 0.9329087007976292,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.802051997633341e-07,
+      "loss": 1.1998,
+      "step": 6060
+    },
+    {
+      "epoch": 0.9344481540992755,
+      "grad_norm": 0.9765625,
+      "learning_rate": 6.494254274711065e-07,
+      "loss": 1.2226,
+      "step": 6070
+    },
+    {
+      "epoch": 0.9359876074009218,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.193491143274999e-07,
+      "loss": 1.1811,
+      "step": 6080
+    },
+    {
+      "epoch": 0.937527060702568,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.899771292014328e-07,
+      "loss": 1.1547,
+      "step": 6090
+    },
+    {
+      "epoch": 0.9390665140042143,
+      "grad_norm": 0.94140625,
+      "learning_rate": 5.613103206146386e-07,
+      "loss": 1.1566,
+      "step": 6100
+    },
+    {
+      "epoch": 0.9406059673058605,
+      "grad_norm": 0.984375,
+      "learning_rate": 5.333495167171353e-07,
+      "loss": 1.1649,
+      "step": 6110
+    },
+    {
+      "epoch": 0.9421454206075067,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.060955252633004e-07,
+      "loss": 1.1907,
+      "step": 6120
+    },
+    {
+      "epoch": 0.943684873909153,
+      "grad_norm": 0.96875,
+      "learning_rate": 4.795491335885471e-07,
+      "loss": 1.164,
+      "step": 6130
+    },
+    {
+      "epoch": 0.9452243272107993,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.537111085865714e-07,
+      "loss": 1.1705,
+      "step": 6140
+    },
+    {
+      "epoch": 0.9467637805124455,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.2858219668719393e-07,
+      "loss": 1.1868,
+      "step": 6150
+    },
+    {
+      "epoch": 0.9483032338140918,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.0416312383480257e-07,
+      "loss": 1.1958,
+      "step": 6160
+    },
+    {
+      "epoch": 0.949842687115738,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.8045459546738604e-07,
+      "loss": 1.1759,
+      "step": 6170
+    },
+    {
+      "epoch": 0.9513821404173842,
+      "grad_norm": 0.96875,
+      "learning_rate": 3.5745729649613034e-07,
+      "loss": 1.1496,
+      "step": 6180
+    },
+    {
+      "epoch": 0.9529215937190305,
+      "grad_norm": 1.0703125,
+      "learning_rate": 3.3517189128567105e-07,
+      "loss": 1.1768,
+      "step": 6190
+    },
+    {
+      "epoch": 0.9544610470206768,
+      "grad_norm": 1.0546875,
+      "learning_rate": 3.1359902363486126e-07,
+      "loss": 1.2276,
+      "step": 6200
+    },
+    {
+      "epoch": 0.9560005003223231,
+      "grad_norm": 1.0234375,
+      "learning_rate": 2.92739316758206e-07,
+      "loss": 1.194,
+      "step": 6210
+    },
+    {
+      "epoch": 0.9575399536239693,
+      "grad_norm": 1.078125,
+      "learning_rate": 2.7259337326783207e-07,
+      "loss": 1.194,
+      "step": 6220
+    },
+    {
+      "epoch": 0.9590794069256156,
+      "grad_norm": 0.93359375,
+      "learning_rate": 2.5316177515609664e-07,
+      "loss": 1.1486,
+      "step": 6230
+    },
+    {
+      "epoch": 0.9606188602272618,
+      "grad_norm": 1.0390625,
+      "learning_rate": 2.3444508377876706e-07,
+      "loss": 1.2341,
+      "step": 6240
+    },
+    {
+      "epoch": 0.962158313528908,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.1644383983880357e-07,
+      "loss": 1.1635,
+      "step": 6250
+    },
+    {
+      "epoch": 0.9636977668305543,
+      "grad_norm": 0.9921875,
+      "learning_rate": 1.991585633707438e-07,
+      "loss": 1.1799,
+      "step": 6260
+    },
+    {
+      "epoch": 0.9652372201322006,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.8258975372567044e-07,
+      "loss": 1.1879,
+      "step": 6270
+    },
+    {
+      "epoch": 0.9667766734338468,
+      "grad_norm": 0.96875,
+      "learning_rate": 1.667378895567978e-07,
+      "loss": 1.1933,
+      "step": 6280
+    },
+    {
+      "epoch": 0.968316126735493,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.5160342880563562e-07,
+      "loss": 1.1989,
+      "step": 6290
+    },
+    {
+      "epoch": 0.9698555800371393,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.3718680868876065e-07,
+      "loss": 1.1976,
+      "step": 6300
+    },
+    {
+      "epoch": 0.9713950333387855,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.2348844568518536e-07,
+      "loss": 1.1379,
+      "step": 6310
+    },
+    {
+      "epoch": 0.9729344866404318,
+      "grad_norm": 0.94140625,
+      "learning_rate": 1.1050873552433394e-07,
+      "loss": 1.2124,
+      "step": 6320
+    },
+    {
+      "epoch": 0.9744739399420781,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.824805317459884e-08,
+      "loss": 1.1379,
+      "step": 6330
+    },
+    {
+      "epoch": 0.9760133932437243,
+      "grad_norm": 0.94140625,
+      "learning_rate": 8.670675283252162e-08,
+      "loss": 1.1701,
+      "step": 6340
+    },
+    {
+      "epoch": 0.9775528465453706,
+      "grad_norm": 0.98046875,
+      "learning_rate": 7.588516791253997e-08,
+      "loss": 1.1907,
+      "step": 6350
+    },
+    {
+      "epoch": 0.9790922998470168,
+      "grad_norm": 1.0,
+      "learning_rate": 6.578361103738162e-08,
+      "loss": 1.194,
+      "step": 6360
+    },
+    {
+      "epoch": 0.9806317531486631,
+      "grad_norm": 1.0,
+      "learning_rate": 5.640237402901316e-08,
+      "loss": 1.1512,
+      "step": 6370
+    },
+    {
+      "epoch": 0.9821712064503093,
+      "grad_norm": 0.9375,
+      "learning_rate": 4.774172790021625e-08,
+      "loss": 1.169,
+      "step": 6380
+    },
+    {
+      "epoch": 0.9837106597519556,
+      "grad_norm": 1.046875,
+      "learning_rate": 3.9801922846766095e-08,
+      "loss": 1.1899,
+      "step": 6390
+    },
+    {
+      "epoch": 0.9852501130536019,
+      "grad_norm": 1.078125,
+      "learning_rate": 3.258318824018447e-08,
+      "loss": 1.2061,
+      "step": 6400
+    },
+    {
+      "epoch": 0.9867895663552481,
+      "grad_norm": 0.98046875,
+      "learning_rate": 2.6085732621133896e-08,
+      "loss": 1.2191,
+      "step": 6410
+    },
+    {
+      "epoch": 0.9883290196568943,
+      "grad_norm": 0.89453125,
+      "learning_rate": 2.030974369338634e-08,
+      "loss": 1.1731,
+      "step": 6420
+    },
+    {
+      "epoch": 0.9898684729585406,
+      "grad_norm": 1.0,
+      "learning_rate": 1.5255388318388685e-08,
+      "loss": 1.1684,
+      "step": 6430
+    },
+    {
+      "epoch": 0.9914079262601868,
+      "grad_norm": 0.93359375,
+      "learning_rate": 1.0922812510463786e-08,
+      "loss": 1.2051,
+      "step": 6440
+    },
+    {
+      "epoch": 0.9929473795618331,
+      "grad_norm": 0.99609375,
+      "learning_rate": 7.3121414325721945e-09,
+      "loss": 1.1839,
+      "step": 6450
+    },
+    {
+      "epoch": 0.9944868328634794,
+      "grad_norm": 0.95703125,
+      "learning_rate": 4.423479392709484e-09,
+      "loss": 1.2251,
+      "step": 6460
+    },
+    {
+      "epoch": 0.9960262861651256,
+      "grad_norm": 1.015625,
+      "learning_rate": 2.256909840886445e-09,
+      "loss": 1.1751,
+      "step": 6470
+    },
+    {
+      "epoch": 0.9975657394667719,
+      "grad_norm": 1.1328125,
+      "learning_rate": 8.12495366725452e-10,
+      "loss": 1.1922,
+      "step": 6480
+    },
+    {
+      "epoch": 0.9991051927684181,
+      "grad_norm": 0.98046875,
+      "learning_rate": 9.027769763969574e-11,
+      "loss": 1.1806,
+      "step": 6490
+    },
+    {
+      "epoch": 0.9998749194192412,
+      "eval_loss": 1.200334906578064,
+      "eval_runtime": 1389.3681,
+      "eval_samples_per_second": 16.633,
+      "eval_steps_per_second": 8.317,
+      "step": 6495
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 6495,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.3551089815460864e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}