diff --git "a/checkpoint-10000/trainer_state.json" "b/checkpoint-10000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-10000/trainer_state.json"
@@ -0,0 +1,7034 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.26753705388196264,
+  "eval_steps": 500,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0002675370538819627,
+      "grad_norm": 20.12447738647461,
+      "learning_rate": 1.8e-07,
+      "loss": 1.138,
+      "step": 10
+    },
+    {
+      "epoch": 0.0005350741077639254,
+      "grad_norm": 50.818824768066406,
+      "learning_rate": 3.8e-07,
+      "loss": 0.8476,
+      "step": 20
+    },
+    {
+      "epoch": 0.0008026111616458879,
+      "grad_norm": 12.44139289855957,
+      "learning_rate": 5.8e-07,
+      "loss": 0.8351,
+      "step": 30
+    },
+    {
+      "epoch": 0.0010701482155278507,
+      "grad_norm": 23.251235961914062,
+      "learning_rate": 7.8e-07,
+      "loss": 0.7812,
+      "step": 40
+    },
+    {
+      "epoch": 0.0013376852694098132,
+      "grad_norm": 24.371976852416992,
+      "learning_rate": 9.8e-07,
+      "loss": 0.7733,
+      "step": 50
+    },
+    {
+      "epoch": 0.0016052223232917758,
+      "grad_norm": 21.278976440429688,
+      "learning_rate": 1.18e-06,
+      "loss": 0.7244,
+      "step": 60
+    },
+    {
+      "epoch": 0.0018727593771737385,
+      "grad_norm": 12.708613395690918,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.69,
+      "step": 70
+    },
+    {
+      "epoch": 0.0021402964310557014,
+      "grad_norm": 8.386579513549805,
+      "learning_rate": 1.5800000000000003e-06,
+      "loss": 0.5016,
+      "step": 80
+    },
+    {
+      "epoch": 0.0024078334849376637,
+      "grad_norm": 4.984817981719971,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.3073,
+      "step": 90
+    },
+    {
+      "epoch": 0.0026753705388196263,
+      "grad_norm": 19.579139709472656,
+      "learning_rate": 1.98e-06,
+      "loss": 0.4469,
+      "step": 100
+    },
+    {
+      "epoch": 0.002942907592701589,
+      "grad_norm": 6.719464302062988,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.227,
+      "step": 110
+    },
+    {
+      "epoch": 0.0032104446465835517,
+      "grad_norm": 10.55249309539795,
+      "learning_rate": 2.38e-06,
+      "loss": 0.8577,
+      "step": 120
+    },
+    {
+      "epoch": 0.0034779817004655144,
+      "grad_norm": 6.482703685760498,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.3466,
+      "step": 130
+    },
+    {
+      "epoch": 0.003745518754347477,
+      "grad_norm": 11.616539001464844,
+      "learning_rate": 2.78e-06,
+      "loss": 0.395,
+      "step": 140
+    },
+    {
+      "epoch": 0.00401305580822944,
+      "grad_norm": 13.077627182006836,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.2734,
+      "step": 150
+    },
+    {
+      "epoch": 0.004280592862111403,
+      "grad_norm": 6.381225109100342,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.1903,
+      "step": 160
+    },
+    {
+      "epoch": 0.004548129915993365,
+      "grad_norm": 11.506539344787598,
+      "learning_rate": 3.38e-06,
+      "loss": 0.2305,
+      "step": 170
+    },
+    {
+      "epoch": 0.004815666969875327,
+      "grad_norm": 3.0193920135498047,
+      "learning_rate": 3.58e-06,
+      "loss": 0.2195,
+      "step": 180
+    },
+    {
+      "epoch": 0.00508320402375729,
+      "grad_norm": 4.8953857421875,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.303,
+      "step": 190
+    },
+    {
+      "epoch": 0.005350741077639253,
+      "grad_norm": 2.152596950531006,
+      "learning_rate": 3.98e-06,
+      "loss": 0.321,
+      "step": 200
+    },
+    {
+      "epoch": 0.005618278131521216,
+      "grad_norm": 6.5669708251953125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.2662,
+      "step": 210
+    },
+    {
+      "epoch": 0.005885815185403178,
+      "grad_norm": 9.632110595703125,
+      "learning_rate": 4.38e-06,
+      "loss": 0.2083,
+      "step": 220
+    },
+    {
+      "epoch": 0.006153352239285141,
+      "grad_norm": 6.781233787536621,
+      "learning_rate": 4.58e-06,
+      "loss": 0.2039,
+      "step": 230
+    },
+    {
+      "epoch": 0.006420889293167103,
+      "grad_norm": 13.589167594909668,
+      "learning_rate": 4.780000000000001e-06,
+      "loss": 0.2832,
+      "step": 240
+    },
+    {
+      "epoch": 0.0066884263470490665,
+      "grad_norm": 8.22282600402832,
+      "learning_rate": 4.98e-06,
+      "loss": 0.2493,
+      "step": 250
+    },
+    {
+      "epoch": 0.006955963400931029,
+      "grad_norm": 3.4578022956848145,
+      "learning_rate": 5.18e-06,
+      "loss": 0.1388,
+      "step": 260
+    },
+    {
+      "epoch": 0.007223500454812992,
+      "grad_norm": 4.4235920906066895,
+      "learning_rate": 5.38e-06,
+      "loss": 0.1599,
+      "step": 270
+    },
+    {
+      "epoch": 0.007491037508694954,
+      "grad_norm": 4.77069091796875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.18,
+      "step": 280
+    },
+    {
+      "epoch": 0.007758574562576917,
+      "grad_norm": 5.606808185577393,
+      "learning_rate": 5.78e-06,
+      "loss": 0.1562,
+      "step": 290
+    },
+    {
+      "epoch": 0.00802611161645888,
+      "grad_norm": 3.3532209396362305,
+      "learning_rate": 5.98e-06,
+      "loss": 0.2299,
+      "step": 300
+    },
+    {
+      "epoch": 0.008293648670340843,
+      "grad_norm": 8.097010612487793,
+      "learning_rate": 6.18e-06,
+      "loss": 0.2286,
+      "step": 310
+    },
+    {
+      "epoch": 0.008561185724222806,
+      "grad_norm": 5.691704273223877,
+      "learning_rate": 6.38e-06,
+      "loss": 0.2212,
+      "step": 320
+    },
+    {
+      "epoch": 0.008828722778104767,
+      "grad_norm": 1.5986082553863525,
+      "learning_rate": 6.58e-06,
+      "loss": 0.2152,
+      "step": 330
+    },
+    {
+      "epoch": 0.00909625983198673,
+      "grad_norm": 2.6581177711486816,
+      "learning_rate": 6.78e-06,
+      "loss": 0.2137,
+      "step": 340
+    },
+    {
+      "epoch": 0.009363796885868693,
+      "grad_norm": 12.409740447998047,
+      "learning_rate": 6.98e-06,
+      "loss": 0.1903,
+      "step": 350
+    },
+    {
+      "epoch": 0.009631333939750655,
+      "grad_norm": 8.475409507751465,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.1799,
+      "step": 360
+    },
+    {
+      "epoch": 0.009898870993632618,
+      "grad_norm": 3.4934091567993164,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.2267,
+      "step": 370
+    },
+    {
+      "epoch": 0.01016640804751458,
+      "grad_norm": 5.401501655578613,
+      "learning_rate": 7.580000000000001e-06,
+      "loss": 0.1488,
+      "step": 380
+    },
+    {
+      "epoch": 0.010433945101396544,
+      "grad_norm": 5.623922824859619,
+      "learning_rate": 7.78e-06,
+      "loss": 0.2007,
+      "step": 390
+    },
+    {
+      "epoch": 0.010701482155278505,
+      "grad_norm": 2.949655532836914,
+      "learning_rate": 7.98e-06,
+      "loss": 0.248,
+      "step": 400
+    },
+    {
+      "epoch": 0.010969019209160468,
+      "grad_norm": 6.962086200714111,
+      "learning_rate": 8.18e-06,
+      "loss": 0.175,
+      "step": 410
+    },
+    {
+      "epoch": 0.011236556263042432,
+      "grad_norm": 4.8697404861450195,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.1695,
+      "step": 420
+    },
+    {
+      "epoch": 0.011504093316924395,
+      "grad_norm": 3.704503297805786,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.1439,
+      "step": 430
+    },
+    {
+      "epoch": 0.011771630370806356,
+      "grad_norm": 3.6497867107391357,
+      "learning_rate": 8.78e-06,
+      "loss": 0.1716,
+      "step": 440
+    },
+    {
+      "epoch": 0.01203916742468832,
+      "grad_norm": 4.83482027053833,
+      "learning_rate": 8.98e-06,
+      "loss": 0.2563,
+      "step": 450
+    },
+    {
+      "epoch": 0.012306704478570282,
+      "grad_norm": 9.308363914489746,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.2905,
+      "step": 460
+    },
+    {
+      "epoch": 0.012574241532452245,
+      "grad_norm": 3.6065757274627686,
+      "learning_rate": 9.38e-06,
+      "loss": 0.2134,
+      "step": 470
+    },
+    {
+      "epoch": 0.012841778586334207,
+      "grad_norm": 4.783448219299316,
+      "learning_rate": 9.58e-06,
+      "loss": 0.2248,
+      "step": 480
+    },
+    {
+      "epoch": 0.01310931564021617,
+      "grad_norm": 7.807690620422363,
+      "learning_rate": 9.78e-06,
+      "loss": 0.1916,
+      "step": 490
+    },
+    {
+      "epoch": 0.013376852694098133,
+      "grad_norm": 5.255489826202393,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.1749,
+      "step": 500
+    },
+    {
+      "epoch": 0.013644389747980096,
+      "grad_norm": 5.288446426391602,
+      "learning_rate": 1.018e-05,
+      "loss": 0.1667,
+      "step": 510
+    },
+    {
+      "epoch": 0.013911926801862057,
+      "grad_norm": 3.8384625911712646,
+      "learning_rate": 1.038e-05,
+      "loss": 0.1845,
+      "step": 520
+    },
+    {
+      "epoch": 0.01417946385574402,
+      "grad_norm": 4.450418472290039,
+      "learning_rate": 1.058e-05,
+      "loss": 0.1391,
+      "step": 530
+    },
+    {
+      "epoch": 0.014447000909625984,
+      "grad_norm": 5.301421165466309,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.1145,
+      "step": 540
+    },
+    {
+      "epoch": 0.014714537963507945,
+      "grad_norm": 6.675409317016602,
+      "learning_rate": 1.098e-05,
+      "loss": 0.1192,
+      "step": 550
+    },
+    {
+      "epoch": 0.014982075017389908,
+      "grad_norm": 3.589747428894043,
+      "learning_rate": 1.118e-05,
+      "loss": 0.3026,
+      "step": 560
+    },
+    {
+      "epoch": 0.015249612071271871,
+      "grad_norm": 5.0948920249938965,
+      "learning_rate": 1.1380000000000001e-05,
+      "loss": 0.1483,
+      "step": 570
+    },
+    {
+      "epoch": 0.015517149125153834,
+      "grad_norm": 4.402790546417236,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.2121,
+      "step": 580
+    },
+    {
+      "epoch": 0.015784686179035796,
+      "grad_norm": 3.459892511367798,
+      "learning_rate": 1.178e-05,
+      "loss": 0.1816,
+      "step": 590
+    },
+    {
+      "epoch": 0.01605222323291776,
+      "grad_norm": 4.197982311248779,
+      "learning_rate": 1.198e-05,
+      "loss": 0.1405,
+      "step": 600
+    },
+    {
+      "epoch": 0.016319760286799722,
+      "grad_norm": 4.815362453460693,
+      "learning_rate": 1.2180000000000002e-05,
+      "loss": 0.1632,
+      "step": 610
+    },
+    {
+      "epoch": 0.016587297340681685,
+      "grad_norm": 5.058053970336914,
+      "learning_rate": 1.238e-05,
+      "loss": 0.2093,
+      "step": 620
+    },
+    {
+      "epoch": 0.016854834394563648,
+      "grad_norm": 11.981295585632324,
+      "learning_rate": 1.258e-05,
+      "loss": 0.2021,
+      "step": 630
+    },
+    {
+      "epoch": 0.01712237144844561,
+      "grad_norm": 2.0436439514160156,
+      "learning_rate": 1.278e-05,
+      "loss": 0.1646,
+      "step": 640
+    },
+    {
+      "epoch": 0.01738990850232757,
+      "grad_norm": 2.7562735080718994,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.1332,
+      "step": 650
+    },
+    {
+      "epoch": 0.017657445556209534,
+      "grad_norm": 1.0508995056152344,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.2284,
+      "step": 660
+    },
+    {
+      "epoch": 0.017924982610091497,
+      "grad_norm": 6.9363226890563965,
+      "learning_rate": 1.338e-05,
+      "loss": 0.1726,
+      "step": 670
+    },
+    {
+      "epoch": 0.01819251966397346,
+      "grad_norm": 1.884033441543579,
+      "learning_rate": 1.358e-05,
+      "loss": 0.1254,
+      "step": 680
+    },
+    {
+      "epoch": 0.018460056717855423,
+      "grad_norm": 5.572957992553711,
+      "learning_rate": 1.3780000000000002e-05,
+      "loss": 0.1166,
+      "step": 690
+    },
+    {
+      "epoch": 0.018727593771737387,
+      "grad_norm": 3.1382200717926025,
+      "learning_rate": 1.3980000000000002e-05,
+      "loss": 0.1522,
+      "step": 700
+    },
+    {
+      "epoch": 0.01899513082561935,
+      "grad_norm": 3.6476964950561523,
+      "learning_rate": 1.4180000000000001e-05,
+      "loss": 0.1857,
+      "step": 710
+    },
+    {
+      "epoch": 0.01926266787950131,
+      "grad_norm": 6.777200222015381,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.1778,
+      "step": 720
+    },
+    {
+      "epoch": 0.019530204933383272,
+      "grad_norm": 4.470036029815674,
+      "learning_rate": 1.4580000000000003e-05,
+      "loss": 0.2573,
+      "step": 730
+    },
+    {
+      "epoch": 0.019797741987265235,
+      "grad_norm": 4.104392051696777,
+      "learning_rate": 1.4779999999999999e-05,
+      "loss": 0.1455,
+      "step": 740
+    },
+    {
+      "epoch": 0.0200652790411472,
+      "grad_norm": 4.102880001068115,
+      "learning_rate": 1.4979999999999999e-05,
+      "loss": 0.1582,
+      "step": 750
+    },
+    {
+      "epoch": 0.02033281609502916,
+      "grad_norm": 2.7197999954223633,
+      "learning_rate": 1.518e-05,
+      "loss": 0.2922,
+      "step": 760
+    },
+    {
+      "epoch": 0.020600353148911125,
+      "grad_norm": 3.6682119369506836,
+      "learning_rate": 1.538e-05,
+      "loss": 0.1688,
+      "step": 770
+    },
+    {
+      "epoch": 0.020867890202793088,
+      "grad_norm": 5.457839012145996,
+      "learning_rate": 1.558e-05,
+      "loss": 0.2239,
+      "step": 780
+    },
+    {
+      "epoch": 0.02113542725667505,
+      "grad_norm": 3.8674941062927246,
+      "learning_rate": 1.578e-05,
+      "loss": 0.1674,
+      "step": 790
+    },
+    {
+      "epoch": 0.02140296431055701,
+      "grad_norm": 6.690402030944824,
+      "learning_rate": 1.598e-05,
+      "loss": 0.1468,
+      "step": 800
+    },
+    {
+      "epoch": 0.021670501364438974,
+      "grad_norm": 8.31381607055664,
+      "learning_rate": 1.618e-05,
+      "loss": 0.2156,
+      "step": 810
+    },
+    {
+      "epoch": 0.021938038418320937,
+      "grad_norm": 4.517109394073486,
+      "learning_rate": 1.6380000000000002e-05,
+      "loss": 0.2704,
+      "step": 820
+    },
+    {
+      "epoch": 0.0222055754722029,
+      "grad_norm": 5.689136505126953,
+      "learning_rate": 1.658e-05,
+      "loss": 0.1111,
+      "step": 830
+    },
+    {
+      "epoch": 0.022473112526084863,
+      "grad_norm": 2.1215598583221436,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.1979,
+      "step": 840
+    },
+    {
+      "epoch": 0.022740649579966826,
+      "grad_norm": 4.404816150665283,
+      "learning_rate": 1.698e-05,
+      "loss": 0.229,
+      "step": 850
+    },
+    {
+      "epoch": 0.02300818663384879,
+      "grad_norm": 3.351616382598877,
+      "learning_rate": 1.718e-05,
+      "loss": 0.2004,
+      "step": 860
+    },
+    {
+      "epoch": 0.02327572368773075,
+      "grad_norm": 4.51784086227417,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.2074,
+      "step": 870
+    },
+    {
+      "epoch": 0.023543260741612712,
+      "grad_norm": 3.863691806793213,
+      "learning_rate": 1.758e-05,
+      "loss": 0.1774,
+      "step": 880
+    },
+    {
+      "epoch": 0.023810797795494675,
+      "grad_norm": 3.5073671340942383,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.1721,
+      "step": 890
+    },
+    {
+      "epoch": 0.02407833484937664,
+      "grad_norm": 2.672020196914673,
+      "learning_rate": 1.798e-05,
+      "loss": 0.2046,
+      "step": 900
+    },
+    {
+      "epoch": 0.0243458719032586,
+      "grad_norm": 6.270020961761475,
+      "learning_rate": 1.818e-05,
+      "loss": 0.1425,
+      "step": 910
+    },
+    {
+      "epoch": 0.024613408957140565,
+      "grad_norm": 7.55366849899292,
+      "learning_rate": 1.838e-05,
+      "loss": 0.1881,
+      "step": 920
+    },
+    {
+      "epoch": 0.024880946011022528,
+      "grad_norm": 3.1182713508605957,
+      "learning_rate": 1.858e-05,
+      "loss": 0.2621,
+      "step": 930
+    },
+    {
+      "epoch": 0.02514848306490449,
+      "grad_norm": 6.137206554412842,
+      "learning_rate": 1.878e-05,
+      "loss": 0.1486,
+      "step": 940
+    },
+    {
+      "epoch": 0.02541602011878645,
+      "grad_norm": 2.543691396713257,
+      "learning_rate": 1.898e-05,
+      "loss": 0.1808,
+      "step": 950
+    },
+    {
+      "epoch": 0.025683557172668414,
+      "grad_norm": 3.0582141876220703,
+      "learning_rate": 1.918e-05,
+      "loss": 0.1482,
+      "step": 960
+    },
+    {
+      "epoch": 0.025951094226550377,
+      "grad_norm": 3.3326706886291504,
+      "learning_rate": 1.938e-05,
+      "loss": 0.1304,
+      "step": 970
+    },
+    {
+      "epoch": 0.02621863128043234,
+      "grad_norm": 12.924772262573242,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.2182,
+      "step": 980
+    },
+    {
+      "epoch": 0.026486168334314303,
+      "grad_norm": 3.6143674850463867,
+      "learning_rate": 1.978e-05,
+      "loss": 0.1618,
+      "step": 990
+    },
+    {
+      "epoch": 0.026753705388196266,
+      "grad_norm": 1.5050777196884155,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.1723,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02702124244207823,
+      "grad_norm": 2.934372901916504,
+      "learning_rate": 2.0180000000000003e-05,
+      "loss": 0.2051,
+      "step": 1010
+    },
+    {
+      "epoch": 0.027288779495960192,
+      "grad_norm": 2.3569283485412598,
+      "learning_rate": 2.038e-05,
+      "loss": 0.2511,
+      "step": 1020
+    },
+    {
+      "epoch": 0.027556316549842152,
+      "grad_norm": 4.302797317504883,
+      "learning_rate": 2.0580000000000003e-05,
+      "loss": 0.0966,
+      "step": 1030
+    },
+    {
+      "epoch": 0.027823853603724115,
+      "grad_norm": 1.8805347681045532,
+      "learning_rate": 2.078e-05,
+      "loss": 0.1336,
+      "step": 1040
+    },
+    {
+      "epoch": 0.028091390657606078,
+      "grad_norm": 5.796336650848389,
+      "learning_rate": 2.098e-05,
+      "loss": 0.0939,
+      "step": 1050
+    },
+    {
+      "epoch": 0.02835892771148804,
+      "grad_norm": 1.4898550510406494,
+      "learning_rate": 2.118e-05,
+      "loss": 0.166,
+      "step": 1060
+    },
+    {
+      "epoch": 0.028626464765370004,
+      "grad_norm": 3.1921794414520264,
+      "learning_rate": 2.138e-05,
+      "loss": 0.267,
+      "step": 1070
+    },
+    {
+      "epoch": 0.028894001819251967,
+      "grad_norm": 4.336124897003174,
+      "learning_rate": 2.158e-05,
+      "loss": 0.1163,
+      "step": 1080
+    },
+    {
+      "epoch": 0.02916153887313393,
+      "grad_norm": 3.3892979621887207,
+      "learning_rate": 2.178e-05,
+      "loss": 0.1439,
+      "step": 1090
+    },
+    {
+      "epoch": 0.02942907592701589,
+      "grad_norm": 2.5495309829711914,
+      "learning_rate": 2.198e-05,
+      "loss": 0.1269,
+      "step": 1100
+    },
+    {
+      "epoch": 0.029696612980897853,
+      "grad_norm": 3.3253257274627686,
+      "learning_rate": 2.218e-05,
+      "loss": 0.1897,
+      "step": 1110
+    },
+    {
+      "epoch": 0.029964150034779816,
+      "grad_norm": 0.9029315114021301,
+      "learning_rate": 2.2380000000000003e-05,
+      "loss": 0.1051,
+      "step": 1120
+    },
+    {
+      "epoch": 0.03023168708866178,
+      "grad_norm": 2.3470475673675537,
+      "learning_rate": 2.258e-05,
+      "loss": 0.1435,
+      "step": 1130
+    },
+    {
+      "epoch": 0.030499224142543743,
+      "grad_norm": 1.7739107608795166,
+      "learning_rate": 2.2780000000000002e-05,
+      "loss": 0.148,
+      "step": 1140
+    },
+    {
+      "epoch": 0.030766761196425706,
+      "grad_norm": 2.145885705947876,
+      "learning_rate": 2.298e-05,
+      "loss": 0.1866,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03103429825030767,
+      "grad_norm": 2.240212917327881,
+      "learning_rate": 2.318e-05,
+      "loss": 0.2587,
+      "step": 1160
+    },
+    {
+      "epoch": 0.03130183530418963,
+      "grad_norm": 3.2969062328338623,
+      "learning_rate": 2.3380000000000003e-05,
+      "loss": 0.1893,
+      "step": 1170
+    },
+    {
+      "epoch": 0.03156937235807159,
+      "grad_norm": 3.984684705734253,
+      "learning_rate": 2.358e-05,
+      "loss": 0.1652,
+      "step": 1180
+    },
+    {
+      "epoch": 0.03183690941195356,
+      "grad_norm": 2.990130662918091,
+      "learning_rate": 2.3780000000000003e-05,
+      "loss": 0.2044,
+      "step": 1190
+    },
+    {
+      "epoch": 0.03210444646583552,
+      "grad_norm": 5.511849403381348,
+      "learning_rate": 2.398e-05,
+      "loss": 0.2477,
+      "step": 1200
+    },
+    {
+      "epoch": 0.032371983519717484,
+      "grad_norm": 5.792886734008789,
+      "learning_rate": 2.418e-05,
+      "loss": 0.1616,
+      "step": 1210
+    },
+    {
+      "epoch": 0.032639520573599444,
+      "grad_norm": 2.6802847385406494,
+      "learning_rate": 2.438e-05,
+      "loss": 0.1518,
+      "step": 1220
+    },
+    {
+      "epoch": 0.032907057627481404,
+      "grad_norm": 5.089447021484375,
+      "learning_rate": 2.4580000000000002e-05,
+      "loss": 0.1942,
+      "step": 1230
+    },
+    {
+      "epoch": 0.03317459468136337,
+      "grad_norm": 4.769607067108154,
+      "learning_rate": 2.478e-05,
+      "loss": 0.1065,
+      "step": 1240
+    },
+    {
+      "epoch": 0.03344213173524533,
+      "grad_norm": 4.552340507507324,
+      "learning_rate": 2.498e-05,
+      "loss": 0.1289,
+      "step": 1250
+    },
+    {
+      "epoch": 0.033709668789127296,
+      "grad_norm": 4.012768745422363,
+      "learning_rate": 2.5180000000000003e-05,
+      "loss": 0.1636,
+      "step": 1260
+    },
+    {
+      "epoch": 0.033977205843009256,
+      "grad_norm": 2.8586196899414062,
+      "learning_rate": 2.5380000000000004e-05,
+      "loss": 0.1839,
+      "step": 1270
+    },
+    {
+      "epoch": 0.03424474289689122,
+      "grad_norm": 5.532242298126221,
+      "learning_rate": 2.5580000000000002e-05,
+      "loss": 0.1079,
+      "step": 1280
+    },
+    {
+      "epoch": 0.03451227995077318,
+      "grad_norm": 3.0744903087615967,
+      "learning_rate": 2.5779999999999997e-05,
+      "loss": 0.1645,
+      "step": 1290
+    },
+    {
+      "epoch": 0.03477981700465514,
+      "grad_norm": 3.0554075241088867,
+      "learning_rate": 2.598e-05,
+      "loss": 0.2259,
+      "step": 1300
+    },
+    {
+      "epoch": 0.03504735405853711,
+      "grad_norm": 2.1796746253967285,
+      "learning_rate": 2.618e-05,
+      "loss": 0.1356,
+      "step": 1310
+    },
+    {
+      "epoch": 0.03531489111241907,
+      "grad_norm": 4.0079827308654785,
+      "learning_rate": 2.6379999999999998e-05,
+      "loss": 0.1659,
+      "step": 1320
+    },
+    {
+      "epoch": 0.035582428166301035,
+      "grad_norm": 1.1575309038162231,
+      "learning_rate": 2.658e-05,
+      "loss": 0.0967,
+      "step": 1330
+    },
+    {
+      "epoch": 0.035849965220182994,
+      "grad_norm": 4.512209892272949,
+      "learning_rate": 2.678e-05,
+      "loss": 0.1074,
+      "step": 1340
+    },
+    {
+      "epoch": 0.03611750227406496,
+      "grad_norm": 1.950435996055603,
+      "learning_rate": 2.698e-05,
+      "loss": 0.1478,
+      "step": 1350
+    },
+    {
+      "epoch": 0.03638503932794692,
+      "grad_norm": 4.1817402839660645,
+      "learning_rate": 2.718e-05,
+      "loss": 0.0916,
+      "step": 1360
+    },
+    {
+      "epoch": 0.03665257638182888,
+      "grad_norm": 5.860038757324219,
+      "learning_rate": 2.738e-05,
+      "loss": 0.1965,
+      "step": 1370
+    },
+    {
+      "epoch": 0.03692011343571085,
+      "grad_norm": 3.4507508277893066,
+      "learning_rate": 2.758e-05,
+      "loss": 0.1074,
+      "step": 1380
+    },
+    {
+      "epoch": 0.037187650489592806,
+      "grad_norm": 3.4122352600097656,
+      "learning_rate": 2.778e-05,
+      "loss": 0.2129,
+      "step": 1390
+    },
+    {
+      "epoch": 0.03745518754347477,
+      "grad_norm": 3.8269567489624023,
+      "learning_rate": 2.798e-05,
+      "loss": 0.1736,
+      "step": 1400
+    },
+    {
+      "epoch": 0.03772272459735673,
+      "grad_norm": 3.13329815864563,
+      "learning_rate": 2.818e-05,
+      "loss": 0.084,
+      "step": 1410
+    },
+    {
+      "epoch": 0.0379902616512387,
+      "grad_norm": 2.2879185676574707,
+      "learning_rate": 2.8380000000000003e-05,
+      "loss": 0.1218,
+      "step": 1420
+    },
+    {
+      "epoch": 0.03825779870512066,
+      "grad_norm": 0.7270588278770447,
+      "learning_rate": 2.858e-05,
+      "loss": 0.1048,
+      "step": 1430
+    },
+    {
+      "epoch": 0.03852533575900262,
+      "grad_norm": 2.9591336250305176,
+      "learning_rate": 2.8780000000000002e-05,
+      "loss": 0.1147,
+      "step": 1440
+    },
+    {
+      "epoch": 0.038792872812884585,
+      "grad_norm": 3.4224274158477783,
+      "learning_rate": 2.898e-05,
+      "loss": 0.0882,
+      "step": 1450
+    },
+    {
+      "epoch": 0.039060409866766545,
+      "grad_norm": 5.170720100402832,
+      "learning_rate": 2.9180000000000002e-05,
+      "loss": 0.0965,
+      "step": 1460
+    },
+    {
+      "epoch": 0.03932794692064851,
+      "grad_norm": 1.7120144367218018,
+      "learning_rate": 2.9380000000000003e-05,
+      "loss": 0.1719,
+      "step": 1470
+    },
+    {
+      "epoch": 0.03959548397453047,
+      "grad_norm": 2.514927625656128,
+      "learning_rate": 2.958e-05,
+      "loss": 0.2015,
+      "step": 1480
+    },
+    {
+      "epoch": 0.03986302102841244,
+      "grad_norm": 2.8296895027160645,
+      "learning_rate": 2.9780000000000003e-05,
+      "loss": 0.1924,
+      "step": 1490
+    },
+    {
+      "epoch": 0.0401305580822944,
+      "grad_norm": 2.131199598312378,
+      "learning_rate": 2.998e-05,
+      "loss": 0.1092,
+      "step": 1500
+    },
+    {
+      "epoch": 0.040398095136176364,
+      "grad_norm": 1.194530725479126,
+      "learning_rate": 3.0180000000000002e-05,
+      "loss": 0.0924,
+      "step": 1510
+    },
+    {
+      "epoch": 0.04066563219005832,
+      "grad_norm": 2.1516709327697754,
+      "learning_rate": 3.0380000000000004e-05,
+      "loss": 0.1845,
+      "step": 1520
+    },
+    {
+      "epoch": 0.04093316924394028,
+      "grad_norm": 1.2379114627838135,
+      "learning_rate": 3.058e-05,
+      "loss": 0.198,
+      "step": 1530
+    },
+    {
+      "epoch": 0.04120070629782225,
+      "grad_norm": 1.4216729402542114,
+      "learning_rate": 3.078e-05,
+      "loss": 0.0661,
+      "step": 1540
+    },
+    {
+      "epoch": 0.04146824335170421,
+      "grad_norm": 1.9382178783416748,
+      "learning_rate": 3.0980000000000005e-05,
+      "loss": 0.1804,
+      "step": 1550
+    },
+    {
+      "epoch": 0.041735780405586176,
+      "grad_norm": 4.734653949737549,
+      "learning_rate": 3.118e-05,
+      "loss": 0.1536,
+      "step": 1560
+    },
+    {
+      "epoch": 0.042003317459468135,
+      "grad_norm": 3.552345037460327,
+      "learning_rate": 3.138e-05,
+      "loss": 0.1179,
+      "step": 1570
+    },
+    {
+      "epoch": 0.0422708545133501,
+      "grad_norm": 2.2107479572296143,
+      "learning_rate": 3.1580000000000006e-05,
+      "loss": 0.0951,
+      "step": 1580
+    },
+    {
+      "epoch": 0.04253839156723206,
+      "grad_norm": 1.98602294921875,
+      "learning_rate": 3.1780000000000004e-05,
+      "loss": 0.2111,
+      "step": 1590
+    },
+    {
+      "epoch": 0.04280592862111402,
+      "grad_norm": 5.028587818145752,
+      "learning_rate": 3.198e-05,
+      "loss": 0.1912,
+      "step": 1600
+    },
+    {
+      "epoch": 0.04307346567499599,
+      "grad_norm": 1.3388959169387817,
+      "learning_rate": 3.218e-05,
+      "loss": 0.1232,
+      "step": 1610
+    },
+    {
+      "epoch": 0.04334100272887795,
+      "grad_norm": 4.712498188018799,
+      "learning_rate": 3.238e-05,
+      "loss": 0.1215,
+      "step": 1620
+    },
+    {
+      "epoch": 0.043608539782759914,
+      "grad_norm": 5.163971424102783,
+      "learning_rate": 3.2579999999999996e-05,
+      "loss": 0.1014,
+      "step": 1630
+    },
+    {
+      "epoch": 0.043876076836641874,
+      "grad_norm": 3.163271427154541,
+      "learning_rate": 3.278e-05,
+      "loss": 0.1007,
+      "step": 1640
+    },
+    {
+      "epoch": 0.04414361389052384,
+      "grad_norm": 4.756677150726318,
+      "learning_rate": 3.298e-05,
+      "loss": 0.1522,
+      "step": 1650
+    },
+    {
+      "epoch": 0.0444111509444058,
+      "grad_norm": 4.2119622230529785,
+      "learning_rate": 3.318e-05,
+      "loss": 0.2771,
+      "step": 1660
+    },
+    {
+      "epoch": 0.04467868799828776,
+      "grad_norm": 3.651808977127075,
+      "learning_rate": 3.338e-05,
+      "loss": 0.11,
+      "step": 1670
+    },
+    {
+      "epoch": 0.044946225052169726,
+      "grad_norm": 6.368651390075684,
+      "learning_rate": 3.358e-05,
+      "loss": 0.1587,
+      "step": 1680
+    },
+    {
+      "epoch": 0.045213762106051686,
+      "grad_norm": 1.9080853462219238,
+      "learning_rate": 3.378e-05,
+      "loss": 0.208,
+      "step": 1690
+    },
+    {
+      "epoch": 0.04548129915993365,
+      "grad_norm": 1.4746414422988892,
+      "learning_rate": 3.398e-05,
+      "loss": 0.1569,
+      "step": 1700
+    },
+    {
+      "epoch": 0.04574883621381561,
+      "grad_norm": 2.3431146144866943,
+      "learning_rate": 3.418e-05,
+      "loss": 0.4569,
+      "step": 1710
+    },
+    {
+      "epoch": 0.04601637326769758,
+      "grad_norm": 2.641139030456543,
+      "learning_rate": 3.438e-05,
+      "loss": 0.1868,
+      "step": 1720
+    },
+    {
+      "epoch": 0.04628391032157954,
+      "grad_norm": 2.590181350708008,
+      "learning_rate": 3.4580000000000004e-05,
+      "loss": 0.1715,
+      "step": 1730
+    },
+    {
+      "epoch": 0.0465514473754615,
+      "grad_norm": 2.873771905899048,
+      "learning_rate": 3.478e-05,
+      "loss": 0.1715,
+      "step": 1740
+    },
+    {
+      "epoch": 0.046818984429343465,
+      "grad_norm": 2.491135835647583,
+      "learning_rate": 3.498e-05,
+      "loss": 0.2051,
+      "step": 1750
+    },
+    {
+      "epoch": 0.047086521483225424,
+      "grad_norm": 8.574052810668945,
+      "learning_rate": 3.518e-05,
+      "loss": 0.2161,
+      "step": 1760
+    },
+    {
+      "epoch": 0.04735405853710739,
+      "grad_norm": 6.06572961807251,
+      "learning_rate": 3.5380000000000003e-05,
+      "loss": 0.1851,
+      "step": 1770
+    },
+    {
+      "epoch": 0.04762159559098935,
+      "grad_norm": 6.41258430480957,
+      "learning_rate": 3.558e-05,
+      "loss": 0.1927,
+      "step": 1780
+    },
+    {
+      "epoch": 0.04788913264487132,
+      "grad_norm": 3.417293071746826,
+      "learning_rate": 3.578e-05,
+      "loss": 0.1651,
+      "step": 1790
+    },
+    {
+      "epoch": 0.04815666969875328,
+      "grad_norm": 3.2387452125549316,
+      "learning_rate": 3.5980000000000004e-05,
+      "loss": 0.1387,
+      "step": 1800
+    },
+    {
+      "epoch": 0.04842420675263524,
+      "grad_norm": 3.6368348598480225,
+      "learning_rate": 3.618e-05,
+      "loss": 0.1465,
+      "step": 1810
+    },
+    {
+      "epoch": 0.0486917438065172,
+      "grad_norm": 1.795356273651123,
+      "learning_rate": 3.638e-05,
+      "loss": 0.1857,
+      "step": 1820
+    },
+    {
+      "epoch": 0.04895928086039916,
+      "grad_norm": 2.705772638320923,
+      "learning_rate": 3.6580000000000006e-05,
+      "loss": 0.1174,
+      "step": 1830
+    },
+    {
+      "epoch": 0.04922681791428113,
+      "grad_norm": 3.347998857498169,
+      "learning_rate": 3.6780000000000004e-05,
+      "loss": 0.187,
+      "step": 1840
+    },
+    {
+      "epoch": 0.04949435496816309,
+      "grad_norm": 4.399221897125244,
+      "learning_rate": 3.698e-05,
+      "loss": 0.134,
+      "step": 1850
+    },
+    {
+      "epoch": 0.049761892022045055,
+      "grad_norm": 2.6708574295043945,
+      "learning_rate": 3.7180000000000007e-05,
+      "loss": 0.1174,
+      "step": 1860
+    },
+    {
+      "epoch": 0.050029429075927015,
+      "grad_norm": 4.130018711090088,
+      "learning_rate": 3.7380000000000005e-05,
+      "loss": 0.0982,
+      "step": 1870
+    },
+    {
+      "epoch": 0.05029696612980898,
+      "grad_norm": 3.975407123565674,
+      "learning_rate": 3.758e-05,
+      "loss": 0.1957,
+      "step": 1880
+    },
+    {
+      "epoch": 0.05056450318369094,
+      "grad_norm": 4.0126142501831055,
+      "learning_rate": 3.778000000000001e-05,
+      "loss": 0.2315,
+      "step": 1890
+    },
+    {
+      "epoch": 0.0508320402375729,
+      "grad_norm": 5.350969314575195,
+      "learning_rate": 3.7980000000000006e-05,
+      "loss": 0.1471,
+      "step": 1900
+    },
+    {
+      "epoch": 0.05109957729145487,
+      "grad_norm": 1.5338300466537476,
+      "learning_rate": 3.818e-05,
+      "loss": 0.2028,
+      "step": 1910
+    },
+    {
+      "epoch": 0.05136711434533683,
+      "grad_norm": 1.7664234638214111,
+      "learning_rate": 3.838e-05,
+      "loss": 0.1067,
+      "step": 1920
+    },
+    {
+      "epoch": 0.051634651399218794,
+      "grad_norm": 3.3167011737823486,
+      "learning_rate": 3.858e-05,
+      "loss": 0.1655,
+      "step": 1930
+    },
+    {
+      "epoch": 0.05190218845310075,
+      "grad_norm": 2.8226985931396484,
+      "learning_rate": 3.878e-05,
+      "loss": 0.0888,
+      "step": 1940
+    },
+    {
+      "epoch": 0.05216972550698272,
+      "grad_norm": 2.0538065433502197,
+      "learning_rate": 3.898e-05,
+      "loss": 0.113,
+      "step": 1950
+    },
+    {
+      "epoch": 0.05243726256086468,
+      "grad_norm": 2.4558589458465576,
+      "learning_rate": 3.918e-05,
+      "loss": 0.1154,
+      "step": 1960
+    },
+    {
+      "epoch": 0.05270479961474664,
+      "grad_norm": 4.267798900604248,
+      "learning_rate": 3.938e-05,
+      "loss": 0.1446,
+      "step": 1970
+    },
+    {
+      "epoch": 0.052972336668628606,
+      "grad_norm": 4.441429615020752,
+      "learning_rate": 3.958e-05,
+      "loss": 0.1801,
+      "step": 1980
+    },
+    {
+      "epoch": 0.053239873722510565,
+      "grad_norm": 3.349006414413452,
+      "learning_rate": 3.978e-05,
+      "loss": 0.0771,
+      "step": 1990
+    },
+    {
+      "epoch": 0.05350741077639253,
+      "grad_norm": 1.323613166809082,
+      "learning_rate": 3.998e-05,
+      "loss": 0.0949,
+      "step": 2000
+    },
+    {
+      "epoch": 0.05377494783027449,
+      "grad_norm": 2.2788097858428955,
+      "learning_rate": 4.018e-05,
+      "loss": 0.0851,
+      "step": 2010
+    },
+    {
+      "epoch": 0.05404248488415646,
+      "grad_norm": 3.694000244140625,
+      "learning_rate": 4.038e-05,
+      "loss": 0.1932,
+      "step": 2020
+    },
+    {
+      "epoch": 0.05431002193803842,
+      "grad_norm": 1.4532541036605835,
+      "learning_rate": 4.058e-05,
+      "loss": 0.1856,
+      "step": 2030
+    },
+    {
+      "epoch": 0.054577558991920384,
+      "grad_norm": 3.8919193744659424,
+      "learning_rate": 4.078e-05,
+      "loss": 0.1306,
+      "step": 2040
+    },
+    {
+      "epoch": 0.054845096045802344,
+      "grad_norm": 3.4200127124786377,
+      "learning_rate": 4.0980000000000004e-05,
+      "loss": 0.0738,
+      "step": 2050
+    },
+    {
+      "epoch": 0.055112633099684304,
+      "grad_norm": 2.4151480197906494,
+      "learning_rate": 4.118e-05,
+      "loss": 0.1256,
+      "step": 2060
+    },
+    {
+      "epoch": 0.05538017015356627,
+      "grad_norm": 2.9375481605529785,
+      "learning_rate": 4.138e-05,
+      "loss": 0.0954,
+      "step": 2070
+    },
+    {
+      "epoch": 0.05564770720744823,
+      "grad_norm": 4.609827995300293,
+      "learning_rate": 4.1580000000000005e-05,
+      "loss": 0.0774,
+      "step": 2080
+    },
+    {
+      "epoch": 0.055915244261330196,
+      "grad_norm": 10.668481826782227,
+      "learning_rate": 4.178e-05,
+      "loss": 0.3049,
+      "step": 2090
+    },
+    {
+      "epoch": 0.056182781315212156,
+      "grad_norm": 1.959334135055542,
+      "learning_rate": 4.198e-05,
+      "loss": 0.1121,
+      "step": 2100
+    },
+    {
+      "epoch": 0.05645031836909412,
+      "grad_norm": 4.011350631713867,
+      "learning_rate": 4.2180000000000006e-05,
+      "loss": 0.147,
+      "step": 2110
+    },
+    {
+      "epoch": 0.05671785542297608,
+      "grad_norm": 2.9794819355010986,
+      "learning_rate": 4.2380000000000004e-05,
+      "loss": 0.1123,
+      "step": 2120
+    },
+    {
+      "epoch": 0.05698539247685804,
+      "grad_norm": 2.453258514404297,
+      "learning_rate": 4.258e-05,
+      "loss": 0.1918,
+      "step": 2130
+    },
+    {
+      "epoch": 0.05725292953074001,
+      "grad_norm": 3.915712833404541,
+      "learning_rate": 4.278e-05,
+      "loss": 0.1388,
+      "step": 2140
+    },
+    {
+      "epoch": 0.05752046658462197,
+      "grad_norm": 2.252056360244751,
+      "learning_rate": 4.2980000000000005e-05,
+      "loss": 0.261,
+      "step": 2150
+    },
+    {
+      "epoch": 0.057788003638503935,
+      "grad_norm": 4.746344566345215,
+      "learning_rate": 4.318e-05,
+      "loss": 0.1204,
+      "step": 2160
+    },
+    {
+      "epoch": 0.058055540692385894,
+      "grad_norm": 4.586584568023682,
+      "learning_rate": 4.338e-05,
+      "loss": 0.1038,
+      "step": 2170
+    },
+    {
+      "epoch": 0.05832307774626786,
+      "grad_norm": 2.486616373062134,
+      "learning_rate": 4.3580000000000006e-05,
+      "loss": 0.1051,
+      "step": 2180
+    },
+    {
+      "epoch": 0.05859061480014982,
+      "grad_norm": 2.557344675064087,
+      "learning_rate": 4.3780000000000004e-05,
+      "loss": 0.1279,
+      "step": 2190
+    },
+    {
+      "epoch": 0.05885815185403178,
+      "grad_norm": 5.2276177406311035,
+      "learning_rate": 4.398e-05,
+      "loss": 0.1612,
+      "step": 2200
+    },
+    {
+      "epoch": 0.05912568890791375,
+      "grad_norm": 3.798539638519287,
+      "learning_rate": 4.418000000000001e-05,
+      "loss": 0.1948,
+      "step": 2210
+    },
+    {
+      "epoch": 0.059393225961795706,
+      "grad_norm": 4.212856292724609,
+      "learning_rate": 4.438e-05,
+      "loss": 0.1493,
+      "step": 2220
+    },
+    {
+      "epoch": 0.05966076301567767,
+      "grad_norm": 3.9480957984924316,
+      "learning_rate": 4.458e-05,
+      "loss": 0.1884,
+      "step": 2230
+    },
+    {
+      "epoch": 0.05992830006955963,
+      "grad_norm": 2.8267641067504883,
+      "learning_rate": 4.478e-05,
+      "loss": 0.1498,
+      "step": 2240
+    },
+    {
+      "epoch": 0.0601958371234416,
+      "grad_norm": 1.5335146188735962,
+      "learning_rate": 4.498e-05,
+      "loss": 0.209,
+      "step": 2250
+    },
+    {
+      "epoch": 0.06046337417732356,
+      "grad_norm": 1.3726667165756226,
+      "learning_rate": 4.518e-05,
+      "loss": 0.1172,
+      "step": 2260
+    },
+    {
+      "epoch": 0.06073091123120552,
+      "grad_norm": 2.558229923248291,
+      "learning_rate": 4.538e-05,
+      "loss": 0.1313,
+      "step": 2270
+    },
+    {
+      "epoch": 0.060998448285087485,
+      "grad_norm": 3.712463855743408,
+      "learning_rate": 4.558e-05,
+      "loss": 0.177,
+      "step": 2280
+    },
+    {
+      "epoch": 0.061265985338969445,
+      "grad_norm": 0.8052666187286377,
+      "learning_rate": 4.578e-05,
+      "loss": 0.1295,
+      "step": 2290
+    },
+    {
+      "epoch": 0.06153352239285141,
+      "grad_norm": 4.44253396987915,
+      "learning_rate": 4.5980000000000004e-05,
+      "loss": 0.1735,
+      "step": 2300
+    },
+    {
+      "epoch": 0.06180105944673337,
+      "grad_norm": 2.840752601623535,
+      "learning_rate": 4.618e-05,
+      "loss": 0.1629,
+      "step": 2310
+    },
+    {
+      "epoch": 0.06206859650061534,
+      "grad_norm": 3.343148708343506,
+      "learning_rate": 4.638e-05,
+      "loss": 0.2366,
+      "step": 2320
+    },
+    {
+      "epoch": 0.0623361335544973,
+      "grad_norm": 3.830004930496216,
+      "learning_rate": 4.6580000000000005e-05,
+      "loss": 0.2666,
+      "step": 2330
+    },
+    {
+      "epoch": 0.06260367060837926,
+      "grad_norm": 3.000541925430298,
+      "learning_rate": 4.678e-05,
+      "loss": 0.1334,
+      "step": 2340
+    },
+    {
+      "epoch": 0.06287120766226122,
+      "grad_norm": 6.345517635345459,
+      "learning_rate": 4.698e-05,
+      "loss": 0.2426,
+      "step": 2350
+    },
+    {
+      "epoch": 0.06313874471614318,
+      "grad_norm": 2.9436278343200684,
+      "learning_rate": 4.718e-05,
+      "loss": 0.1206,
+      "step": 2360
+    },
+    {
+      "epoch": 0.06340628177002515,
+      "grad_norm": 3.979114294052124,
+      "learning_rate": 4.7380000000000004e-05,
+      "loss": 0.0709,
+      "step": 2370
+    },
+    {
+      "epoch": 0.06367381882390712,
+      "grad_norm": 2.7514750957489014,
+      "learning_rate": 4.758e-05,
+      "loss": 0.1522,
+      "step": 2380
+    },
+    {
+      "epoch": 0.06394135587778907,
+      "grad_norm": 3.2403063774108887,
+      "learning_rate": 4.778e-05,
+      "loss": 0.1979,
+      "step": 2390
+    },
+    {
+      "epoch": 0.06420889293167104,
+      "grad_norm": 1.9483088254928589,
+      "learning_rate": 4.7980000000000005e-05,
+      "loss": 0.154,
+      "step": 2400
+    },
+    {
+      "epoch": 0.064476429985553,
+      "grad_norm": 5.368651390075684,
+      "learning_rate": 4.818e-05,
+      "loss": 0.1932,
+      "step": 2410
+    },
+    {
+      "epoch": 0.06474396703943497,
+      "grad_norm": 5.0648884773254395,
+      "learning_rate": 4.838e-05,
+      "loss": 0.1612,
+      "step": 2420
+    },
+    {
+      "epoch": 0.06501150409331692,
+      "grad_norm": 2.842226505279541,
+      "learning_rate": 4.8580000000000006e-05,
+      "loss": 0.1693,
+      "step": 2430
+    },
+    {
+      "epoch": 0.06527904114719889,
+      "grad_norm": 3.5029146671295166,
+      "learning_rate": 4.8780000000000004e-05,
+      "loss": 0.1925,
+      "step": 2440
+    },
+    {
+      "epoch": 0.06554657820108085,
+      "grad_norm": 1.2403169870376587,
+      "learning_rate": 4.898e-05,
+      "loss": 0.098,
+      "step": 2450
+    },
+    {
+      "epoch": 0.06581411525496281,
+      "grad_norm": 1.368030309677124,
+      "learning_rate": 4.918000000000001e-05,
+      "loss": 0.1227,
+      "step": 2460
+    },
+    {
+      "epoch": 0.06608165230884477,
+      "grad_norm": 4.0152411460876465,
+      "learning_rate": 4.9380000000000005e-05,
+      "loss": 0.1707,
+      "step": 2470
+    },
+    {
+      "epoch": 0.06634918936272674,
+      "grad_norm": 1.8496848344802856,
+      "learning_rate": 4.958e-05,
+      "loss": 0.1207,
+      "step": 2480
+    },
+    {
+      "epoch": 0.06661672641660871,
+      "grad_norm": 3.3116636276245117,
+      "learning_rate": 4.978e-05,
+      "loss": 0.1914,
+      "step": 2490
+    },
+    {
+      "epoch": 0.06688426347049066,
+      "grad_norm": 3.8023338317871094,
+      "learning_rate": 4.9980000000000006e-05,
+      "loss": 0.2265,
+      "step": 2500
+    },
+    {
+      "epoch": 0.06715180052437263,
+      "grad_norm": 5.016868591308594,
+      "learning_rate": 5.0180000000000004e-05,
+      "loss": 0.1509,
+      "step": 2510
+    },
+    {
+      "epoch": 0.06741933757825459,
+      "grad_norm": 4.325119972229004,
+      "learning_rate": 5.038e-05,
+      "loss": 0.1367,
+      "step": 2520
+    },
+    {
+      "epoch": 0.06768687463213655,
+      "grad_norm": 3.81866455078125,
+      "learning_rate": 5.058000000000001e-05,
+      "loss": 0.1432,
+      "step": 2530
+    },
+    {
+      "epoch": 0.06795441168601851,
+      "grad_norm": 2.2937376499176025,
+      "learning_rate": 5.0780000000000005e-05,
+      "loss": 0.0782,
+      "step": 2540
+    },
+    {
+      "epoch": 0.06822194873990048,
+      "grad_norm": 4.269251346588135,
+      "learning_rate": 5.098e-05,
+      "loss": 0.1415,
+      "step": 2550
+    },
+    {
+      "epoch": 0.06848948579378245,
+      "grad_norm": 2.6492388248443604,
+      "learning_rate": 5.118000000000001e-05,
+      "loss": 0.1077,
+      "step": 2560
+    },
+    {
+      "epoch": 0.0687570228476644,
+      "grad_norm": 2.7712574005126953,
+      "learning_rate": 5.1380000000000006e-05,
+      "loss": 0.114,
+      "step": 2570
+    },
+    {
+      "epoch": 0.06902455990154636,
+      "grad_norm": 2.260787010192871,
+      "learning_rate": 5.1580000000000004e-05,
+      "loss": 0.1559,
+      "step": 2580
+    },
+    {
+      "epoch": 0.06929209695542833,
+      "grad_norm": 2.370650053024292,
+      "learning_rate": 5.178000000000001e-05,
+      "loss": 0.1612,
+      "step": 2590
+    },
+    {
+      "epoch": 0.06955963400931028,
+      "grad_norm": 2.600682258605957,
+      "learning_rate": 5.198000000000001e-05,
+      "loss": 0.1543,
+      "step": 2600
+    },
+    {
+      "epoch": 0.06982717106319225,
+      "grad_norm": 2.978938341140747,
+      "learning_rate": 5.2180000000000005e-05,
+      "loss": 0.103,
+      "step": 2610
+    },
+    {
+      "epoch": 0.07009470811707422,
+      "grad_norm": 7.752751350402832,
+      "learning_rate": 5.238000000000001e-05,
+      "loss": 0.1765,
+      "step": 2620
+    },
+    {
+      "epoch": 0.07036224517095618,
+      "grad_norm": 2.527935028076172,
+      "learning_rate": 5.258000000000001e-05,
+      "loss": 0.1292,
+      "step": 2630
+    },
+    {
+      "epoch": 0.07062978222483814,
+      "grad_norm": 5.200439453125,
+      "learning_rate": 5.2780000000000006e-05,
+      "loss": 0.1535,
+      "step": 2640
+    },
+    {
+      "epoch": 0.0708973192787201,
+      "grad_norm": 1.9635539054870605,
+      "learning_rate": 5.2980000000000004e-05,
+      "loss": 0.2297,
+      "step": 2650
+    },
+    {
+      "epoch": 0.07116485633260207,
+      "grad_norm": 3.156805992126465,
+      "learning_rate": 5.318000000000001e-05,
+      "loss": 0.1235,
+      "step": 2660
+    },
+    {
+      "epoch": 0.07143239338648402,
+      "grad_norm": 4.129756927490234,
+      "learning_rate": 5.338000000000001e-05,
+      "loss": 0.0935,
+      "step": 2670
+    },
+    {
+      "epoch": 0.07169993044036599,
+      "grad_norm": 2.2804853916168213,
+      "learning_rate": 5.3580000000000005e-05,
+      "loss": 0.1124,
+      "step": 2680
+    },
+    {
+      "epoch": 0.07196746749424796,
+      "grad_norm": 3.8742973804473877,
+      "learning_rate": 5.378e-05,
+      "loss": 0.1472,
+      "step": 2690
+    },
+    {
+      "epoch": 0.07223500454812992,
+      "grad_norm": 4.7431111335754395,
+      "learning_rate": 5.3979999999999995e-05,
+      "loss": 0.1389,
+      "step": 2700
+    },
+    {
+      "epoch": 0.07250254160201187,
+      "grad_norm": 2.2268123626708984,
+      "learning_rate": 5.418e-05,
+      "loss": 0.1566,
+      "step": 2710
+    },
+    {
+      "epoch": 0.07277007865589384,
+      "grad_norm": 3.2736880779266357,
+      "learning_rate": 5.438e-05,
+      "loss": 0.2286,
+      "step": 2720
+    },
+    {
+      "epoch": 0.07303761570977581,
+      "grad_norm": 2.897988796234131,
+      "learning_rate": 5.4579999999999996e-05,
+      "loss": 0.1208,
+      "step": 2730
+    },
+    {
+      "epoch": 0.07330515276365776,
+      "grad_norm": 2.223618268966675,
+      "learning_rate": 5.478e-05,
+      "loss": 0.1683,
+      "step": 2740
+    },
+    {
+      "epoch": 0.07357268981753973,
+      "grad_norm": 2.9427027702331543,
+      "learning_rate": 5.498e-05,
+      "loss": 0.1574,
+      "step": 2750
+    },
+    {
+      "epoch": 0.0738402268714217,
+      "grad_norm": 3.3442776203155518,
+      "learning_rate": 5.518e-05,
+      "loss": 0.127,
+      "step": 2760
+    },
+    {
+      "epoch": 0.07410776392530366,
+      "grad_norm": 4.0944504737854,
+      "learning_rate": 5.538e-05,
+      "loss": 0.1684,
+      "step": 2770
+    },
+    {
+      "epoch": 0.07437530097918561,
+      "grad_norm": 4.976080894470215,
+      "learning_rate": 5.558e-05,
+      "loss": 0.1983,
+      "step": 2780
+    },
+    {
+      "epoch": 0.07464283803306758,
+      "grad_norm": 3.0538461208343506,
+      "learning_rate": 5.578e-05,
+      "loss": 0.1598,
+      "step": 2790
+    },
+    {
+      "epoch": 0.07491037508694955,
+      "grad_norm": 1.4887174367904663,
+      "learning_rate": 5.5979999999999996e-05,
+      "loss": 0.1285,
+      "step": 2800
+    },
+    {
+      "epoch": 0.0751779121408315,
+      "grad_norm": 3.3375604152679443,
+      "learning_rate": 5.618e-05,
+      "loss": 0.1393,
+      "step": 2810
+    },
+    {
+      "epoch": 0.07544544919471347,
+      "grad_norm": 3.2729783058166504,
+      "learning_rate": 5.638e-05,
+      "loss": 0.129,
+      "step": 2820
+    },
+    {
+      "epoch": 0.07571298624859543,
+      "grad_norm": 1.9873477220535278,
+      "learning_rate": 5.658e-05,
+      "loss": 0.1264,
+      "step": 2830
+    },
+    {
+      "epoch": 0.0759805233024774,
+      "grad_norm": 2.9893558025360107,
+      "learning_rate": 5.678e-05,
+      "loss": 0.1134,
+      "step": 2840
+    },
+    {
+      "epoch": 0.07624806035635935,
+      "grad_norm": 2.33949613571167,
+      "learning_rate": 5.698e-05,
+      "loss": 0.1116,
+      "step": 2850
+    },
+    {
+      "epoch": 0.07651559741024132,
+      "grad_norm": 3.723203420639038,
+      "learning_rate": 5.718e-05,
+      "loss": 0.1179,
+      "step": 2860
+    },
+    {
+      "epoch": 0.07678313446412328,
+      "grad_norm": 2.6315486431121826,
+      "learning_rate": 5.738e-05,
+      "loss": 0.1199,
+      "step": 2870
+    },
+    {
+      "epoch": 0.07705067151800524,
+      "grad_norm": 5.381216049194336,
+      "learning_rate": 5.758e-05,
+      "loss": 0.1191,
+      "step": 2880
+    },
+    {
+      "epoch": 0.0773182085718872,
+      "grad_norm": 2.997633457183838,
+      "learning_rate": 5.778e-05,
+      "loss": 0.1411,
+      "step": 2890
+    },
+    {
+      "epoch": 0.07758574562576917,
+      "grad_norm": 7.40579891204834,
+      "learning_rate": 5.7980000000000004e-05,
+      "loss": 0.2547,
+      "step": 2900
+    },
+    {
+      "epoch": 0.07785328267965114,
+      "grad_norm": 2.1498169898986816,
+      "learning_rate": 5.818e-05,
+      "loss": 0.1664,
+      "step": 2910
+    },
+    {
+      "epoch": 0.07812081973353309,
+      "grad_norm": 1.9432194232940674,
+      "learning_rate": 5.838e-05,
+      "loss": 0.1746,
+      "step": 2920
+    },
+    {
+      "epoch": 0.07838835678741506,
+      "grad_norm": 2.1177845001220703,
+      "learning_rate": 5.858e-05,
+      "loss": 0.1255,
+      "step": 2930
+    },
+    {
+      "epoch": 0.07865589384129702,
+      "grad_norm": 3.5222561359405518,
+      "learning_rate": 5.878e-05,
+      "loss": 0.1541,
+      "step": 2940
+    },
+    {
+      "epoch": 0.07892343089517898,
+      "grad_norm": 2.0651638507843018,
+      "learning_rate": 5.898e-05,
+      "loss": 0.1325,
+      "step": 2950
+    },
+    {
+      "epoch": 0.07919096794906094,
+      "grad_norm": 3.389939785003662,
+      "learning_rate": 5.918e-05,
+      "loss": 0.119,
+      "step": 2960
+    },
+    {
+      "epoch": 0.07945850500294291,
+      "grad_norm": 1.6713917255401611,
+      "learning_rate": 5.9380000000000004e-05,
+      "loss": 0.114,
+      "step": 2970
+    },
+    {
+      "epoch": 0.07972604205682488,
+      "grad_norm": 3.1877686977386475,
+      "learning_rate": 5.958e-05,
+      "loss": 0.1004,
+      "step": 2980
+    },
+    {
+      "epoch": 0.07999357911070683,
+      "grad_norm": 4.905401229858398,
+      "learning_rate": 5.978e-05,
+      "loss": 0.2193,
+      "step": 2990
+    },
+    {
+      "epoch": 0.0802611161645888,
+      "grad_norm": 3.805263042449951,
+      "learning_rate": 5.9980000000000005e-05,
+      "loss": 0.1851,
+      "step": 3000
+    },
+    {
+      "epoch": 0.08052865321847076,
+      "grad_norm": 3.809326171875,
+      "learning_rate": 6.018e-05,
+      "loss": 0.127,
+      "step": 3010
+    },
+    {
+      "epoch": 0.08079619027235273,
+      "grad_norm": 1.868795394897461,
+      "learning_rate": 6.038e-05,
+      "loss": 0.1507,
+      "step": 3020
+    },
+    {
+      "epoch": 0.08106372732623468,
+      "grad_norm": 0.9657459259033203,
+      "learning_rate": 6.0580000000000006e-05,
+      "loss": 0.1264,
+      "step": 3030
+    },
+    {
+      "epoch": 0.08133126438011665,
+      "grad_norm": 2.2487008571624756,
+      "learning_rate": 6.0780000000000004e-05,
+      "loss": 0.1875,
+      "step": 3040
+    },
+    {
+      "epoch": 0.08159880143399861,
+      "grad_norm": 3.458157539367676,
+      "learning_rate": 6.098e-05,
+      "loss": 0.1909,
+      "step": 3050
+    },
+    {
+      "epoch": 0.08186633848788057,
+      "grad_norm": 2.5686893463134766,
+      "learning_rate": 6.118000000000001e-05,
+      "loss": 0.248,
+      "step": 3060
+    },
+    {
+      "epoch": 0.08213387554176253,
+      "grad_norm": 4.621330738067627,
+      "learning_rate": 6.138e-05,
+      "loss": 0.1482,
+      "step": 3070
+    },
+    {
+      "epoch": 0.0824014125956445,
+      "grad_norm": 1.2376431226730347,
+      "learning_rate": 6.158e-05,
+      "loss": 0.1638,
+      "step": 3080
+    },
+    {
+      "epoch": 0.08266894964952647,
+      "grad_norm": 1.3856208324432373,
+      "learning_rate": 6.178000000000001e-05,
+      "loss": 0.1997,
+      "step": 3090
+    },
+    {
+      "epoch": 0.08293648670340842,
+      "grad_norm": 6.965011119842529,
+      "learning_rate": 6.198e-05,
+      "loss": 0.2021,
+      "step": 3100
+    },
+    {
+      "epoch": 0.08320402375729039,
+      "grad_norm": 2.3318867683410645,
+      "learning_rate": 6.218e-05,
+      "loss": 0.1597,
+      "step": 3110
+    },
+    {
+      "epoch": 0.08347156081117235,
+      "grad_norm": 2.4535961151123047,
+      "learning_rate": 6.238000000000001e-05,
+      "loss": 0.1431,
+      "step": 3120
+    },
+    {
+      "epoch": 0.0837390978650543,
+      "grad_norm": 2.1204936504364014,
+      "learning_rate": 6.258e-05,
+      "loss": 0.1319,
+      "step": 3130
+    },
+    {
+      "epoch": 0.08400663491893627,
+      "grad_norm": 2.6752521991729736,
+      "learning_rate": 6.278e-05,
+      "loss": 0.1601,
+      "step": 3140
+    },
+    {
+      "epoch": 0.08427417197281824,
+      "grad_norm": 5.335787296295166,
+      "learning_rate": 6.298000000000001e-05,
+      "loss": 0.1614,
+      "step": 3150
+    },
+    {
+      "epoch": 0.0845417090267002,
+      "grad_norm": 2.19464111328125,
+      "learning_rate": 6.318e-05,
+      "loss": 0.1907,
+      "step": 3160
+    },
+    {
+      "epoch": 0.08480924608058216,
+      "grad_norm": 7.111641883850098,
+      "learning_rate": 6.338e-05,
+      "loss": 0.2074,
+      "step": 3170
+    },
+    {
+      "epoch": 0.08507678313446412,
+      "grad_norm": 1.9513574838638306,
+      "learning_rate": 6.358000000000001e-05,
+      "loss": 0.1313,
+      "step": 3180
+    },
+    {
+      "epoch": 0.08534432018834609,
+      "grad_norm": 4.8329339027404785,
+      "learning_rate": 6.378e-05,
+      "loss": 0.2665,
+      "step": 3190
+    },
+    {
+      "epoch": 0.08561185724222804,
+      "grad_norm": 1.450269341468811,
+      "learning_rate": 6.398000000000001e-05,
+      "loss": 0.1705,
+      "step": 3200
+    },
+    {
+      "epoch": 0.08587939429611001,
+      "grad_norm": 1.9194834232330322,
+      "learning_rate": 6.418000000000001e-05,
+      "loss": 0.1235,
+      "step": 3210
+    },
+    {
+      "epoch": 0.08614693134999198,
+      "grad_norm": 3.5588901042938232,
+      "learning_rate": 6.438e-05,
+      "loss": 0.1982,
+      "step": 3220
+    },
+    {
+      "epoch": 0.08641446840387394,
+      "grad_norm": 4.20267391204834,
+      "learning_rate": 6.458000000000001e-05,
+      "loss": 0.1949,
+      "step": 3230
+    },
+    {
+      "epoch": 0.0866820054577559,
+      "grad_norm": 2.3553080558776855,
+      "learning_rate": 6.478000000000001e-05,
+      "loss": 0.2022,
+      "step": 3240
+    },
+    {
+      "epoch": 0.08694954251163786,
+      "grad_norm": 2.344111680984497,
+      "learning_rate": 6.498e-05,
+      "loss": 0.1789,
+      "step": 3250
+    },
+    {
+      "epoch": 0.08721707956551983,
+      "grad_norm": 2.5806593894958496,
+      "learning_rate": 6.518000000000001e-05,
+      "loss": 0.1051,
+      "step": 3260
+    },
+    {
+      "epoch": 0.08748461661940178,
+      "grad_norm": 2.5859286785125732,
+      "learning_rate": 6.538000000000001e-05,
+      "loss": 0.1157,
+      "step": 3270
+    },
+    {
+      "epoch": 0.08775215367328375,
+      "grad_norm": 4.3466949462890625,
+      "learning_rate": 6.558e-05,
+      "loss": 0.1833,
+      "step": 3280
+    },
+    {
+      "epoch": 0.08801969072716571,
+      "grad_norm": 3.0861928462982178,
+      "learning_rate": 6.578000000000001e-05,
+      "loss": 0.1271,
+      "step": 3290
+    },
+    {
+      "epoch": 0.08828722778104768,
+      "grad_norm": 3.5505292415618896,
+      "learning_rate": 6.598e-05,
+      "loss": 0.1444,
+      "step": 3300
+    },
+    {
+      "epoch": 0.08855476483492963,
+      "grad_norm": 2.558872938156128,
+      "learning_rate": 6.618e-05,
+      "loss": 0.1682,
+      "step": 3310
+    },
+    {
+      "epoch": 0.0888223018888116,
+      "grad_norm": 2.6185920238494873,
+      "learning_rate": 6.638e-05,
+      "loss": 0.1753,
+      "step": 3320
+    },
+    {
+      "epoch": 0.08908983894269357,
+      "grad_norm": 2.045041084289551,
+      "learning_rate": 6.658e-05,
+      "loss": 0.1702,
+      "step": 3330
+    },
+    {
+      "epoch": 0.08935737599657552,
+      "grad_norm": 2.7430830001831055,
+      "learning_rate": 6.678e-05,
+      "loss": 0.1827,
+      "step": 3340
+    },
+    {
+      "epoch": 0.08962491305045749,
+      "grad_norm": 3.024043321609497,
+      "learning_rate": 6.698e-05,
+      "loss": 0.1383,
+      "step": 3350
+    },
+    {
+      "epoch": 0.08989245010433945,
+      "grad_norm": 4.093254566192627,
+      "learning_rate": 6.718e-05,
+      "loss": 0.149,
+      "step": 3360
+    },
+    {
+      "epoch": 0.09015998715822142,
+      "grad_norm": 2.1039535999298096,
+      "learning_rate": 6.738e-05,
+      "loss": 0.2029,
+      "step": 3370
+    },
+    {
+      "epoch": 0.09042752421210337,
+      "grad_norm": 1.3505908250808716,
+      "learning_rate": 6.758e-05,
+      "loss": 0.1221,
+      "step": 3380
+    },
+    {
+      "epoch": 0.09069506126598534,
+      "grad_norm": 1.6860965490341187,
+      "learning_rate": 6.778e-05,
+      "loss": 0.1289,
+      "step": 3390
+    },
+    {
+      "epoch": 0.0909625983198673,
+      "grad_norm": 3.6444787979125977,
+      "learning_rate": 6.798e-05,
+      "loss": 0.1753,
+      "step": 3400
+    },
+    {
+      "epoch": 0.09123013537374926,
+      "grad_norm": 2.001826524734497,
+      "learning_rate": 6.818e-05,
+      "loss": 0.1341,
+      "step": 3410
+    },
+    {
+      "epoch": 0.09149767242763122,
+      "grad_norm": 2.6169731616973877,
+      "learning_rate": 6.838e-05,
+      "loss": 0.1356,
+      "step": 3420
+    },
+    {
+      "epoch": 0.09176520948151319,
+      "grad_norm": 3.243393898010254,
+      "learning_rate": 6.858e-05,
+      "loss": 0.1925,
+      "step": 3430
+    },
+    {
+      "epoch": 0.09203274653539516,
+      "grad_norm": 1.9603872299194336,
+      "learning_rate": 6.878e-05,
+      "loss": 0.1296,
+      "step": 3440
+    },
+    {
+      "epoch": 0.09230028358927711,
+      "grad_norm": 2.140632390975952,
+      "learning_rate": 6.898e-05,
+      "loss": 0.0963,
+      "step": 3450
+    },
+    {
+      "epoch": 0.09256782064315908,
+      "grad_norm": 4.610437393188477,
+      "learning_rate": 6.918e-05,
+      "loss": 0.201,
+      "step": 3460
+    },
+    {
+      "epoch": 0.09283535769704104,
+      "grad_norm": 1.385136604309082,
+      "learning_rate": 6.938e-05,
+      "loss": 0.2361,
+      "step": 3470
+    },
+    {
+      "epoch": 0.093102894750923,
+      "grad_norm": 1.304279088973999,
+      "learning_rate": 6.958e-05,
+      "loss": 0.0969,
+      "step": 3480
+    },
+    {
+      "epoch": 0.09337043180480496,
+      "grad_norm": 1.3633532524108887,
+      "learning_rate": 6.978e-05,
+      "loss": 0.1538,
+      "step": 3490
+    },
+    {
+      "epoch": 0.09363796885868693,
+      "grad_norm": 1.7657570838928223,
+      "learning_rate": 6.998e-05,
+      "loss": 0.1386,
+      "step": 3500
+    },
+    {
+      "epoch": 0.0939055059125689,
+      "grad_norm": 2.1220507621765137,
+      "learning_rate": 7.018e-05,
+      "loss": 0.1231,
+      "step": 3510
+    },
+    {
+      "epoch": 0.09417304296645085,
+      "grad_norm": 2.7939164638519287,
+      "learning_rate": 7.038e-05,
+      "loss": 0.173,
+      "step": 3520
+    },
+    {
+      "epoch": 0.09444058002033281,
+      "grad_norm": 3.7730629444122314,
+      "learning_rate": 7.058e-05,
+      "loss": 0.0976,
+      "step": 3530
+    },
+    {
+      "epoch": 0.09470811707421478,
+      "grad_norm": 2.624497413635254,
+      "learning_rate": 7.078e-05,
+      "loss": 0.103,
+      "step": 3540
+    },
+    {
+      "epoch": 0.09497565412809675,
+      "grad_norm": 3.076930046081543,
+      "learning_rate": 7.098e-05,
+      "loss": 0.1236,
+      "step": 3550
+    },
+    {
+      "epoch": 0.0952431911819787,
+      "grad_norm": 1.985494613647461,
+      "learning_rate": 7.118e-05,
+      "loss": 0.1453,
+      "step": 3560
+    },
+    {
+      "epoch": 0.09551072823586067,
+      "grad_norm": 4.473679065704346,
+      "learning_rate": 7.138e-05,
+      "loss": 0.1924,
+      "step": 3570
+    },
+    {
+      "epoch": 0.09577826528974263,
+      "grad_norm": 2.5547404289245605,
+      "learning_rate": 7.158e-05,
+      "loss": 0.1341,
+      "step": 3580
+    },
+    {
+      "epoch": 0.09604580234362459,
+      "grad_norm": 1.9036720991134644,
+      "learning_rate": 7.178000000000001e-05,
+      "loss": 0.1413,
+      "step": 3590
+    },
+    {
+      "epoch": 0.09631333939750655,
+      "grad_norm": 3.6082117557525635,
+      "learning_rate": 7.198e-05,
+      "loss": 0.1313,
+      "step": 3600
+    },
+    {
+      "epoch": 0.09658087645138852,
+      "grad_norm": 4.738934516906738,
+      "learning_rate": 7.218e-05,
+      "loss": 0.1509,
+      "step": 3610
+    },
+    {
+      "epoch": 0.09684841350527049,
+      "grad_norm": 1.226082444190979,
+      "learning_rate": 7.238000000000001e-05,
+      "loss": 0.1513,
+      "step": 3620
+    },
+    {
+      "epoch": 0.09711595055915244,
+      "grad_norm": 4.781921863555908,
+      "learning_rate": 7.258e-05,
+      "loss": 0.189,
+      "step": 3630
+    },
+    {
+      "epoch": 0.0973834876130344,
+      "grad_norm": 2.045872688293457,
+      "learning_rate": 7.278e-05,
+      "loss": 0.1196,
+      "step": 3640
+    },
+    {
+      "epoch": 0.09765102466691637,
+      "grad_norm": 2.475872039794922,
+      "learning_rate": 7.298000000000001e-05,
+      "loss": 0.2424,
+      "step": 3650
+    },
+    {
+      "epoch": 0.09791856172079832,
+      "grad_norm": 2.875009059906006,
+      "learning_rate": 7.318e-05,
+      "loss": 0.1266,
+      "step": 3660
+    },
+    {
+      "epoch": 0.09818609877468029,
+      "grad_norm": 2.1114628314971924,
+      "learning_rate": 7.338e-05,
+      "loss": 0.1699,
+      "step": 3670
+    },
+    {
+      "epoch": 0.09845363582856226,
+      "grad_norm": 3.4419660568237305,
+      "learning_rate": 7.358000000000001e-05,
+      "loss": 0.1443,
+      "step": 3680
+    },
+    {
+      "epoch": 0.09872117288244422,
+      "grad_norm": 2.928575038909912,
+      "learning_rate": 7.378e-05,
+      "loss": 0.1543,
+      "step": 3690
+    },
+    {
+      "epoch": 0.09898870993632618,
+      "grad_norm": 3.3291118144989014,
+      "learning_rate": 7.398e-05,
+      "loss": 0.1842,
+      "step": 3700
+    },
+    {
+      "epoch": 0.09925624699020814,
+      "grad_norm": 1.5365654230117798,
+      "learning_rate": 7.418000000000001e-05,
+      "loss": 0.0911,
+      "step": 3710
+    },
+    {
+      "epoch": 0.09952378404409011,
+      "grad_norm": 2.320582151412964,
+      "learning_rate": 7.438e-05,
+      "loss": 0.0917,
+      "step": 3720
+    },
+    {
+      "epoch": 0.09979132109797206,
+      "grad_norm": 4.383909225463867,
+      "learning_rate": 7.458000000000001e-05,
+      "loss": 0.1238,
+      "step": 3730
+    },
+    {
+      "epoch": 0.10005885815185403,
+      "grad_norm": 4.141207695007324,
+      "learning_rate": 7.478e-05,
+      "loss": 0.1364,
+      "step": 3740
+    },
+    {
+      "epoch": 0.100326395205736,
+      "grad_norm": 0.9993315935134888,
+      "learning_rate": 7.498e-05,
+      "loss": 0.1612,
+      "step": 3750
+    },
+    {
+      "epoch": 0.10059393225961796,
+      "grad_norm": 2.4344987869262695,
+      "learning_rate": 7.518000000000001e-05,
+      "loss": 0.1468,
+      "step": 3760
+    },
+    {
+      "epoch": 0.10086146931349992,
+      "grad_norm": 1.858894944190979,
+      "learning_rate": 7.538e-05,
+      "loss": 0.1759,
+      "step": 3770
+    },
+    {
+      "epoch": 0.10112900636738188,
+      "grad_norm": 3.0989270210266113,
+      "learning_rate": 7.558e-05,
+      "loss": 0.2947,
+      "step": 3780
+    },
+    {
+      "epoch": 0.10139654342126385,
+      "grad_norm": 2.3697149753570557,
+      "learning_rate": 7.578000000000001e-05,
+      "loss": 0.1418,
+      "step": 3790
+    },
+    {
+      "epoch": 0.1016640804751458,
+      "grad_norm": 2.451291084289551,
+      "learning_rate": 7.598e-05,
+      "loss": 0.1306,
+      "step": 3800
+    },
+    {
+      "epoch": 0.10193161752902777,
+      "grad_norm": 2.572941303253174,
+      "learning_rate": 7.618e-05,
+      "loss": 0.1597,
+      "step": 3810
+    },
+    {
+      "epoch": 0.10219915458290973,
+      "grad_norm": 6.167232513427734,
+      "learning_rate": 7.638000000000001e-05,
+      "loss": 0.3566,
+      "step": 3820
+    },
+    {
+      "epoch": 0.1024666916367917,
+      "grad_norm": 3.944796085357666,
+      "learning_rate": 7.658e-05,
+      "loss": 0.1691,
+      "step": 3830
+    },
+    {
+      "epoch": 0.10273422869067365,
+      "grad_norm": 2.700469970703125,
+      "learning_rate": 7.678000000000001e-05,
+      "loss": 0.1929,
+      "step": 3840
+    },
+    {
+      "epoch": 0.10300176574455562,
+      "grad_norm": 4.325595378875732,
+      "learning_rate": 7.698000000000001e-05,
+      "loss": 0.217,
+      "step": 3850
+    },
+    {
+      "epoch": 0.10326930279843759,
+      "grad_norm": 1.9190975427627563,
+      "learning_rate": 7.718e-05,
+      "loss": 0.1418,
+      "step": 3860
+    },
+    {
+      "epoch": 0.10353683985231954,
+      "grad_norm": 3.0678229331970215,
+      "learning_rate": 7.738000000000001e-05,
+      "loss": 0.165,
+      "step": 3870
+    },
+    {
+      "epoch": 0.1038043769062015,
+      "grad_norm": 0.7696755528450012,
+      "learning_rate": 7.758000000000001e-05,
+      "loss": 0.106,
+      "step": 3880
+    },
+    {
+      "epoch": 0.10407191396008347,
+      "grad_norm": 2.2603869438171387,
+      "learning_rate": 7.778e-05,
+      "loss": 0.1712,
+      "step": 3890
+    },
+    {
+      "epoch": 0.10433945101396544,
+      "grad_norm": 1.4615178108215332,
+      "learning_rate": 7.798000000000001e-05,
+      "loss": 0.1273,
+      "step": 3900
+    },
+    {
+      "epoch": 0.10460698806784739,
+      "grad_norm": 0.9426511526107788,
+      "learning_rate": 7.818000000000001e-05,
+      "loss": 0.2164,
+      "step": 3910
+    },
+    {
+      "epoch": 0.10487452512172936,
+      "grad_norm": 1.930358648300171,
+      "learning_rate": 7.838e-05,
+      "loss": 0.1068,
+      "step": 3920
+    },
+    {
+      "epoch": 0.10514206217561133,
+      "grad_norm": 3.182863473892212,
+      "learning_rate": 7.858000000000001e-05,
+      "loss": 0.2181,
+      "step": 3930
+    },
+    {
+      "epoch": 0.10540959922949328,
+      "grad_norm": 3.415813446044922,
+      "learning_rate": 7.878e-05,
+      "loss": 0.141,
+      "step": 3940
+    },
+    {
+      "epoch": 0.10567713628337524,
+      "grad_norm": 2.487356185913086,
+      "learning_rate": 7.897999999999999e-05,
+      "loss": 0.0939,
+      "step": 3950
+    },
+    {
+      "epoch": 0.10594467333725721,
+      "grad_norm": 3.0631377696990967,
+      "learning_rate": 7.918e-05,
+      "loss": 0.208,
+      "step": 3960
+    },
+    {
+      "epoch": 0.10621221039113918,
+      "grad_norm": 2.9028663635253906,
+      "learning_rate": 7.938e-05,
+      "loss": 0.1122,
+      "step": 3970
+    },
+    {
+      "epoch": 0.10647974744502113,
+      "grad_norm": 3.122224807739258,
+      "learning_rate": 7.958e-05,
+      "loss": 0.107,
+      "step": 3980
+    },
+    {
+      "epoch": 0.1067472844989031,
+      "grad_norm": 2.823211669921875,
+      "learning_rate": 7.978e-05,
+      "loss": 0.1805,
+      "step": 3990
+    },
+    {
+      "epoch": 0.10701482155278506,
+      "grad_norm": 1.976285457611084,
+      "learning_rate": 7.998e-05,
+      "loss": 0.1434,
+      "step": 4000
+    },
+    {
+      "epoch": 0.10728235860666702,
+      "grad_norm": 2.1536288261413574,
+      "learning_rate": 8.018e-05,
+      "loss": 0.1343,
+      "step": 4010
+    },
+    {
+      "epoch": 0.10754989566054898,
+      "grad_norm": 3.523444652557373,
+      "learning_rate": 8.038e-05,
+      "loss": 0.1774,
+      "step": 4020
+    },
+    {
+      "epoch": 0.10781743271443095,
+      "grad_norm": 1.3530638217926025,
+      "learning_rate": 8.058e-05,
+      "loss": 0.1381,
+      "step": 4030
+    },
+    {
+      "epoch": 0.10808496976831292,
+      "grad_norm": 2.807088613510132,
+      "learning_rate": 8.078e-05,
+      "loss": 0.164,
+      "step": 4040
+    },
+    {
+      "epoch": 0.10835250682219487,
+      "grad_norm": 1.9341635704040527,
+      "learning_rate": 8.098e-05,
+      "loss": 0.1296,
+      "step": 4050
+    },
+    {
+      "epoch": 0.10862004387607684,
+      "grad_norm": 2.8526663780212402,
+      "learning_rate": 8.118e-05,
+      "loss": 0.2234,
+      "step": 4060
+    },
+    {
+      "epoch": 0.1088875809299588,
+      "grad_norm": 2.8787147998809814,
+      "learning_rate": 8.138e-05,
+      "loss": 0.2589,
+      "step": 4070
+    },
+    {
+      "epoch": 0.10915511798384077,
+      "grad_norm": 4.703817844390869,
+      "learning_rate": 8.158e-05,
+      "loss": 0.158,
+      "step": 4080
+    },
+    {
+      "epoch": 0.10942265503772272,
+      "grad_norm": 1.5755696296691895,
+      "learning_rate": 8.178e-05,
+      "loss": 0.1364,
+      "step": 4090
+    },
+    {
+      "epoch": 0.10969019209160469,
+      "grad_norm": 3.4550106525421143,
+      "learning_rate": 8.198e-05,
+      "loss": 0.1414,
+      "step": 4100
+    },
+    {
+      "epoch": 0.10995772914548665,
+      "grad_norm": 1.3362661600112915,
+      "learning_rate": 8.218e-05,
+      "loss": 0.1646,
+      "step": 4110
+    },
+    {
+      "epoch": 0.11022526619936861,
+      "grad_norm": 3.385833263397217,
+      "learning_rate": 8.238000000000001e-05,
+      "loss": 0.1485,
+      "step": 4120
+    },
+    {
+      "epoch": 0.11049280325325057,
+      "grad_norm": 4.89804744720459,
+      "learning_rate": 8.258e-05,
+      "loss": 0.2225,
+      "step": 4130
+    },
+    {
+      "epoch": 0.11076034030713254,
+      "grad_norm": 2.8741190433502197,
+      "learning_rate": 8.278e-05,
+      "loss": 0.1027,
+      "step": 4140
+    },
+    {
+      "epoch": 0.11102787736101451,
+      "grad_norm": 2.650700569152832,
+      "learning_rate": 8.298000000000001e-05,
+      "loss": 0.1899,
+      "step": 4150
+    },
+    {
+      "epoch": 0.11129541441489646,
+      "grad_norm": 1.3462910652160645,
+      "learning_rate": 8.318e-05,
+      "loss": 0.2081,
+      "step": 4160
+    },
+    {
+      "epoch": 0.11156295146877843,
+      "grad_norm": 0.8906551003456116,
+      "learning_rate": 8.338e-05,
+      "loss": 0.2036,
+      "step": 4170
+    },
+    {
+      "epoch": 0.11183048852266039,
+      "grad_norm": 2.3604013919830322,
+      "learning_rate": 8.358e-05,
+      "loss": 0.1264,
+      "step": 4180
+    },
+    {
+      "epoch": 0.11209802557654235,
+      "grad_norm": 2.070692300796509,
+      "learning_rate": 8.378e-05,
+      "loss": 0.1612,
+      "step": 4190
+    },
+    {
+      "epoch": 0.11236556263042431,
+      "grad_norm": 2.6971616744995117,
+      "learning_rate": 8.398e-05,
+      "loss": 0.2902,
+      "step": 4200
+    },
+    {
+      "epoch": 0.11263309968430628,
+      "grad_norm": 1.6872801780700684,
+      "learning_rate": 8.418e-05,
+      "loss": 0.1882,
+      "step": 4210
+    },
+    {
+      "epoch": 0.11290063673818825,
+      "grad_norm": 1.9647997617721558,
+      "learning_rate": 8.438e-05,
+      "loss": 0.1936,
+      "step": 4220
+    },
+    {
+      "epoch": 0.1131681737920702,
+      "grad_norm": 2.1070916652679443,
+      "learning_rate": 8.458e-05,
+      "loss": 0.1403,
+      "step": 4230
+    },
+    {
+      "epoch": 0.11343571084595216,
+      "grad_norm": 4.372876167297363,
+      "learning_rate": 8.478e-05,
+      "loss": 0.1364,
+      "step": 4240
+    },
+    {
+      "epoch": 0.11370324789983413,
+      "grad_norm": 2.14023756980896,
+      "learning_rate": 8.498e-05,
+      "loss": 0.1468,
+      "step": 4250
+    },
+    {
+      "epoch": 0.11397078495371608,
+      "grad_norm": 3.9535181522369385,
+      "learning_rate": 8.518000000000001e-05,
+      "loss": 0.253,
+      "step": 4260
+    },
+    {
+      "epoch": 0.11423832200759805,
+      "grad_norm": 2.7740018367767334,
+      "learning_rate": 8.538e-05,
+      "loss": 0.1944,
+      "step": 4270
+    },
+    {
+      "epoch": 0.11450585906148002,
+      "grad_norm": 2.68082857131958,
+      "learning_rate": 8.558e-05,
+      "loss": 0.2087,
+      "step": 4280
+    },
+    {
+      "epoch": 0.11477339611536198,
+      "grad_norm": 1.1226370334625244,
+      "learning_rate": 8.578000000000001e-05,
+      "loss": 0.0912,
+      "step": 4290
+    },
+    {
+      "epoch": 0.11504093316924394,
+      "grad_norm": 1.156408429145813,
+      "learning_rate": 8.598e-05,
+      "loss": 0.1081,
+      "step": 4300
+    },
+    {
+      "epoch": 0.1153084702231259,
+      "grad_norm": 1.394780158996582,
+      "learning_rate": 8.618e-05,
+      "loss": 0.325,
+      "step": 4310
+    },
+    {
+      "epoch": 0.11557600727700787,
+      "grad_norm": 1.0326213836669922,
+      "learning_rate": 8.638000000000001e-05,
+      "loss": 0.0864,
+      "step": 4320
+    },
+    {
+      "epoch": 0.11584354433088982,
+      "grad_norm": 2.0394978523254395,
+      "learning_rate": 8.658e-05,
+      "loss": 0.1512,
+      "step": 4330
+    },
+    {
+      "epoch": 0.11611108138477179,
+      "grad_norm": 2.894214630126953,
+      "learning_rate": 8.678e-05,
+      "loss": 0.2289,
+      "step": 4340
+    },
+    {
+      "epoch": 0.11637861843865376,
+      "grad_norm": 2.2017617225646973,
+      "learning_rate": 8.698000000000001e-05,
+      "loss": 0.1207,
+      "step": 4350
+    },
+    {
+      "epoch": 0.11664615549253572,
+      "grad_norm": 2.0121042728424072,
+      "learning_rate": 8.718e-05,
+      "loss": 0.3134,
+      "step": 4360
+    },
+    {
+      "epoch": 0.11691369254641767,
+      "grad_norm": 4.714865684509277,
+      "learning_rate": 8.738000000000001e-05,
+      "loss": 0.2362,
+      "step": 4370
+    },
+    {
+      "epoch": 0.11718122960029964,
+      "grad_norm": 2.4241867065429688,
+      "learning_rate": 8.758000000000001e-05,
+      "loss": 0.1983,
+      "step": 4380
+    },
+    {
+      "epoch": 0.11744876665418161,
+      "grad_norm": 4.4691619873046875,
+      "learning_rate": 8.778e-05,
+      "loss": 0.1294,
+      "step": 4390
+    },
+    {
+      "epoch": 0.11771630370806356,
+      "grad_norm": 1.3408023118972778,
+      "learning_rate": 8.798000000000001e-05,
+      "loss": 0.1846,
+      "step": 4400
+    },
+    {
+      "epoch": 0.11798384076194553,
+      "grad_norm": 1.9288387298583984,
+      "learning_rate": 8.818000000000001e-05,
+      "loss": 0.2097,
+      "step": 4410
+    },
+    {
+      "epoch": 0.1182513778158275,
+      "grad_norm": 3.456092119216919,
+      "learning_rate": 8.838e-05,
+      "loss": 0.2219,
+      "step": 4420
+    },
+    {
+      "epoch": 0.11851891486970946,
+      "grad_norm": 1.2042790651321411,
+      "learning_rate": 8.858000000000001e-05,
+      "loss": 0.2569,
+      "step": 4430
+    },
+    {
+      "epoch": 0.11878645192359141,
+      "grad_norm": 3.0921735763549805,
+      "learning_rate": 8.878000000000001e-05,
+      "loss": 0.2574,
+      "step": 4440
+    },
+    {
+      "epoch": 0.11905398897747338,
+      "grad_norm": 2.4361231327056885,
+      "learning_rate": 8.898e-05,
+      "loss": 0.287,
+      "step": 4450
+    },
+    {
+      "epoch": 0.11932152603135535,
+      "grad_norm": 2.4821317195892334,
+      "learning_rate": 8.918000000000001e-05,
+      "loss": 0.1532,
+      "step": 4460
+    },
+    {
+      "epoch": 0.1195890630852373,
+      "grad_norm": 2.5308728218078613,
+      "learning_rate": 8.938e-05,
+      "loss": 0.235,
+      "step": 4470
+    },
+    {
+      "epoch": 0.11985660013911927,
+      "grad_norm": 2.5199458599090576,
+      "learning_rate": 8.958e-05,
+      "loss": 0.1243,
+      "step": 4480
+    },
+    {
+      "epoch": 0.12012413719300123,
+      "grad_norm": 1.007845401763916,
+      "learning_rate": 8.978000000000001e-05,
+      "loss": 0.1379,
+      "step": 4490
+    },
+    {
+      "epoch": 0.1203916742468832,
+      "grad_norm": 1.4686565399169922,
+      "learning_rate": 8.998e-05,
+      "loss": 0.2431,
+      "step": 4500
+    },
+    {
+      "epoch": 0.12065921130076515,
+      "grad_norm": 3.999821424484253,
+      "learning_rate": 9.018000000000001e-05,
+      "loss": 0.199,
+      "step": 4510
+    },
+    {
+      "epoch": 0.12092674835464712,
+      "grad_norm": 3.2069876194000244,
+      "learning_rate": 9.038000000000001e-05,
+      "loss": 0.1178,
+      "step": 4520
+    },
+    {
+      "epoch": 0.12119428540852908,
+      "grad_norm": 2.92010498046875,
+      "learning_rate": 9.058e-05,
+      "loss": 0.1554,
+      "step": 4530
+    },
+    {
+      "epoch": 0.12146182246241104,
+      "grad_norm": 4.208976745605469,
+      "learning_rate": 9.078000000000001e-05,
+      "loss": 0.1459,
+      "step": 4540
+    },
+    {
+      "epoch": 0.121729359516293,
+      "grad_norm": 6.488468170166016,
+      "learning_rate": 9.098000000000001e-05,
+      "loss": 0.2168,
+      "step": 4550
+    },
+    {
+      "epoch": 0.12199689657017497,
+      "grad_norm": 1.4069709777832031,
+      "learning_rate": 9.118e-05,
+      "loss": 0.1663,
+      "step": 4560
+    },
+    {
+      "epoch": 0.12226443362405694,
+      "grad_norm": 1.4715473651885986,
+      "learning_rate": 9.138e-05,
+      "loss": 0.131,
+      "step": 4570
+    },
+    {
+      "epoch": 0.12253197067793889,
+      "grad_norm": 3.105782985687256,
+      "learning_rate": 9.158e-05,
+      "loss": 0.1781,
+      "step": 4580
+    },
+    {
+      "epoch": 0.12279950773182086,
+      "grad_norm": 1.352236032485962,
+      "learning_rate": 9.178e-05,
+      "loss": 0.1601,
+      "step": 4590
+    },
+    {
+      "epoch": 0.12306704478570282,
+      "grad_norm": 1.6743556261062622,
+      "learning_rate": 9.198e-05,
+      "loss": 0.0842,
+      "step": 4600
+    },
+    {
+      "epoch": 0.12333458183958479,
+      "grad_norm": 2.0349812507629395,
+      "learning_rate": 9.218e-05,
+      "loss": 0.1416,
+      "step": 4610
+    },
+    {
+      "epoch": 0.12360211889346674,
+      "grad_norm": 4.281164169311523,
+      "learning_rate": 9.238e-05,
+      "loss": 0.1621,
+      "step": 4620
+    },
+    {
+      "epoch": 0.12386965594734871,
+      "grad_norm": 3.861701488494873,
+      "learning_rate": 9.258e-05,
+      "loss": 0.2104,
+      "step": 4630
+    },
+    {
+      "epoch": 0.12413719300123068,
+      "grad_norm": 1.3111672401428223,
+      "learning_rate": 9.278e-05,
+      "loss": 0.1447,
+      "step": 4640
+    },
+    {
+      "epoch": 0.12440473005511263,
+      "grad_norm": 2.911105155944824,
+      "learning_rate": 9.298e-05,
+      "loss": 0.1732,
+      "step": 4650
+    },
+    {
+      "epoch": 0.1246722671089946,
+      "grad_norm": 1.139471411705017,
+      "learning_rate": 9.318e-05,
+      "loss": 0.1466,
+      "step": 4660
+    },
+    {
+      "epoch": 0.12493980416287656,
+      "grad_norm": 4.321771621704102,
+      "learning_rate": 9.338e-05,
+      "loss": 0.1769,
+      "step": 4670
+    },
+    {
+      "epoch": 0.12520734121675853,
+      "grad_norm": 2.669766426086426,
+      "learning_rate": 9.358e-05,
+      "loss": 0.1842,
+      "step": 4680
+    },
+    {
+      "epoch": 0.1254748782706405,
+      "grad_norm": 1.4404703378677368,
+      "learning_rate": 9.378e-05,
+      "loss": 0.1126,
+      "step": 4690
+    },
+    {
+      "epoch": 0.12574241532452243,
+      "grad_norm": 1.628858208656311,
+      "learning_rate": 9.398e-05,
+      "loss": 0.186,
+      "step": 4700
+    },
+    {
+      "epoch": 0.1260099523784044,
+      "grad_norm": 2.5880675315856934,
+      "learning_rate": 9.418e-05,
+      "loss": 0.1263,
+      "step": 4710
+    },
+    {
+      "epoch": 0.12627748943228637,
+      "grad_norm": 2.4773612022399902,
+      "learning_rate": 9.438e-05,
+      "loss": 0.1673,
+      "step": 4720
+    },
+    {
+      "epoch": 0.12654502648616833,
+      "grad_norm": 1.3746280670166016,
+      "learning_rate": 9.458e-05,
+      "loss": 0.1377,
+      "step": 4730
+    },
+    {
+      "epoch": 0.1268125635400503,
+      "grad_norm": 3.6017401218414307,
+      "learning_rate": 9.478e-05,
+      "loss": 0.3163,
+      "step": 4740
+    },
+    {
+      "epoch": 0.12708010059393227,
+      "grad_norm": 11.797868728637695,
+      "learning_rate": 9.498e-05,
+      "loss": 0.3201,
+      "step": 4750
+    },
+    {
+      "epoch": 0.12734763764781423,
+      "grad_norm": 2.2882864475250244,
+      "learning_rate": 9.518000000000001e-05,
+      "loss": 0.1577,
+      "step": 4760
+    },
+    {
+      "epoch": 0.12761517470169617,
+      "grad_norm": 3.5163676738739014,
+      "learning_rate": 9.538e-05,
+      "loss": 0.233,
+      "step": 4770
+    },
+    {
+      "epoch": 0.12788271175557814,
+      "grad_norm": 2.9472875595092773,
+      "learning_rate": 9.558e-05,
+      "loss": 0.2279,
+      "step": 4780
+    },
+    {
+      "epoch": 0.1281502488094601,
+      "grad_norm": 2.0604071617126465,
+      "learning_rate": 9.578000000000001e-05,
+      "loss": 0.1901,
+      "step": 4790
+    },
+    {
+      "epoch": 0.12841778586334207,
+      "grad_norm": 1.970211148262024,
+      "learning_rate": 9.598e-05,
+      "loss": 0.2273,
+      "step": 4800
+    },
+    {
+      "epoch": 0.12868532291722404,
+      "grad_norm": 2.7023608684539795,
+      "learning_rate": 9.618e-05,
+      "loss": 0.1602,
+      "step": 4810
+    },
+    {
+      "epoch": 0.128952859971106,
+      "grad_norm": 2.152268171310425,
+      "learning_rate": 9.638000000000001e-05,
+      "loss": 0.1757,
+      "step": 4820
+    },
+    {
+      "epoch": 0.12922039702498797,
+      "grad_norm": 1.715922236442566,
+      "learning_rate": 9.658e-05,
+      "loss": 0.1632,
+      "step": 4830
+    },
+    {
+      "epoch": 0.12948793407886994,
+      "grad_norm": 1.7317090034484863,
+      "learning_rate": 9.678e-05,
+      "loss": 0.1411,
+      "step": 4840
+    },
+    {
+      "epoch": 0.12975547113275188,
+      "grad_norm": 2.4071249961853027,
+      "learning_rate": 9.698000000000001e-05,
+      "loss": 0.1364,
+      "step": 4850
+    },
+    {
+      "epoch": 0.13002300818663384,
+      "grad_norm": 1.5090610980987549,
+      "learning_rate": 9.718e-05,
+      "loss": 0.1244,
+      "step": 4860
+    },
+    {
+      "epoch": 0.1302905452405158,
+      "grad_norm": 3.5730698108673096,
+      "learning_rate": 9.738e-05,
+      "loss": 0.1873,
+      "step": 4870
+    },
+    {
+      "epoch": 0.13055808229439778,
+      "grad_norm": 1.974471926689148,
+      "learning_rate": 9.758000000000001e-05,
+      "loss": 0.1811,
+      "step": 4880
+    },
+    {
+      "epoch": 0.13082561934827974,
+      "grad_norm": 2.309720516204834,
+      "learning_rate": 9.778e-05,
+      "loss": 0.1591,
+      "step": 4890
+    },
+    {
+      "epoch": 0.1310931564021617,
+      "grad_norm": 2.7792000770568848,
+      "learning_rate": 9.798000000000001e-05,
+      "loss": 0.1733,
+      "step": 4900
+    },
+    {
+      "epoch": 0.13136069345604368,
+      "grad_norm": 3.2748286724090576,
+      "learning_rate": 9.818000000000001e-05,
+      "loss": 0.1417,
+      "step": 4910
+    },
+    {
+      "epoch": 0.13162823050992561,
+      "grad_norm": 3.990873336791992,
+      "learning_rate": 9.838e-05,
+      "loss": 0.2102,
+      "step": 4920
+    },
+    {
+      "epoch": 0.13189576756380758,
+      "grad_norm": 0.9762261509895325,
+      "learning_rate": 9.858000000000001e-05,
+      "loss": 0.1072,
+      "step": 4930
+    },
+    {
+      "epoch": 0.13216330461768955,
+      "grad_norm": 1.8507840633392334,
+      "learning_rate": 9.878e-05,
+      "loss": 0.1542,
+      "step": 4940
+    },
+    {
+      "epoch": 0.13243084167157151,
+      "grad_norm": 3.078033685684204,
+      "learning_rate": 9.898e-05,
+      "loss": 0.1069,
+      "step": 4950
+    },
+    {
+      "epoch": 0.13269837872545348,
+      "grad_norm": 1.834459900856018,
+      "learning_rate": 9.918000000000001e-05,
+      "loss": 0.1747,
+      "step": 4960
+    },
+    {
+      "epoch": 0.13296591577933545,
+      "grad_norm": 1.2193723917007446,
+      "learning_rate": 9.938e-05,
+      "loss": 0.1624,
+      "step": 4970
+    },
+    {
+      "epoch": 0.13323345283321741,
+      "grad_norm": 2.8759982585906982,
+      "learning_rate": 9.958e-05,
+      "loss": 0.1364,
+      "step": 4980
+    },
+    {
+      "epoch": 0.13350098988709935,
+      "grad_norm": 3.7162201404571533,
+      "learning_rate": 9.978000000000001e-05,
+      "loss": 0.1077,
+      "step": 4990
+    },
+    {
+      "epoch": 0.13376852694098132,
+      "grad_norm": 2.0040194988250732,
+      "learning_rate": 9.998e-05,
+      "loss": 0.1892,
+      "step": 5000
+    },
+    {
+      "epoch": 0.13403606399486329,
+      "grad_norm": 2.0531482696533203,
+      "learning_rate": 9.999999778549045e-05,
+      "loss": 0.1051,
+      "step": 5010
+    },
+    {
+      "epoch": 0.13430360104874525,
+      "grad_norm": 3.5830023288726807,
+      "learning_rate": 9.999999013039593e-05,
+      "loss": 0.2221,
+      "step": 5020
+    },
+    {
+      "epoch": 0.13457113810262722,
+      "grad_norm": 2.4252774715423584,
+      "learning_rate": 9.999997700737766e-05,
+      "loss": 0.1668,
+      "step": 5030
+    },
+    {
+      "epoch": 0.13483867515650919,
+      "grad_norm": 4.013788223266602,
+      "learning_rate": 9.999995841643709e-05,
+      "loss": 0.1935,
+      "step": 5040
+    },
+    {
+      "epoch": 0.13510621221039115,
+      "grad_norm": 2.3152918815612793,
+      "learning_rate": 9.999993435757623e-05,
+      "loss": 0.1444,
+      "step": 5050
+    },
+    {
+      "epoch": 0.1353737492642731,
+      "grad_norm": 2.018033742904663,
+      "learning_rate": 9.999990483079773e-05,
+      "loss": 0.2242,
+      "step": 5060
+    },
+    {
+      "epoch": 0.13564128631815506,
+      "grad_norm": 4.115313529968262,
+      "learning_rate": 9.999986983610481e-05,
+      "loss": 0.2535,
+      "step": 5070
+    },
+    {
+      "epoch": 0.13590882337203702,
+      "grad_norm": 1.875641942024231,
+      "learning_rate": 9.99998293735013e-05,
+      "loss": 0.2741,
+      "step": 5080
+    },
+    {
+      "epoch": 0.136176360425919,
+      "grad_norm": 1.2649637460708618,
+      "learning_rate": 9.999978344299161e-05,
+      "loss": 0.1216,
+      "step": 5090
+    },
+    {
+      "epoch": 0.13644389747980096,
+      "grad_norm": 2.053818702697754,
+      "learning_rate": 9.99997320445808e-05,
+      "loss": 0.1372,
+      "step": 5100
+    },
+    {
+      "epoch": 0.13671143453368292,
+      "grad_norm": 1.8366016149520874,
+      "learning_rate": 9.999967517827444e-05,
+      "loss": 0.1847,
+      "step": 5110
+    },
+    {
+      "epoch": 0.1369789715875649,
+      "grad_norm": 2.342142343521118,
+      "learning_rate": 9.999961284407879e-05,
+      "loss": 0.1719,
+      "step": 5120
+    },
+    {
+      "epoch": 0.13724650864144683,
+      "grad_norm": 2.8020803928375244,
+      "learning_rate": 9.999954504200067e-05,
+      "loss": 0.1743,
+      "step": 5130
+    },
+    {
+      "epoch": 0.1375140456953288,
+      "grad_norm": 3.340163469314575,
+      "learning_rate": 9.999947177204744e-05,
+      "loss": 0.1742,
+      "step": 5140
+    },
+    {
+      "epoch": 0.13778158274921076,
+      "grad_norm": 2.347142219543457,
+      "learning_rate": 9.999939303422718e-05,
+      "loss": 0.1377,
+      "step": 5150
+    },
+    {
+      "epoch": 0.13804911980309273,
+      "grad_norm": 3.386178493499756,
+      "learning_rate": 9.999930882854847e-05,
+      "loss": 0.1877,
+      "step": 5160
+    },
+    {
+      "epoch": 0.1383166568569747,
+      "grad_norm": 3.4735794067382812,
+      "learning_rate": 9.999921915502051e-05,
+      "loss": 0.1638,
+      "step": 5170
+    },
+    {
+      "epoch": 0.13858419391085666,
+      "grad_norm": 2.099531650543213,
+      "learning_rate": 9.99991240136531e-05,
+      "loss": 0.2111,
+      "step": 5180
+    },
+    {
+      "epoch": 0.13885173096473863,
+      "grad_norm": 2.795299530029297,
+      "learning_rate": 9.999902340445668e-05,
+      "loss": 0.1546,
+      "step": 5190
+    },
+    {
+      "epoch": 0.13911926801862057,
+      "grad_norm": 4.0125932693481445,
+      "learning_rate": 9.999891732744224e-05,
+      "loss": 0.1673,
+      "step": 5200
+    },
+    {
+      "epoch": 0.13938680507250253,
+      "grad_norm": 0.9259185194969177,
+      "learning_rate": 9.999880578262135e-05,
+      "loss": 0.147,
+      "step": 5210
+    },
+    {
+      "epoch": 0.1396543421263845,
+      "grad_norm": 1.5301170349121094,
+      "learning_rate": 9.999868877000624e-05,
+      "loss": 0.2476,
+      "step": 5220
+    },
+    {
+      "epoch": 0.13992187918026647,
+      "grad_norm": 2.6077611446380615,
+      "learning_rate": 9.99985662896097e-05,
+      "loss": 0.1286,
+      "step": 5230
+    },
+    {
+      "epoch": 0.14018941623414843,
+      "grad_norm": 3.602152109146118,
+      "learning_rate": 9.999843834144513e-05,
+      "loss": 0.1783,
+      "step": 5240
+    },
+    {
+      "epoch": 0.1404569532880304,
+      "grad_norm": 3.170705556869507,
+      "learning_rate": 9.99983049255265e-05,
+      "loss": 0.2678,
+      "step": 5250
+    },
+    {
+      "epoch": 0.14072449034191237,
+      "grad_norm": 2.108248472213745,
+      "learning_rate": 9.999816604186843e-05,
+      "loss": 0.2215,
+      "step": 5260
+    },
+    {
+      "epoch": 0.1409920273957943,
+      "grad_norm": 1.956078290939331,
+      "learning_rate": 9.999802169048609e-05,
+      "loss": 0.1097,
+      "step": 5270
+    },
+    {
+      "epoch": 0.14125956444967627,
+      "grad_norm": 2.360588788986206,
+      "learning_rate": 9.999787187139527e-05,
+      "loss": 0.19,
+      "step": 5280
+    },
+    {
+      "epoch": 0.14152710150355824,
+      "grad_norm": 1.4083774089813232,
+      "learning_rate": 9.999771658461234e-05,
+      "loss": 0.1885,
+      "step": 5290
+    },
+    {
+      "epoch": 0.1417946385574402,
+      "grad_norm": 1.6136811971664429,
+      "learning_rate": 9.999755583015431e-05,
+      "loss": 0.1234,
+      "step": 5300
+    },
+    {
+      "epoch": 0.14206217561132217,
+      "grad_norm": 3.5447802543640137,
+      "learning_rate": 9.999738960803874e-05,
+      "loss": 0.1398,
+      "step": 5310
+    },
+    {
+      "epoch": 0.14232971266520414,
+      "grad_norm": 3.7510509490966797,
+      "learning_rate": 9.99972179182838e-05,
+      "loss": 0.1812,
+      "step": 5320
+    },
+    {
+      "epoch": 0.1425972497190861,
+      "grad_norm": 4.443079948425293,
+      "learning_rate": 9.99970407609083e-05,
+      "loss": 0.1348,
+      "step": 5330
+    },
+    {
+      "epoch": 0.14286478677296804,
+      "grad_norm": 2.5534844398498535,
+      "learning_rate": 9.999685813593159e-05,
+      "loss": 0.1038,
+      "step": 5340
+    },
+    {
+      "epoch": 0.14313232382685,
+      "grad_norm": 2.7219581604003906,
+      "learning_rate": 9.999667004337362e-05,
+      "loss": 0.1483,
+      "step": 5350
+    },
+    {
+      "epoch": 0.14339986088073198,
+      "grad_norm": 1.1570942401885986,
+      "learning_rate": 9.9996476483255e-05,
+      "loss": 0.1443,
+      "step": 5360
+    },
+    {
+      "epoch": 0.14366739793461394,
+      "grad_norm": 1.465549111366272,
+      "learning_rate": 9.999627745559688e-05,
+      "loss": 0.1159,
+      "step": 5370
+    },
+    {
+      "epoch": 0.1439349349884959,
+      "grad_norm": 2.2212634086608887,
+      "learning_rate": 9.999607296042101e-05,
+      "loss": 0.1579,
+      "step": 5380
+    },
+    {
+      "epoch": 0.14420247204237788,
+      "grad_norm": 1.4482470750808716,
+      "learning_rate": 9.99958629977498e-05,
+      "loss": 0.1695,
+      "step": 5390
+    },
+    {
+      "epoch": 0.14447000909625984,
+      "grad_norm": 1.292111873626709,
+      "learning_rate": 9.999564756760615e-05,
+      "loss": 0.0729,
+      "step": 5400
+    },
+    {
+      "epoch": 0.14473754615014178,
+      "grad_norm": 1.5690926313400269,
+      "learning_rate": 9.999542667001366e-05,
+      "loss": 0.278,
+      "step": 5410
+    },
+    {
+      "epoch": 0.14500508320402375,
+      "grad_norm": 4.719433784484863,
+      "learning_rate": 9.999520030499647e-05,
+      "loss": 0.1865,
+      "step": 5420
+    },
+    {
+      "epoch": 0.14527262025790572,
+      "grad_norm": 1.538076400756836,
+      "learning_rate": 9.999496847257936e-05,
+      "loss": 0.132,
+      "step": 5430
+    },
+    {
+      "epoch": 0.14554015731178768,
+      "grad_norm": 3.731153726577759,
+      "learning_rate": 9.999473117278764e-05,
+      "loss": 0.1999,
+      "step": 5440
+    },
+    {
+      "epoch": 0.14580769436566965,
+      "grad_norm": 4.914045810699463,
+      "learning_rate": 9.999448840564731e-05,
+      "loss": 0.194,
+      "step": 5450
+    },
+    {
+      "epoch": 0.14607523141955162,
+      "grad_norm": 4.517213344573975,
+      "learning_rate": 9.999424017118488e-05,
+      "loss": 0.1391,
+      "step": 5460
+    },
+    {
+      "epoch": 0.14634276847343358,
+      "grad_norm": 1.270505666732788,
+      "learning_rate": 9.999398646942751e-05,
+      "loss": 0.1162,
+      "step": 5470
+    },
+    {
+      "epoch": 0.14661030552731552,
+      "grad_norm": 1.648106575012207,
+      "learning_rate": 9.999372730040296e-05,
+      "loss": 0.0972,
+      "step": 5480
+    },
+    {
+      "epoch": 0.1468778425811975,
+      "grad_norm": 1.0745735168457031,
+      "learning_rate": 9.999346266413953e-05,
+      "loss": 0.163,
+      "step": 5490
+    },
+    {
+      "epoch": 0.14714537963507945,
+      "grad_norm": 2.679692506790161,
+      "learning_rate": 9.99931925606662e-05,
+      "loss": 0.1738,
+      "step": 5500
+    },
+    {
+      "epoch": 0.14741291668896142,
+      "grad_norm": 2.4323172569274902,
+      "learning_rate": 9.99929169900125e-05,
+      "loss": 0.2223,
+      "step": 5510
+    },
+    {
+      "epoch": 0.1476804537428434,
+      "grad_norm": 3.3761966228485107,
+      "learning_rate": 9.999263595220855e-05,
+      "loss": 0.1729,
+      "step": 5520
+    },
+    {
+      "epoch": 0.14794799079672535,
+      "grad_norm": 2.320732593536377,
+      "learning_rate": 9.99923494472851e-05,
+      "loss": 0.1642,
+      "step": 5530
+    },
+    {
+      "epoch": 0.14821552785060732,
+      "grad_norm": 1.310529351234436,
+      "learning_rate": 9.999205747527348e-05,
+      "loss": 0.1301,
+      "step": 5540
+    },
+    {
+      "epoch": 0.14848306490448926,
+      "grad_norm": 2.591639518737793,
+      "learning_rate": 9.999176003620561e-05,
+      "loss": 0.1894,
+      "step": 5550
+    },
+    {
+      "epoch": 0.14875060195837123,
+      "grad_norm": 1.3016331195831299,
+      "learning_rate": 9.999145713011405e-05,
+      "loss": 0.1596,
+      "step": 5560
+    },
+    {
+      "epoch": 0.1490181390122532,
+      "grad_norm": 3.2038493156433105,
+      "learning_rate": 9.999114875703186e-05,
+      "loss": 0.3051,
+      "step": 5570
+    },
+    {
+      "epoch": 0.14928567606613516,
+      "grad_norm": 3.315134048461914,
+      "learning_rate": 9.999083491699281e-05,
+      "loss": 0.1885,
+      "step": 5580
+    },
+    {
+      "epoch": 0.14955321312001713,
+      "grad_norm": 2.2435102462768555,
+      "learning_rate": 9.999051561003123e-05,
+      "loss": 0.1455,
+      "step": 5590
+    },
+    {
+      "epoch": 0.1498207501738991,
+      "grad_norm": 2.442758798599243,
+      "learning_rate": 9.999019083618202e-05,
+      "loss": 0.113,
+      "step": 5600
+    },
+    {
+      "epoch": 0.15008828722778106,
+      "grad_norm": 1.9697140455245972,
+      "learning_rate": 9.99898605954807e-05,
+      "loss": 0.1323,
+      "step": 5610
+    },
+    {
+      "epoch": 0.150355824281663,
+      "grad_norm": 2.1731460094451904,
+      "learning_rate": 9.998952488796338e-05,
+      "loss": 0.1123,
+      "step": 5620
+    },
+    {
+      "epoch": 0.15062336133554496,
+      "grad_norm": 2.5453693866729736,
+      "learning_rate": 9.998918371366676e-05,
+      "loss": 0.2346,
+      "step": 5630
+    },
+    {
+      "epoch": 0.15089089838942693,
+      "grad_norm": 2.107804298400879,
+      "learning_rate": 9.99888370726282e-05,
+      "loss": 0.1735,
+      "step": 5640
+    },
+    {
+      "epoch": 0.1511584354433089,
+      "grad_norm": 1.4639204740524292,
+      "learning_rate": 9.998848496488556e-05,
+      "loss": 0.1123,
+      "step": 5650
+    },
+    {
+      "epoch": 0.15142597249719086,
+      "grad_norm": 1.343528151512146,
+      "learning_rate": 9.998812739047736e-05,
+      "loss": 0.1733,
+      "step": 5660
+    },
+    {
+      "epoch": 0.15169350955107283,
+      "grad_norm": 3.0273730754852295,
+      "learning_rate": 9.99877643494427e-05,
+      "loss": 0.2701,
+      "step": 5670
+    },
+    {
+      "epoch": 0.1519610466049548,
+      "grad_norm": 2.7359254360198975,
+      "learning_rate": 9.998739584182128e-05,
+      "loss": 0.2504,
+      "step": 5680
+    },
+    {
+      "epoch": 0.15222858365883674,
+      "grad_norm": 2.0538694858551025,
+      "learning_rate": 9.998702186765342e-05,
+      "loss": 0.171,
+      "step": 5690
+    },
+    {
+      "epoch": 0.1524961207127187,
+      "grad_norm": 1.2736024856567383,
+      "learning_rate": 9.998664242698e-05,
+      "loss": 0.1766,
+      "step": 5700
+    },
+    {
+      "epoch": 0.15276365776660067,
+      "grad_norm": 1.9293681383132935,
+      "learning_rate": 9.998625751984251e-05,
+      "loss": 0.0985,
+      "step": 5710
+    },
+    {
+      "epoch": 0.15303119482048264,
+      "grad_norm": 3.1050310134887695,
+      "learning_rate": 9.998586714628307e-05,
+      "loss": 0.1715,
+      "step": 5720
+    },
+    {
+      "epoch": 0.1532987318743646,
+      "grad_norm": 1.673852801322937,
+      "learning_rate": 9.998547130634432e-05,
+      "loss": 0.1029,
+      "step": 5730
+    },
+    {
+      "epoch": 0.15356626892824657,
+      "grad_norm": 2.09529447555542,
+      "learning_rate": 9.99850700000696e-05,
+      "loss": 0.1264,
+      "step": 5740
+    },
+    {
+      "epoch": 0.15383380598212854,
+      "grad_norm": 2.9502627849578857,
+      "learning_rate": 9.998466322750278e-05,
+      "loss": 0.1756,
+      "step": 5750
+    },
+    {
+      "epoch": 0.15410134303601047,
+      "grad_norm": 1.5892301797866821,
+      "learning_rate": 9.998425098868834e-05,
+      "loss": 0.1649,
+      "step": 5760
+    },
+    {
+      "epoch": 0.15436888008989244,
+      "grad_norm": 5.697965145111084,
+      "learning_rate": 9.998383328367136e-05,
+      "loss": 0.1885,
+      "step": 5770
+    },
+    {
+      "epoch": 0.1546364171437744,
+      "grad_norm": 2.209225654602051,
+      "learning_rate": 9.99834101124975e-05,
+      "loss": 0.1116,
+      "step": 5780
+    },
+    {
+      "epoch": 0.15490395419765637,
+      "grad_norm": 2.075218915939331,
+      "learning_rate": 9.998298147521309e-05,
+      "loss": 0.0822,
+      "step": 5790
+    },
+    {
+      "epoch": 0.15517149125153834,
+      "grad_norm": 2.243316411972046,
+      "learning_rate": 9.998254737186496e-05,
+      "loss": 0.0965,
+      "step": 5800
+    },
+    {
+      "epoch": 0.1554390283054203,
+      "grad_norm": 1.544592261314392,
+      "learning_rate": 9.99821078025006e-05,
+      "loss": 0.318,
+      "step": 5810
+    },
+    {
+      "epoch": 0.15570656535930227,
+      "grad_norm": 2.7073616981506348,
+      "learning_rate": 9.998166276716807e-05,
+      "loss": 0.105,
+      "step": 5820
+    },
+    {
+      "epoch": 0.1559741024131842,
+      "grad_norm": 2.192981243133545,
+      "learning_rate": 9.998121226591606e-05,
+      "loss": 0.1588,
+      "step": 5830
+    },
+    {
+      "epoch": 0.15624163946706618,
+      "grad_norm": 2.3334310054779053,
+      "learning_rate": 9.998075629879382e-05,
+      "loss": 0.0896,
+      "step": 5840
+    },
+    {
+      "epoch": 0.15650917652094815,
+      "grad_norm": 1.3266490697860718,
+      "learning_rate": 9.99802948658512e-05,
+      "loss": 0.074,
+      "step": 5850
+    },
+    {
+      "epoch": 0.1567767135748301,
+      "grad_norm": 1.9787209033966064,
+      "learning_rate": 9.99798279671387e-05,
+      "loss": 0.0733,
+      "step": 5860
+    },
+    {
+      "epoch": 0.15704425062871208,
+      "grad_norm": 1.947495460510254,
+      "learning_rate": 9.997935560270734e-05,
+      "loss": 0.1607,
+      "step": 5870
+    },
+    {
+      "epoch": 0.15731178768259405,
+      "grad_norm": 1.7948633432388306,
+      "learning_rate": 9.997887777260879e-05,
+      "loss": 0.1143,
+      "step": 5880
+    },
+    {
+      "epoch": 0.157579324736476,
+      "grad_norm": 0.9118947982788086,
+      "learning_rate": 9.997839447689532e-05,
+      "loss": 0.1137,
+      "step": 5890
+    },
+    {
+      "epoch": 0.15784686179035795,
+      "grad_norm": 1.8234741687774658,
+      "learning_rate": 9.997790571561978e-05,
+      "loss": 0.1139,
+      "step": 5900
+    },
+    {
+      "epoch": 0.15811439884423992,
+      "grad_norm": 2.655151128768921,
+      "learning_rate": 9.99774114888356e-05,
+      "loss": 0.1097,
+      "step": 5910
+    },
+    {
+      "epoch": 0.15838193589812188,
+      "grad_norm": 2.8651068210601807,
+      "learning_rate": 9.997691179659684e-05,
+      "loss": 0.1479,
+      "step": 5920
+    },
+    {
+      "epoch": 0.15864947295200385,
+      "grad_norm": 2.1915700435638428,
+      "learning_rate": 9.997640663895815e-05,
+      "loss": 0.2111,
+      "step": 5930
+    },
+    {
+      "epoch": 0.15891701000588582,
+      "grad_norm": 1.545310378074646,
+      "learning_rate": 9.997589601597477e-05,
+      "loss": 0.2479,
+      "step": 5940
+    },
+    {
+      "epoch": 0.15918454705976778,
+      "grad_norm": 2.5394134521484375,
+      "learning_rate": 9.997537992770252e-05,
+      "loss": 0.1552,
+      "step": 5950
+    },
+    {
+      "epoch": 0.15945208411364975,
+      "grad_norm": 1.4568824768066406,
+      "learning_rate": 9.997485837419788e-05,
+      "loss": 0.1104,
+      "step": 5960
+    },
+    {
+      "epoch": 0.15971962116753172,
+      "grad_norm": 1.9367064237594604,
+      "learning_rate": 9.997433135551786e-05,
+      "loss": 0.2204,
+      "step": 5970
+    },
+    {
+      "epoch": 0.15998715822141366,
+      "grad_norm": 2.2885844707489014,
+      "learning_rate": 9.997379887172009e-05,
+      "loss": 0.098,
+      "step": 5980
+    },
+    {
+      "epoch": 0.16025469527529562,
+      "grad_norm": 1.3138468265533447,
+      "learning_rate": 9.997326092286281e-05,
+      "loss": 0.1025,
+      "step": 5990
+    },
+    {
+      "epoch": 0.1605222323291776,
+      "grad_norm": 1.3327690362930298,
+      "learning_rate": 9.997271750900486e-05,
+      "loss": 0.1147,
+      "step": 6000
+    },
+    {
+      "epoch": 0.16078976938305956,
+      "grad_norm": 1.8419421911239624,
+      "learning_rate": 9.997216863020565e-05,
+      "loss": 0.165,
+      "step": 6010
+    },
+    {
+      "epoch": 0.16105730643694152,
+      "grad_norm": 5.050527095794678,
+      "learning_rate": 9.99716142865252e-05,
+      "loss": 0.1891,
+      "step": 6020
+    },
+    {
+      "epoch": 0.1613248434908235,
+      "grad_norm": 1.0112247467041016,
+      "learning_rate": 9.997105447802415e-05,
+      "loss": 0.1051,
+      "step": 6030
+    },
+    {
+      "epoch": 0.16159238054470546,
+      "grad_norm": 0.9817419052124023,
+      "learning_rate": 9.997048920476373e-05,
+      "loss": 0.1432,
+      "step": 6040
+    },
+    {
+      "epoch": 0.1618599175985874,
+      "grad_norm": 2.5745348930358887,
+      "learning_rate": 9.996991846680572e-05,
+      "loss": 0.1094,
+      "step": 6050
+    },
+    {
+      "epoch": 0.16212745465246936,
+      "grad_norm": 3.707491874694824,
+      "learning_rate": 9.996934226421257e-05,
+      "loss": 0.2097,
+      "step": 6060
+    },
+    {
+      "epoch": 0.16239499170635133,
+      "grad_norm": 1.3008679151535034,
+      "learning_rate": 9.996876059704726e-05,
+      "loss": 0.0831,
+      "step": 6070
+    },
+    {
+      "epoch": 0.1626625287602333,
+      "grad_norm": 1.9833636283874512,
+      "learning_rate": 9.996817346537343e-05,
+      "loss": 0.0829,
+      "step": 6080
+    },
+    {
+      "epoch": 0.16293006581411526,
+      "grad_norm": 2.1460869312286377,
+      "learning_rate": 9.996758086925526e-05,
+      "loss": 0.1706,
+      "step": 6090
+    },
+    {
+      "epoch": 0.16319760286799723,
+      "grad_norm": 1.9522624015808105,
+      "learning_rate": 9.996698280875759e-05,
+      "loss": 0.0956,
+      "step": 6100
+    },
+    {
+      "epoch": 0.1634651399218792,
+      "grad_norm": 5.923405170440674,
+      "learning_rate": 9.99663792839458e-05,
+      "loss": 0.1033,
+      "step": 6110
+    },
+    {
+      "epoch": 0.16373267697576113,
+      "grad_norm": 1.8757652044296265,
+      "learning_rate": 9.99657702948859e-05,
+      "loss": 0.1025,
+      "step": 6120
+    },
+    {
+      "epoch": 0.1640002140296431,
+      "grad_norm": 4.116968154907227,
+      "learning_rate": 9.996515584164448e-05,
+      "loss": 0.1742,
+      "step": 6130
+    },
+    {
+      "epoch": 0.16426775108352507,
+      "grad_norm": 1.9392778873443604,
+      "learning_rate": 9.996453592428873e-05,
+      "loss": 0.1864,
+      "step": 6140
+    },
+    {
+      "epoch": 0.16453528813740703,
+      "grad_norm": 2.479210138320923,
+      "learning_rate": 9.996391054288646e-05,
+      "loss": 0.1553,
+      "step": 6150
+    },
+    {
+      "epoch": 0.164802825191289,
+      "grad_norm": 0.9382572770118713,
+      "learning_rate": 9.996327969750605e-05,
+      "loss": 0.1362,
+      "step": 6160
+    },
+    {
+      "epoch": 0.16507036224517097,
+      "grad_norm": 1.6960062980651855,
+      "learning_rate": 9.996264338821649e-05,
+      "loss": 0.3839,
+      "step": 6170
+    },
+    {
+      "epoch": 0.16533789929905293,
+      "grad_norm": 1.8536770343780518,
+      "learning_rate": 9.996200161508735e-05,
+      "loss": 0.1583,
+      "step": 6180
+    },
+    {
+      "epoch": 0.16560543635293487,
+      "grad_norm": 3.539649724960327,
+      "learning_rate": 9.996135437818885e-05,
+      "loss": 0.3171,
+      "step": 6190
+    },
+    {
+      "epoch": 0.16587297340681684,
+      "grad_norm": 2.909716844558716,
+      "learning_rate": 9.996070167759175e-05,
+      "loss": 0.201,
+      "step": 6200
+    },
+    {
+      "epoch": 0.1661405104606988,
+      "grad_norm": 3.0367794036865234,
+      "learning_rate": 9.996004351336743e-05,
+      "loss": 0.1128,
+      "step": 6210
+    },
+    {
+      "epoch": 0.16640804751458077,
+      "grad_norm": 1.4757113456726074,
+      "learning_rate": 9.995937988558785e-05,
+      "loss": 0.1657,
+      "step": 6220
+    },
+    {
+      "epoch": 0.16667558456846274,
+      "grad_norm": 2.7811591625213623,
+      "learning_rate": 9.995871079432561e-05,
+      "loss": 0.2722,
+      "step": 6230
+    },
+    {
+      "epoch": 0.1669431216223447,
+      "grad_norm": 3.620941638946533,
+      "learning_rate": 9.995803623965389e-05,
+      "loss": 0.1228,
+      "step": 6240
+    },
+    {
+      "epoch": 0.16721065867622667,
+      "grad_norm": 1.204426646232605,
+      "learning_rate": 9.995735622164641e-05,
+      "loss": 0.1291,
+      "step": 6250
+    },
+    {
+      "epoch": 0.1674781957301086,
+      "grad_norm": 4.050158977508545,
+      "learning_rate": 9.995667074037758e-05,
+      "loss": 0.1451,
+      "step": 6260
+    },
+    {
+      "epoch": 0.16774573278399058,
+      "grad_norm": 2.1068108081817627,
+      "learning_rate": 9.995597979592232e-05,
+      "loss": 0.176,
+      "step": 6270
+    },
+    {
+      "epoch": 0.16801326983787254,
+      "grad_norm": 2.194674253463745,
+      "learning_rate": 9.995528338835625e-05,
+      "loss": 0.1189,
+      "step": 6280
+    },
+    {
+      "epoch": 0.1682808068917545,
+      "grad_norm": 2.3699734210968018,
+      "learning_rate": 9.995458151775547e-05,
+      "loss": 0.0954,
+      "step": 6290
+    },
+    {
+      "epoch": 0.16854834394563648,
+      "grad_norm": 2.639029026031494,
+      "learning_rate": 9.995387418419677e-05,
+      "loss": 0.1288,
+      "step": 6300
+    },
+    {
+      "epoch": 0.16881588099951844,
+      "grad_norm": 1.497971773147583,
+      "learning_rate": 9.99531613877575e-05,
+      "loss": 0.1716,
+      "step": 6310
+    },
+    {
+      "epoch": 0.1690834180534004,
+      "grad_norm": 1.3116201162338257,
+      "learning_rate": 9.995244312851559e-05,
+      "loss": 0.1137,
+      "step": 6320
+    },
+    {
+      "epoch": 0.16935095510728235,
+      "grad_norm": 4.220026969909668,
+      "learning_rate": 9.995171940654961e-05,
+      "loss": 0.1399,
+      "step": 6330
+    },
+    {
+      "epoch": 0.1696184921611643,
+      "grad_norm": 1.2118412256240845,
+      "learning_rate": 9.995099022193871e-05,
+      "loss": 0.156,
+      "step": 6340
+    },
+    {
+      "epoch": 0.16988602921504628,
+      "grad_norm": 1.1523243188858032,
+      "learning_rate": 9.995025557476261e-05,
+      "loss": 0.1429,
+      "step": 6350
+    },
+    {
+      "epoch": 0.17015356626892825,
+      "grad_norm": 1.2214714288711548,
+      "learning_rate": 9.994951546510165e-05,
+      "loss": 0.1294,
+      "step": 6360
+    },
+    {
+      "epoch": 0.1704211033228102,
+      "grad_norm": 2.4627785682678223,
+      "learning_rate": 9.994876989303679e-05,
+      "loss": 0.1222,
+      "step": 6370
+    },
+    {
+      "epoch": 0.17068864037669218,
+      "grad_norm": 1.4282948970794678,
+      "learning_rate": 9.994801885864955e-05,
+      "loss": 0.1184,
+      "step": 6380
+    },
+    {
+      "epoch": 0.17095617743057415,
+      "grad_norm": 2.346067428588867,
+      "learning_rate": 9.994726236202205e-05,
+      "loss": 0.1137,
+      "step": 6390
+    },
+    {
+      "epoch": 0.17122371448445609,
+      "grad_norm": 9.943742752075195,
+      "learning_rate": 9.994650040323704e-05,
+      "loss": 0.2346,
+      "step": 6400
+    },
+    {
+      "epoch": 0.17149125153833805,
+      "grad_norm": 1.4862421751022339,
+      "learning_rate": 9.994573298237784e-05,
+      "loss": 0.1858,
+      "step": 6410
+    },
+    {
+      "epoch": 0.17175878859222002,
+      "grad_norm": 3.2627735137939453,
+      "learning_rate": 9.994496009952837e-05,
+      "loss": 0.1162,
+      "step": 6420
+    },
+    {
+      "epoch": 0.17202632564610199,
+      "grad_norm": 2.259890556335449,
+      "learning_rate": 9.994418175477316e-05,
+      "loss": 0.1604,
+      "step": 6430
+    },
+    {
+      "epoch": 0.17229386269998395,
+      "grad_norm": 1.619936466217041,
+      "learning_rate": 9.994339794819733e-05,
+      "loss": 0.0845,
+      "step": 6440
+    },
+    {
+      "epoch": 0.17256139975386592,
+      "grad_norm": 1.7823067903518677,
+      "learning_rate": 9.994260867988658e-05,
+      "loss": 0.1116,
+      "step": 6450
+    },
+    {
+      "epoch": 0.17282893680774788,
+      "grad_norm": 2.049935817718506,
+      "learning_rate": 9.994181394992723e-05,
+      "loss": 0.1323,
+      "step": 6460
+    },
+    {
+      "epoch": 0.17309647386162982,
+      "grad_norm": 1.2583856582641602,
+      "learning_rate": 9.994101375840618e-05,
+      "loss": 0.1781,
+      "step": 6470
+    },
+    {
+      "epoch": 0.1733640109155118,
+      "grad_norm": 2.0496578216552734,
+      "learning_rate": 9.994020810541098e-05,
+      "loss": 0.1161,
+      "step": 6480
+    },
+    {
+      "epoch": 0.17363154796939376,
+      "grad_norm": 3.0217502117156982,
+      "learning_rate": 9.99393969910297e-05,
+      "loss": 0.0971,
+      "step": 6490
+    },
+    {
+      "epoch": 0.17389908502327572,
+      "grad_norm": 1.6570594310760498,
+      "learning_rate": 9.993858041535104e-05,
+      "loss": 0.1779,
+      "step": 6500
+    },
+    {
+      "epoch": 0.1741666220771577,
+      "grad_norm": 1.653567910194397,
+      "learning_rate": 9.99377583784643e-05,
+      "loss": 0.1232,
+      "step": 6510
+    },
+    {
+      "epoch": 0.17443415913103966,
+      "grad_norm": 1.6445246934890747,
+      "learning_rate": 9.993693088045939e-05,
+      "loss": 0.1192,
+      "step": 6520
+    },
+    {
+      "epoch": 0.17470169618492162,
+      "grad_norm": 3.442699670791626,
+      "learning_rate": 9.99360979214268e-05,
+      "loss": 0.1773,
+      "step": 6530
+    },
+    {
+      "epoch": 0.17496923323880356,
+      "grad_norm": 4.271001815795898,
+      "learning_rate": 9.99352595014576e-05,
+      "loss": 0.1374,
+      "step": 6540
+    },
+    {
+      "epoch": 0.17523677029268553,
+      "grad_norm": 1.6697121858596802,
+      "learning_rate": 9.993441562064354e-05,
+      "loss": 0.1205,
+      "step": 6550
+    },
+    {
+      "epoch": 0.1755043073465675,
+      "grad_norm": 2.266055107116699,
+      "learning_rate": 9.993356627907685e-05,
+      "loss": 0.3071,
+      "step": 6560
+    },
+    {
+      "epoch": 0.17577184440044946,
+      "grad_norm": 1.7424613237380981,
+      "learning_rate": 9.99327114768504e-05,
+      "loss": 0.1868,
+      "step": 6570
+    },
+    {
+      "epoch": 0.17603938145433143,
+      "grad_norm": 1.0938756465911865,
+      "learning_rate": 9.99318512140577e-05,
+      "loss": 0.2358,
+      "step": 6580
+    },
+    {
+      "epoch": 0.1763069185082134,
+      "grad_norm": 0.9428962469100952,
+      "learning_rate": 9.993098549079284e-05,
+      "loss": 0.6003,
+      "step": 6590
+    },
+    {
+      "epoch": 0.17657445556209536,
+      "grad_norm": 2.517547845840454,
+      "learning_rate": 9.993011430715047e-05,
+      "loss": 0.1172,
+      "step": 6600
+    },
+    {
+      "epoch": 0.1768419926159773,
+      "grad_norm": 2.2673802375793457,
+      "learning_rate": 9.992923766322586e-05,
+      "loss": 0.2005,
+      "step": 6610
+    },
+    {
+      "epoch": 0.17710952966985927,
+      "grad_norm": 2.8414878845214844,
+      "learning_rate": 9.99283555591149e-05,
+      "loss": 0.1305,
+      "step": 6620
+    },
+    {
+      "epoch": 0.17737706672374123,
+      "grad_norm": 4.3837690353393555,
+      "learning_rate": 9.992746799491404e-05,
+      "loss": 0.1718,
+      "step": 6630
+    },
+    {
+      "epoch": 0.1776446037776232,
+      "grad_norm": 3.3346669673919678,
+      "learning_rate": 9.992657497072033e-05,
+      "loss": 0.1624,
+      "step": 6640
+    },
+    {
+      "epoch": 0.17791214083150517,
+      "grad_norm": 2.348367691040039,
+      "learning_rate": 9.992567648663147e-05,
+      "loss": 0.178,
+      "step": 6650
+    },
+    {
+      "epoch": 0.17817967788538713,
+      "grad_norm": 1.740557312965393,
+      "learning_rate": 9.992477254274568e-05,
+      "loss": 0.1581,
+      "step": 6660
+    },
+    {
+      "epoch": 0.1784472149392691,
+      "grad_norm": 2.3376238346099854,
+      "learning_rate": 9.992386313916183e-05,
+      "loss": 0.12,
+      "step": 6670
+    },
+    {
+      "epoch": 0.17871475199315104,
+      "grad_norm": 2.591935634613037,
+      "learning_rate": 9.992294827597934e-05,
+      "loss": 0.136,
+      "step": 6680
+    },
+    {
+      "epoch": 0.178982289047033,
+      "grad_norm": 2.6830694675445557,
+      "learning_rate": 9.992202795329831e-05,
+      "loss": 0.1621,
+      "step": 6690
+    },
+    {
+      "epoch": 0.17924982610091497,
+      "grad_norm": 1.596716046333313,
+      "learning_rate": 9.992110217121936e-05,
+      "loss": 0.0927,
+      "step": 6700
+    },
+    {
+      "epoch": 0.17951736315479694,
+      "grad_norm": 1.2287269830703735,
+      "learning_rate": 9.992017092984372e-05,
+      "loss": 0.1639,
+      "step": 6710
+    },
+    {
+      "epoch": 0.1797849002086789,
+      "grad_norm": 2.034308671951294,
+      "learning_rate": 9.991923422927326e-05,
+      "loss": 0.1197,
+      "step": 6720
+    },
+    {
+      "epoch": 0.18005243726256087,
+      "grad_norm": 2.3157830238342285,
+      "learning_rate": 9.991829206961037e-05,
+      "loss": 0.1392,
+      "step": 6730
+    },
+    {
+      "epoch": 0.18031997431644284,
+      "grad_norm": 2.0063724517822266,
+      "learning_rate": 9.991734445095813e-05,
+      "loss": 0.1629,
+      "step": 6740
+    },
+    {
+      "epoch": 0.18058751137032478,
+      "grad_norm": 2.8792333602905273,
+      "learning_rate": 9.991639137342015e-05,
+      "loss": 0.1738,
+      "step": 6750
+    },
+    {
+      "epoch": 0.18085504842420674,
+      "grad_norm": 2.9331159591674805,
+      "learning_rate": 9.991543283710064e-05,
+      "loss": 0.1153,
+      "step": 6760
+    },
+    {
+      "epoch": 0.1811225854780887,
+      "grad_norm": 2.293959617614746,
+      "learning_rate": 9.991446884210445e-05,
+      "loss": 0.243,
+      "step": 6770
+    },
+    {
+      "epoch": 0.18139012253197068,
+      "grad_norm": 0.9256312847137451,
+      "learning_rate": 9.9913499388537e-05,
+      "loss": 0.1656,
+      "step": 6780
+    },
+    {
+      "epoch": 0.18165765958585264,
+      "grad_norm": 3.3486201763153076,
+      "learning_rate": 9.99125244765043e-05,
+      "loss": 0.1405,
+      "step": 6790
+    },
+    {
+      "epoch": 0.1819251966397346,
+      "grad_norm": 3.2146096229553223,
+      "learning_rate": 9.991154410611296e-05,
+      "loss": 0.2232,
+      "step": 6800
+    },
+    {
+      "epoch": 0.18219273369361658,
+      "grad_norm": 2.8801708221435547,
+      "learning_rate": 9.99105582774702e-05,
+      "loss": 0.1898,
+      "step": 6810
+    },
+    {
+      "epoch": 0.18246027074749852,
+      "grad_norm": 1.5311106443405151,
+      "learning_rate": 9.990956699068384e-05,
+      "loss": 0.1532,
+      "step": 6820
+    },
+    {
+      "epoch": 0.18272780780138048,
+      "grad_norm": 1.6631443500518799,
+      "learning_rate": 9.990857024586224e-05,
+      "loss": 0.1002,
+      "step": 6830
+    },
+    {
+      "epoch": 0.18299534485526245,
+      "grad_norm": 3.4409236907958984,
+      "learning_rate": 9.990756804311446e-05,
+      "loss": 0.218,
+      "step": 6840
+    },
+    {
+      "epoch": 0.18326288190914442,
+      "grad_norm": 1.279374361038208,
+      "learning_rate": 9.990656038255006e-05,
+      "loss": 0.1782,
+      "step": 6850
+    },
+    {
+      "epoch": 0.18353041896302638,
+      "grad_norm": 1.5665621757507324,
+      "learning_rate": 9.990554726427926e-05,
+      "loss": 0.1655,
+      "step": 6860
+    },
+    {
+      "epoch": 0.18379795601690835,
+      "grad_norm": 2.5189313888549805,
+      "learning_rate": 9.990452868841284e-05,
+      "loss": 0.2024,
+      "step": 6870
+    },
+    {
+      "epoch": 0.18406549307079031,
+      "grad_norm": 4.390002250671387,
+      "learning_rate": 9.99035046550622e-05,
+      "loss": 0.1071,
+      "step": 6880
+    },
+    {
+      "epoch": 0.18433303012467225,
+      "grad_norm": 1.7407352924346924,
+      "learning_rate": 9.99024751643393e-05,
+      "loss": 0.1317,
+      "step": 6890
+    },
+    {
+      "epoch": 0.18460056717855422,
+      "grad_norm": 3.700486421585083,
+      "learning_rate": 9.990144021635677e-05,
+      "loss": 0.2573,
+      "step": 6900
+    },
+    {
+      "epoch": 0.1848681042324362,
+      "grad_norm": 2.320997476577759,
+      "learning_rate": 9.990039981122775e-05,
+      "loss": 0.1235,
+      "step": 6910
+    },
+    {
+      "epoch": 0.18513564128631815,
+      "grad_norm": 0.824160099029541,
+      "learning_rate": 9.989935394906602e-05,
+      "loss": 0.1508,
+      "step": 6920
+    },
+    {
+      "epoch": 0.18540317834020012,
+      "grad_norm": 1.376738429069519,
+      "learning_rate": 9.989830262998598e-05,
+      "loss": 0.2177,
+      "step": 6930
+    },
+    {
+      "epoch": 0.1856707153940821,
+      "grad_norm": 2.217259407043457,
+      "learning_rate": 9.989724585410259e-05,
+      "loss": 0.1004,
+      "step": 6940
+    },
+    {
+      "epoch": 0.18593825244796405,
+      "grad_norm": 2.9117519855499268,
+      "learning_rate": 9.989618362153139e-05,
+      "loss": 0.1109,
+      "step": 6950
+    },
+    {
+      "epoch": 0.186205789501846,
+      "grad_norm": 2.463655948638916,
+      "learning_rate": 9.989511593238859e-05,
+      "loss": 0.1334,
+      "step": 6960
+    },
+    {
+      "epoch": 0.18647332655572796,
+      "grad_norm": 2.435532808303833,
+      "learning_rate": 9.98940427867909e-05,
+      "loss": 0.1177,
+      "step": 6970
+    },
+    {
+      "epoch": 0.18674086360960993,
+      "grad_norm": 1.447106957435608,
+      "learning_rate": 9.989296418485573e-05,
+      "loss": 0.0779,
+      "step": 6980
+    },
+    {
+      "epoch": 0.1870084006634919,
+      "grad_norm": 2.0865917205810547,
+      "learning_rate": 9.989188012670101e-05,
+      "loss": 0.1393,
+      "step": 6990
+    },
+    {
+      "epoch": 0.18727593771737386,
+      "grad_norm": 1.1041184663772583,
+      "learning_rate": 9.989079061244528e-05,
+      "loss": 0.1481,
+      "step": 7000
+    },
+    {
+      "epoch": 0.18754347477125582,
+      "grad_norm": 2.073221445083618,
+      "learning_rate": 9.988969564220769e-05,
+      "loss": 0.1943,
+      "step": 7010
+    },
+    {
+      "epoch": 0.1878110118251378,
+      "grad_norm": 2.546874523162842,
+      "learning_rate": 9.988859521610801e-05,
+      "loss": 0.1213,
+      "step": 7020
+    },
+    {
+      "epoch": 0.18807854887901976,
+      "grad_norm": 2.575896978378296,
+      "learning_rate": 9.988748933426656e-05,
+      "loss": 0.1487,
+      "step": 7030
+    },
+    {
+      "epoch": 0.1883460859329017,
+      "grad_norm": 1.0710291862487793,
+      "learning_rate": 9.988637799680428e-05,
+      "loss": 0.1022,
+      "step": 7040
+    },
+    {
+      "epoch": 0.18861362298678366,
+      "grad_norm": 3.497037410736084,
+      "learning_rate": 9.98852612038427e-05,
+      "loss": 0.1334,
+      "step": 7050
+    },
+    {
+      "epoch": 0.18888116004066563,
+      "grad_norm": 1.337948203086853,
+      "learning_rate": 9.988413895550397e-05,
+      "loss": 0.2089,
+      "step": 7060
+    },
+    {
+      "epoch": 0.1891486970945476,
+      "grad_norm": 2.703920841217041,
+      "learning_rate": 9.98830112519108e-05,
+      "loss": 0.2338,
+      "step": 7070
+    },
+    {
+      "epoch": 0.18941623414842956,
+      "grad_norm": 4.077207088470459,
+      "learning_rate": 9.98818780931865e-05,
+      "loss": 0.1185,
+      "step": 7080
+    },
+    {
+      "epoch": 0.18968377120231153,
+      "grad_norm": 6.045157432556152,
+      "learning_rate": 9.988073947945502e-05,
+      "loss": 0.1911,
+      "step": 7090
+    },
+    {
+      "epoch": 0.1899513082561935,
+      "grad_norm": 2.3209457397460938,
+      "learning_rate": 9.987959541084087e-05,
+      "loss": 0.1434,
+      "step": 7100
+    },
+    {
+      "epoch": 0.19021884531007544,
+      "grad_norm": 2.526503562927246,
+      "learning_rate": 9.987844588746915e-05,
+      "loss": 0.1389,
+      "step": 7110
+    },
+    {
+      "epoch": 0.1904863823639574,
+      "grad_norm": 1.652001142501831,
+      "learning_rate": 9.987729090946558e-05,
+      "loss": 0.1199,
+      "step": 7120
+    },
+    {
+      "epoch": 0.19075391941783937,
+      "grad_norm": 1.1682806015014648,
+      "learning_rate": 9.987613047695647e-05,
+      "loss": 0.1505,
+      "step": 7130
+    },
+    {
+      "epoch": 0.19102145647172133,
+      "grad_norm": 1.1065540313720703,
+      "learning_rate": 9.987496459006871e-05,
+      "loss": 0.1705,
+      "step": 7140
+    },
+    {
+      "epoch": 0.1912889935256033,
+      "grad_norm": 2.5070431232452393,
+      "learning_rate": 9.987379324892982e-05,
+      "loss": 0.1474,
+      "step": 7150
+    },
+    {
+      "epoch": 0.19155653057948527,
+      "grad_norm": 2.21379017829895,
+      "learning_rate": 9.987261645366788e-05,
+      "loss": 0.1292,
+      "step": 7160
+    },
+    {
+      "epoch": 0.19182406763336723,
+      "grad_norm": 3.1800389289855957,
+      "learning_rate": 9.987143420441158e-05,
+      "loss": 0.1721,
+      "step": 7170
+    },
+    {
+      "epoch": 0.19209160468724917,
+      "grad_norm": 1.814699649810791,
+      "learning_rate": 9.987024650129022e-05,
+      "loss": 0.0947,
+      "step": 7180
+    },
+    {
+      "epoch": 0.19235914174113114,
+      "grad_norm": 2.052980661392212,
+      "learning_rate": 9.986905334443368e-05,
+      "loss": 0.3476,
+      "step": 7190
+    },
+    {
+      "epoch": 0.1926266787950131,
+      "grad_norm": 3.141508102416992,
+      "learning_rate": 9.986785473397245e-05,
+      "loss": 0.1952,
+      "step": 7200
+    },
+    {
+      "epoch": 0.19289421584889507,
+      "grad_norm": 1.8159253597259521,
+      "learning_rate": 9.98666506700376e-05,
+      "loss": 0.1144,
+      "step": 7210
+    },
+    {
+      "epoch": 0.19316175290277704,
+      "grad_norm": 1.1704528331756592,
+      "learning_rate": 9.986544115276081e-05,
+      "loss": 0.0903,
+      "step": 7220
+    },
+    {
+      "epoch": 0.193429289956659,
+      "grad_norm": 1.503147840499878,
+      "learning_rate": 9.986422618227433e-05,
+      "loss": 0.2094,
+      "step": 7230
+    },
+    {
+      "epoch": 0.19369682701054097,
+      "grad_norm": 2.017632484436035,
+      "learning_rate": 9.986300575871106e-05,
+      "loss": 0.2177,
+      "step": 7240
+    },
+    {
+      "epoch": 0.1939643640644229,
+      "grad_norm": 2.145207405090332,
+      "learning_rate": 9.986177988220444e-05,
+      "loss": 0.1146,
+      "step": 7250
+    },
+    {
+      "epoch": 0.19423190111830488,
+      "grad_norm": 2.4722487926483154,
+      "learning_rate": 9.986054855288856e-05,
+      "loss": 0.1546,
+      "step": 7260
+    },
+    {
+      "epoch": 0.19449943817218684,
+      "grad_norm": 5.0568437576293945,
+      "learning_rate": 9.985931177089802e-05,
+      "loss": 0.2796,
+      "step": 7270
+    },
+    {
+      "epoch": 0.1947669752260688,
+      "grad_norm": 2.16977596282959,
+      "learning_rate": 9.985806953636814e-05,
+      "loss": 0.1131,
+      "step": 7280
+    },
+    {
+      "epoch": 0.19503451227995078,
+      "grad_norm": 1.044188380241394,
+      "learning_rate": 9.985682184943471e-05,
+      "loss": 0.1261,
+      "step": 7290
+    },
+    {
+      "epoch": 0.19530204933383274,
+      "grad_norm": 4.040148735046387,
+      "learning_rate": 9.98555687102342e-05,
+      "loss": 0.1074,
+      "step": 7300
+    },
+    {
+      "epoch": 0.1955695863877147,
+      "grad_norm": 2.1772263050079346,
+      "learning_rate": 9.985431011890367e-05,
+      "loss": 0.151,
+      "step": 7310
+    },
+    {
+      "epoch": 0.19583712344159665,
+      "grad_norm": 1.8664695024490356,
+      "learning_rate": 9.985304607558075e-05,
+      "loss": 0.1414,
+      "step": 7320
+    },
+    {
+      "epoch": 0.19610466049547862,
+      "grad_norm": 1.637865662574768,
+      "learning_rate": 9.985177658040364e-05,
+      "loss": 0.1554,
+      "step": 7330
+    },
+    {
+      "epoch": 0.19637219754936058,
+      "grad_norm": 1.7290970087051392,
+      "learning_rate": 9.985050163351119e-05,
+      "loss": 0.1086,
+      "step": 7340
+    },
+    {
+      "epoch": 0.19663973460324255,
+      "grad_norm": 1.3838223218917847,
+      "learning_rate": 9.984922123504286e-05,
+      "loss": 0.1228,
+      "step": 7350
+    },
+    {
+      "epoch": 0.19690727165712452,
+      "grad_norm": 3.301490545272827,
+      "learning_rate": 9.984793538513862e-05,
+      "loss": 0.1134,
+      "step": 7360
+    },
+    {
+      "epoch": 0.19717480871100648,
+      "grad_norm": 1.3136628866195679,
+      "learning_rate": 9.984664408393912e-05,
+      "loss": 0.1499,
+      "step": 7370
+    },
+    {
+      "epoch": 0.19744234576488845,
+      "grad_norm": 1.5475877523422241,
+      "learning_rate": 9.984534733158556e-05,
+      "loss": 0.1225,
+      "step": 7380
+    },
+    {
+      "epoch": 0.1977098828187704,
+      "grad_norm": 2.770557165145874,
+      "learning_rate": 9.984404512821977e-05,
+      "loss": 0.1152,
+      "step": 7390
+    },
+    {
+      "epoch": 0.19797741987265235,
+      "grad_norm": 2.8461718559265137,
+      "learning_rate": 9.984273747398411e-05,
+      "loss": 0.1644,
+      "step": 7400
+    },
+    {
+      "epoch": 0.19824495692653432,
+      "grad_norm": 1.1358819007873535,
+      "learning_rate": 9.984142436902165e-05,
+      "loss": 0.102,
+      "step": 7410
+    },
+    {
+      "epoch": 0.1985124939804163,
+      "grad_norm": 2.672268867492676,
+      "learning_rate": 9.984010581347596e-05,
+      "loss": 0.1462,
+      "step": 7420
+    },
+    {
+      "epoch": 0.19878003103429825,
+      "grad_norm": 1.8550987243652344,
+      "learning_rate": 9.983878180749121e-05,
+      "loss": 0.1979,
+      "step": 7430
+    },
+    {
+      "epoch": 0.19904756808818022,
+      "grad_norm": 1.277213454246521,
+      "learning_rate": 9.983745235121222e-05,
+      "loss": 0.1504,
+      "step": 7440
+    },
+    {
+      "epoch": 0.1993151051420622,
+      "grad_norm": 0.9310991168022156,
+      "learning_rate": 9.983611744478438e-05,
+      "loss": 0.0492,
+      "step": 7450
+    },
+    {
+      "epoch": 0.19958264219594413,
+      "grad_norm": 2.839970827102661,
+      "learning_rate": 9.983477708835365e-05,
+      "loss": 0.1667,
+      "step": 7460
+    },
+    {
+      "epoch": 0.1998501792498261,
+      "grad_norm": 1.2352454662322998,
+      "learning_rate": 9.983343128206664e-05,
+      "loss": 0.234,
+      "step": 7470
+    },
+    {
+      "epoch": 0.20011771630370806,
+      "grad_norm": 1.8263136148452759,
+      "learning_rate": 9.983208002607049e-05,
+      "loss": 0.1668,
+      "step": 7480
+    },
+    {
+      "epoch": 0.20038525335759003,
+      "grad_norm": 2.528616189956665,
+      "learning_rate": 9.9830723320513e-05,
+      "loss": 0.089,
+      "step": 7490
+    },
+    {
+      "epoch": 0.200652790411472,
+      "grad_norm": 0.8657050728797913,
+      "learning_rate": 9.982936116554254e-05,
+      "loss": 0.0908,
+      "step": 7500
+    },
+    {
+      "epoch": 0.20092032746535396,
+      "grad_norm": 2.6577534675598145,
+      "learning_rate": 9.982799356130803e-05,
+      "loss": 0.209,
+      "step": 7510
+    },
+    {
+      "epoch": 0.20118786451923593,
+      "grad_norm": 1.6340969800949097,
+      "learning_rate": 9.982662050795908e-05,
+      "loss": 0.1444,
+      "step": 7520
+    },
+    {
+      "epoch": 0.20145540157311786,
+      "grad_norm": 3.911717414855957,
+      "learning_rate": 9.982524200564583e-05,
+      "loss": 0.1113,
+      "step": 7530
+    },
+    {
+      "epoch": 0.20172293862699983,
+      "grad_norm": 1.975175380706787,
+      "learning_rate": 9.982385805451901e-05,
+      "loss": 0.0939,
+      "step": 7540
+    },
+    {
+      "epoch": 0.2019904756808818,
+      "grad_norm": 2.2767598628997803,
+      "learning_rate": 9.982246865472998e-05,
+      "loss": 0.1814,
+      "step": 7550
+    },
+    {
+      "epoch": 0.20225801273476376,
+      "grad_norm": 2.0023155212402344,
+      "learning_rate": 9.982107380643069e-05,
+      "loss": 0.0966,
+      "step": 7560
+    },
+    {
+      "epoch": 0.20252554978864573,
+      "grad_norm": 1.2372088432312012,
+      "learning_rate": 9.981967350977368e-05,
+      "loss": 0.1364,
+      "step": 7570
+    },
+    {
+      "epoch": 0.2027930868425277,
+      "grad_norm": 1.5896562337875366,
+      "learning_rate": 9.981826776491208e-05,
+      "loss": 0.1619,
+      "step": 7580
+    },
+    {
+      "epoch": 0.20306062389640966,
+      "grad_norm": 1.312674880027771,
+      "learning_rate": 9.98168565719996e-05,
+      "loss": 0.144,
+      "step": 7590
+    },
+    {
+      "epoch": 0.2033281609502916,
+      "grad_norm": 1.8131041526794434,
+      "learning_rate": 9.98154399311906e-05,
+      "loss": 0.1061,
+      "step": 7600
+    },
+    {
+      "epoch": 0.20359569800417357,
+      "grad_norm": 3.194160223007202,
+      "learning_rate": 9.981401784263997e-05,
+      "loss": 0.1603,
+      "step": 7610
+    },
+    {
+      "epoch": 0.20386323505805554,
+      "grad_norm": 1.643049955368042,
+      "learning_rate": 9.981259030650326e-05,
+      "loss": 0.1501,
+      "step": 7620
+    },
+    {
+      "epoch": 0.2041307721119375,
+      "grad_norm": 1.0593957901000977,
+      "learning_rate": 9.981115732293655e-05,
+      "loss": 0.1742,
+      "step": 7630
+    },
+    {
+      "epoch": 0.20439830916581947,
+      "grad_norm": 1.6894831657409668,
+      "learning_rate": 9.980971889209659e-05,
+      "loss": 0.1429,
+      "step": 7640
+    },
+    {
+      "epoch": 0.20466584621970144,
+      "grad_norm": 1.8601562976837158,
+      "learning_rate": 9.980827501414064e-05,
+      "loss": 0.1454,
+      "step": 7650
+    },
+    {
+      "epoch": 0.2049333832735834,
+      "grad_norm": 2.392519474029541,
+      "learning_rate": 9.980682568922663e-05,
+      "loss": 0.0662,
+      "step": 7660
+    },
+    {
+      "epoch": 0.20520092032746534,
+      "grad_norm": 1.206600546836853,
+      "learning_rate": 9.980537091751304e-05,
+      "loss": 0.0811,
+      "step": 7670
+    },
+    {
+      "epoch": 0.2054684573813473,
+      "grad_norm": 2.8217122554779053,
+      "learning_rate": 9.980391069915897e-05,
+      "loss": 0.185,
+      "step": 7680
+    },
+    {
+      "epoch": 0.20573599443522927,
+      "grad_norm": 2.9728426933288574,
+      "learning_rate": 9.98024450343241e-05,
+      "loss": 0.0867,
+      "step": 7690
+    },
+    {
+      "epoch": 0.20600353148911124,
+      "grad_norm": 1.755336880683899,
+      "learning_rate": 9.980097392316872e-05,
+      "loss": 0.2587,
+      "step": 7700
+    },
+    {
+      "epoch": 0.2062710685429932,
+      "grad_norm": 2.352666139602661,
+      "learning_rate": 9.97994973658537e-05,
+      "loss": 0.1506,
+      "step": 7710
+    },
+    {
+      "epoch": 0.20653860559687517,
+      "grad_norm": 1.6363210678100586,
+      "learning_rate": 9.979801536254054e-05,
+      "loss": 0.1476,
+      "step": 7720
+    },
+    {
+      "epoch": 0.20680614265075714,
+      "grad_norm": 2.8456661701202393,
+      "learning_rate": 9.979652791339127e-05,
+      "loss": 0.1706,
+      "step": 7730
+    },
+    {
+      "epoch": 0.20707367970463908,
+      "grad_norm": 1.8753719329833984,
+      "learning_rate": 9.97950350185686e-05,
+      "loss": 0.1096,
+      "step": 7740
+    },
+    {
+      "epoch": 0.20734121675852105,
+      "grad_norm": 3.2885262966156006,
+      "learning_rate": 9.979353667823574e-05,
+      "loss": 0.1784,
+      "step": 7750
+    },
+    {
+      "epoch": 0.207608753812403,
+      "grad_norm": 1.4597461223602295,
+      "learning_rate": 9.979203289255658e-05,
+      "loss": 0.1394,
+      "step": 7760
+    },
+    {
+      "epoch": 0.20787629086628498,
+      "grad_norm": 6.734646797180176,
+      "learning_rate": 9.979052366169557e-05,
+      "loss": 0.2664,
+      "step": 7770
+    },
+    {
+      "epoch": 0.20814382792016695,
+      "grad_norm": 2.5346851348876953,
+      "learning_rate": 9.978900898581775e-05,
+      "loss": 0.125,
+      "step": 7780
+    },
+    {
+      "epoch": 0.2084113649740489,
+      "grad_norm": 1.3623498678207397,
+      "learning_rate": 9.978748886508875e-05,
+      "loss": 0.1064,
+      "step": 7790
+    },
+    {
+      "epoch": 0.20867890202793088,
+      "grad_norm": 1.5502254962921143,
+      "learning_rate": 9.978596329967484e-05,
+      "loss": 0.1247,
+      "step": 7800
+    },
+    {
+      "epoch": 0.20894643908181282,
+      "grad_norm": 7.470516204833984,
+      "learning_rate": 9.978443228974284e-05,
+      "loss": 0.2708,
+      "step": 7810
+    },
+    {
+      "epoch": 0.20921397613569478,
+      "grad_norm": 3.429856300354004,
+      "learning_rate": 9.978289583546015e-05,
+      "loss": 0.1733,
+      "step": 7820
+    },
+    {
+      "epoch": 0.20948151318957675,
+      "grad_norm": 1.1992024183273315,
+      "learning_rate": 9.978135393699484e-05,
+      "loss": 0.1948,
+      "step": 7830
+    },
+    {
+      "epoch": 0.20974905024345872,
+      "grad_norm": 1.0708253383636475,
+      "learning_rate": 9.977980659451548e-05,
+      "loss": 0.0907,
+      "step": 7840
+    },
+    {
+      "epoch": 0.21001658729734068,
+      "grad_norm": 1.7950924634933472,
+      "learning_rate": 9.977825380819135e-05,
+      "loss": 0.17,
+      "step": 7850
+    },
+    {
+      "epoch": 0.21028412435122265,
+      "grad_norm": 1.2497552633285522,
+      "learning_rate": 9.97766955781922e-05,
+      "loss": 0.1281,
+      "step": 7860
+    },
+    {
+      "epoch": 0.21055166140510462,
+      "grad_norm": 1.7946873903274536,
+      "learning_rate": 9.977513190468848e-05,
+      "loss": 0.0967,
+      "step": 7870
+    },
+    {
+      "epoch": 0.21081919845898656,
+      "grad_norm": 3.4223475456237793,
+      "learning_rate": 9.977356278785116e-05,
+      "loss": 0.1358,
+      "step": 7880
+    },
+    {
+      "epoch": 0.21108673551286852,
+      "grad_norm": 2.149904727935791,
+      "learning_rate": 9.977198822785184e-05,
+      "loss": 0.1244,
+      "step": 7890
+    },
+    {
+      "epoch": 0.2113542725667505,
+      "grad_norm": 1.2185226678848267,
+      "learning_rate": 9.977040822486273e-05,
+      "loss": 0.2516,
+      "step": 7900
+    },
+    {
+      "epoch": 0.21162180962063246,
+      "grad_norm": 0.6435996294021606,
+      "learning_rate": 9.97688227790566e-05,
+      "loss": 0.0785,
+      "step": 7910
+    },
+    {
+      "epoch": 0.21188934667451442,
+      "grad_norm": 1.117307186126709,
+      "learning_rate": 9.976723189060684e-05,
+      "loss": 0.1178,
+      "step": 7920
+    },
+    {
+      "epoch": 0.2121568837283964,
+      "grad_norm": 3.7456748485565186,
+      "learning_rate": 9.976563555968742e-05,
+      "loss": 0.1691,
+      "step": 7930
+    },
+    {
+      "epoch": 0.21242442078227836,
+      "grad_norm": 3.309863328933716,
+      "learning_rate": 9.976403378647292e-05,
+      "loss": 0.165,
+      "step": 7940
+    },
+    {
+      "epoch": 0.2126919578361603,
+      "grad_norm": 1.6981874704360962,
+      "learning_rate": 9.97624265711385e-05,
+      "loss": 0.0843,
+      "step": 7950
+    },
+    {
+      "epoch": 0.21295949489004226,
+      "grad_norm": 2.435800075531006,
+      "learning_rate": 9.976081391385993e-05,
+      "loss": 0.1836,
+      "step": 7960
+    },
+    {
+      "epoch": 0.21322703194392423,
+      "grad_norm": 1.0009814500808716,
+      "learning_rate": 9.975919581481356e-05,
+      "loss": 0.0763,
+      "step": 7970
+    },
+    {
+      "epoch": 0.2134945689978062,
+      "grad_norm": 3.4932539463043213,
+      "learning_rate": 9.975757227417634e-05,
+      "loss": 0.1474,
+      "step": 7980
+    },
+    {
+      "epoch": 0.21376210605168816,
+      "grad_norm": 1.3953343629837036,
+      "learning_rate": 9.975594329212586e-05,
+      "loss": 0.1418,
+      "step": 7990
+    },
+    {
+      "epoch": 0.21402964310557013,
+      "grad_norm": 4.332700729370117,
+      "learning_rate": 9.97543088688402e-05,
+      "loss": 0.1655,
+      "step": 8000
+    },
+    {
+      "epoch": 0.2142971801594521,
+      "grad_norm": 1.3221843242645264,
+      "learning_rate": 9.975266900449814e-05,
+      "loss": 0.1413,
+      "step": 8010
+    },
+    {
+      "epoch": 0.21456471721333403,
+      "grad_norm": 1.4961203336715698,
+      "learning_rate": 9.975102369927898e-05,
+      "loss": 0.1242,
+      "step": 8020
+    },
+    {
+      "epoch": 0.214832254267216,
+      "grad_norm": 1.5934072732925415,
+      "learning_rate": 9.974937295336269e-05,
+      "loss": 0.1352,
+      "step": 8030
+    },
+    {
+      "epoch": 0.21509979132109797,
+      "grad_norm": 1.574845552444458,
+      "learning_rate": 9.974771676692975e-05,
+      "loss": 0.1106,
+      "step": 8040
+    },
+    {
+      "epoch": 0.21536732837497993,
+      "grad_norm": 3.7656021118164062,
+      "learning_rate": 9.974605514016131e-05,
+      "loss": 0.1264,
+      "step": 8050
+    },
+    {
+      "epoch": 0.2156348654288619,
+      "grad_norm": 1.5008454322814941,
+      "learning_rate": 9.974438807323907e-05,
+      "loss": 0.1084,
+      "step": 8060
+    },
+    {
+      "epoch": 0.21590240248274387,
+      "grad_norm": 1.8874163627624512,
+      "learning_rate": 9.974271556634535e-05,
+      "loss": 0.0964,
+      "step": 8070
+    },
+    {
+      "epoch": 0.21616993953662583,
+      "grad_norm": 2.371215343475342,
+      "learning_rate": 9.974103761966302e-05,
+      "loss": 0.0917,
+      "step": 8080
+    },
+    {
+      "epoch": 0.2164374765905078,
+      "grad_norm": 2.1461682319641113,
+      "learning_rate": 9.973935423337563e-05,
+      "loss": 0.1342,
+      "step": 8090
+    },
+    {
+      "epoch": 0.21670501364438974,
+      "grad_norm": 2.4077165126800537,
+      "learning_rate": 9.973766540766722e-05,
+      "loss": 0.1737,
+      "step": 8100
+    },
+    {
+      "epoch": 0.2169725506982717,
+      "grad_norm": 1.1834337711334229,
+      "learning_rate": 9.97359711427225e-05,
+      "loss": 0.1706,
+      "step": 8110
+    },
+    {
+      "epoch": 0.21724008775215367,
+      "grad_norm": 1.552042007446289,
+      "learning_rate": 9.973427143872677e-05,
+      "loss": 0.1863,
+      "step": 8120
+    },
+    {
+      "epoch": 0.21750762480603564,
+      "grad_norm": 3.15468168258667,
+      "learning_rate": 9.973256629586589e-05,
+      "loss": 0.0887,
+      "step": 8130
+    },
+    {
+      "epoch": 0.2177751618599176,
+      "grad_norm": 1.5166351795196533,
+      "learning_rate": 9.973085571432632e-05,
+      "loss": 0.1142,
+      "step": 8140
+    },
+    {
+      "epoch": 0.21804269891379957,
+      "grad_norm": 1.9280108213424683,
+      "learning_rate": 9.972913969429513e-05,
+      "loss": 0.1181,
+      "step": 8150
+    },
+    {
+      "epoch": 0.21831023596768154,
+      "grad_norm": 4.926897048950195,
+      "learning_rate": 9.972741823596e-05,
+      "loss": 0.1851,
+      "step": 8160
+    },
+    {
+      "epoch": 0.21857777302156348,
+      "grad_norm": 2.470914840698242,
+      "learning_rate": 9.972569133950917e-05,
+      "loss": 0.1242,
+      "step": 8170
+    },
+    {
+      "epoch": 0.21884531007544544,
+      "grad_norm": 2.4203662872314453,
+      "learning_rate": 9.972395900513151e-05,
+      "loss": 0.0901,
+      "step": 8180
+    },
+    {
+      "epoch": 0.2191128471293274,
+      "grad_norm": 2.907017469406128,
+      "learning_rate": 9.972222123301645e-05,
+      "loss": 0.1514,
+      "step": 8190
+    },
+    {
+      "epoch": 0.21938038418320938,
+      "grad_norm": 2.2339048385620117,
+      "learning_rate": 9.972047802335403e-05,
+      "loss": 0.1962,
+      "step": 8200
+    },
+    {
+      "epoch": 0.21964792123709134,
+      "grad_norm": 1.4709056615829468,
+      "learning_rate": 9.971872937633488e-05,
+      "loss": 0.1591,
+      "step": 8210
+    },
+    {
+      "epoch": 0.2199154582909733,
+      "grad_norm": 2.2633144855499268,
+      "learning_rate": 9.971697529215024e-05,
+      "loss": 0.1504,
+      "step": 8220
+    },
+    {
+      "epoch": 0.22018299534485528,
+      "grad_norm": 1.713781476020813,
+      "learning_rate": 9.971521577099192e-05,
+      "loss": 0.2171,
+      "step": 8230
+    },
+    {
+      "epoch": 0.22045053239873721,
+      "grad_norm": 2.537506103515625,
+      "learning_rate": 9.971345081305236e-05,
+      "loss": 0.074,
+      "step": 8240
+    },
+    {
+      "epoch": 0.22071806945261918,
+      "grad_norm": 3.1921417713165283,
+      "learning_rate": 9.971168041852456e-05,
+      "loss": 0.1463,
+      "step": 8250
+    },
+    {
+      "epoch": 0.22098560650650115,
+      "grad_norm": 1.4856102466583252,
+      "learning_rate": 9.970990458760215e-05,
+      "loss": 0.1472,
+      "step": 8260
+    },
+    {
+      "epoch": 0.22125314356038311,
+      "grad_norm": 5.3148193359375,
+      "learning_rate": 9.970812332047929e-05,
+      "loss": 0.1457,
+      "step": 8270
+    },
+    {
+      "epoch": 0.22152068061426508,
+      "grad_norm": 2.3844265937805176,
+      "learning_rate": 9.97063366173508e-05,
+      "loss": 0.1999,
+      "step": 8280
+    },
+    {
+      "epoch": 0.22178821766814705,
+      "grad_norm": 2.6845767498016357,
+      "learning_rate": 9.970454447841207e-05,
+      "loss": 0.182,
+      "step": 8290
+    },
+    {
+      "epoch": 0.22205575472202901,
+      "grad_norm": 2.9240570068359375,
+      "learning_rate": 9.970274690385909e-05,
+      "loss": 0.2113,
+      "step": 8300
+    },
+    {
+      "epoch": 0.22232329177591095,
+      "grad_norm": 1.857032060623169,
+      "learning_rate": 9.970094389388844e-05,
+      "loss": 0.1065,
+      "step": 8310
+    },
+    {
+      "epoch": 0.22259082882979292,
+      "grad_norm": 1.6749366521835327,
+      "learning_rate": 9.969913544869728e-05,
+      "loss": 0.127,
+      "step": 8320
+    },
+    {
+      "epoch": 0.22285836588367489,
+      "grad_norm": 3.4930453300476074,
+      "learning_rate": 9.96973215684834e-05,
+      "loss": 0.1721,
+      "step": 8330
+    },
+    {
+      "epoch": 0.22312590293755685,
+      "grad_norm": 1.9800903797149658,
+      "learning_rate": 9.969550225344513e-05,
+      "loss": 0.1067,
+      "step": 8340
+    },
+    {
+      "epoch": 0.22339343999143882,
+      "grad_norm": 2.3057103157043457,
+      "learning_rate": 9.969367750378147e-05,
+      "loss": 0.1697,
+      "step": 8350
+    },
+    {
+      "epoch": 0.22366097704532079,
+      "grad_norm": 2.512395143508911,
+      "learning_rate": 9.969184731969194e-05,
+      "loss": 0.1292,
+      "step": 8360
+    },
+    {
+      "epoch": 0.22392851409920275,
+      "grad_norm": 2.9143612384796143,
+      "learning_rate": 9.96900117013767e-05,
+      "loss": 0.1198,
+      "step": 8370
+    },
+    {
+      "epoch": 0.2241960511530847,
+      "grad_norm": 1.6493200063705444,
+      "learning_rate": 9.96881706490365e-05,
+      "loss": 0.066,
+      "step": 8380
+    },
+    {
+      "epoch": 0.22446358820696666,
+      "grad_norm": 2.0659213066101074,
+      "learning_rate": 9.968632416287265e-05,
+      "loss": 0.286,
+      "step": 8390
+    },
+    {
+      "epoch": 0.22473112526084862,
+      "grad_norm": 2.4055933952331543,
+      "learning_rate": 9.96844722430871e-05,
+      "loss": 0.1421,
+      "step": 8400
+    },
+    {
+      "epoch": 0.2249986623147306,
+      "grad_norm": 1.5484590530395508,
+      "learning_rate": 9.968261488988235e-05,
+      "loss": 0.1041,
+      "step": 8410
+    },
+    {
+      "epoch": 0.22526619936861256,
+      "grad_norm": 2.197449207305908,
+      "learning_rate": 9.968075210346155e-05,
+      "loss": 0.1195,
+      "step": 8420
+    },
+    {
+      "epoch": 0.22553373642249452,
+      "grad_norm": 2.2526016235351562,
+      "learning_rate": 9.967888388402839e-05,
+      "loss": 0.1455,
+      "step": 8430
+    },
+    {
+      "epoch": 0.2258012734763765,
+      "grad_norm": 6.885036945343018,
+      "learning_rate": 9.967701023178717e-05,
+      "loss": 0.2459,
+      "step": 8440
+    },
+    {
+      "epoch": 0.22606881053025843,
+      "grad_norm": 2.2770023345947266,
+      "learning_rate": 9.967513114694282e-05,
+      "loss": 0.1441,
+      "step": 8450
+    },
+    {
+      "epoch": 0.2263363475841404,
+      "grad_norm": 4.527496337890625,
+      "learning_rate": 9.967324662970079e-05,
+      "loss": 0.2382,
+      "step": 8460
+    },
+    {
+      "epoch": 0.22660388463802236,
+      "grad_norm": 3.3003056049346924,
+      "learning_rate": 9.96713566802672e-05,
+      "loss": 0.2716,
+      "step": 8470
+    },
+    {
+      "epoch": 0.22687142169190433,
+      "grad_norm": 2.496507167816162,
+      "learning_rate": 9.966946129884873e-05,
+      "loss": 0.1034,
+      "step": 8480
+    },
+    {
+      "epoch": 0.2271389587457863,
+      "grad_norm": 1.8746832609176636,
+      "learning_rate": 9.966756048565265e-05,
+      "loss": 0.1324,
+      "step": 8490
+    },
+    {
+      "epoch": 0.22740649579966826,
+      "grad_norm": 2.8011257648468018,
+      "learning_rate": 9.966565424088681e-05,
+      "loss": 0.1763,
+      "step": 8500
+    },
+    {
+      "epoch": 0.22767403285355023,
+      "grad_norm": 4.417097568511963,
+      "learning_rate": 9.96637425647597e-05,
+      "loss": 0.1768,
+      "step": 8510
+    },
+    {
+      "epoch": 0.22794156990743217,
+      "grad_norm": 2.854245185852051,
+      "learning_rate": 9.966182545748038e-05,
+      "loss": 0.148,
+      "step": 8520
+    },
+    {
+      "epoch": 0.22820910696131413,
+      "grad_norm": 1.4566303491592407,
+      "learning_rate": 9.96599029192585e-05,
+      "loss": 0.1523,
+      "step": 8530
+    },
+    {
+      "epoch": 0.2284766440151961,
+      "grad_norm": 0.8059036135673523,
+      "learning_rate": 9.965797495030428e-05,
+      "loss": 0.0774,
+      "step": 8540
+    },
+    {
+      "epoch": 0.22874418106907807,
+      "grad_norm": 3.247154474258423,
+      "learning_rate": 9.96560415508286e-05,
+      "loss": 0.1455,
+      "step": 8550
+    },
+    {
+      "epoch": 0.22901171812296003,
+      "grad_norm": 1.217594027519226,
+      "learning_rate": 9.965410272104286e-05,
+      "loss": 0.1923,
+      "step": 8560
+    },
+    {
+      "epoch": 0.229279255176842,
+      "grad_norm": 2.5005578994750977,
+      "learning_rate": 9.96521584611591e-05,
+      "loss": 0.1989,
+      "step": 8570
+    },
+    {
+      "epoch": 0.22954679223072397,
+      "grad_norm": 2.0639429092407227,
+      "learning_rate": 9.965020877138994e-05,
+      "loss": 0.0922,
+      "step": 8580
+    },
+    {
+      "epoch": 0.2298143292846059,
+      "grad_norm": 2.657850503921509,
+      "learning_rate": 9.964825365194861e-05,
+      "loss": 0.139,
+      "step": 8590
+    },
+    {
+      "epoch": 0.23008186633848787,
+      "grad_norm": 1.6098079681396484,
+      "learning_rate": 9.96462931030489e-05,
+      "loss": 0.195,
+      "step": 8600
+    },
+    {
+      "epoch": 0.23034940339236984,
+      "grad_norm": 2.1659111976623535,
+      "learning_rate": 9.96443271249052e-05,
+      "loss": 0.0911,
+      "step": 8610
+    },
+    {
+      "epoch": 0.2306169404462518,
+      "grad_norm": 3.4701461791992188,
+      "learning_rate": 9.964235571773255e-05,
+      "loss": 0.1303,
+      "step": 8620
+    },
+    {
+      "epoch": 0.23088447750013377,
+      "grad_norm": 1.885931372642517,
+      "learning_rate": 9.96403788817465e-05,
+      "loss": 0.1873,
+      "step": 8630
+    },
+    {
+      "epoch": 0.23115201455401574,
+      "grad_norm": 0.8512374758720398,
+      "learning_rate": 9.963839661716325e-05,
+      "loss": 0.1806,
+      "step": 8640
+    },
+    {
+      "epoch": 0.2314195516078977,
+      "grad_norm": 2.988626480102539,
+      "learning_rate": 9.963640892419958e-05,
+      "loss": 0.1751,
+      "step": 8650
+    },
+    {
+      "epoch": 0.23168708866177964,
+      "grad_norm": 8.322284698486328,
+      "learning_rate": 9.963441580307286e-05,
+      "loss": 0.1477,
+      "step": 8660
+    },
+    {
+      "epoch": 0.2319546257156616,
+      "grad_norm": 3.3678483963012695,
+      "learning_rate": 9.963241725400104e-05,
+      "loss": 0.1503,
+      "step": 8670
+    },
+    {
+      "epoch": 0.23222216276954358,
+      "grad_norm": 1.8299932479858398,
+      "learning_rate": 9.963041327720271e-05,
+      "loss": 0.1695,
+      "step": 8680
+    },
+    {
+      "epoch": 0.23248969982342554,
+      "grad_norm": 3.6056690216064453,
+      "learning_rate": 9.962840387289697e-05,
+      "loss": 0.1889,
+      "step": 8690
+    },
+    {
+      "epoch": 0.2327572368773075,
+      "grad_norm": 2.272538900375366,
+      "learning_rate": 9.962638904130363e-05,
+      "loss": 0.2213,
+      "step": 8700
+    },
+    {
+      "epoch": 0.23302477393118948,
+      "grad_norm": 5.639734745025635,
+      "learning_rate": 9.962436878264298e-05,
+      "loss": 0.1682,
+      "step": 8710
+    },
+    {
+      "epoch": 0.23329231098507144,
+      "grad_norm": 3.698225259780884,
+      "learning_rate": 9.962234309713598e-05,
+      "loss": 0.1066,
+      "step": 8720
+    },
+    {
+      "epoch": 0.23355984803895338,
+      "grad_norm": 2.027630090713501,
+      "learning_rate": 9.962031198500414e-05,
+      "loss": 0.1215,
+      "step": 8730
+    },
+    {
+      "epoch": 0.23382738509283535,
+      "grad_norm": 2.47491455078125,
+      "learning_rate": 9.961827544646958e-05,
+      "loss": 0.1261,
+      "step": 8740
+    },
+    {
+      "epoch": 0.23409492214671732,
+      "grad_norm": 2.39471697807312,
+      "learning_rate": 9.961623348175501e-05,
+      "loss": 0.1623,
+      "step": 8750
+    },
+    {
+      "epoch": 0.23436245920059928,
+      "grad_norm": 1.5266059637069702,
+      "learning_rate": 9.961418609108377e-05,
+      "loss": 0.1141,
+      "step": 8760
+    },
+    {
+      "epoch": 0.23462999625448125,
+      "grad_norm": 0.969753086566925,
+      "learning_rate": 9.961213327467971e-05,
+      "loss": 0.133,
+      "step": 8770
+    },
+    {
+      "epoch": 0.23489753330836322,
+      "grad_norm": 3.3299427032470703,
+      "learning_rate": 9.961007503276736e-05,
+      "loss": 0.1025,
+      "step": 8780
+    },
+    {
+      "epoch": 0.23516507036224518,
+      "grad_norm": 1.3253065347671509,
+      "learning_rate": 9.960801136557179e-05,
+      "loss": 0.2266,
+      "step": 8790
+    },
+    {
+      "epoch": 0.23543260741612712,
+      "grad_norm": 1.6259866952896118,
+      "learning_rate": 9.960594227331866e-05,
+      "loss": 0.1226,
+      "step": 8800
+    },
+    {
+      "epoch": 0.2357001444700091,
+      "grad_norm": 0.9904751181602478,
+      "learning_rate": 9.960386775623429e-05,
+      "loss": 0.1717,
+      "step": 8810
+    },
+    {
+      "epoch": 0.23596768152389105,
+      "grad_norm": 3.213036298751831,
+      "learning_rate": 9.96017878145455e-05,
+      "loss": 0.163,
+      "step": 8820
+    },
+    {
+      "epoch": 0.23623521857777302,
+      "grad_norm": 1.3302127122879028,
+      "learning_rate": 9.959970244847977e-05,
+      "loss": 0.086,
+      "step": 8830
+    },
+    {
+      "epoch": 0.236502755631655,
+      "grad_norm": 1.9095669984817505,
+      "learning_rate": 9.959761165826518e-05,
+      "loss": 0.1839,
+      "step": 8840
+    },
+    {
+      "epoch": 0.23677029268553695,
+      "grad_norm": 1.0545399188995361,
+      "learning_rate": 9.959551544413033e-05,
+      "loss": 0.1612,
+      "step": 8850
+    },
+    {
+      "epoch": 0.23703782973941892,
+      "grad_norm": 3.9593687057495117,
+      "learning_rate": 9.959341380630448e-05,
+      "loss": 0.1349,
+      "step": 8860
+    },
+    {
+      "epoch": 0.23730536679330086,
+      "grad_norm": 2.001040458679199,
+      "learning_rate": 9.959130674501746e-05,
+      "loss": 0.2481,
+      "step": 8870
+    },
+    {
+      "epoch": 0.23757290384718283,
+      "grad_norm": 2.8113150596618652,
+      "learning_rate": 9.958919426049968e-05,
+      "loss": 0.1462,
+      "step": 8880
+    },
+    {
+      "epoch": 0.2378404409010648,
+      "grad_norm": 1.5897421836853027,
+      "learning_rate": 9.958707635298219e-05,
+      "loss": 0.1051,
+      "step": 8890
+    },
+    {
+      "epoch": 0.23810797795494676,
+      "grad_norm": 2.644256353378296,
+      "learning_rate": 9.958495302269657e-05,
+      "loss": 0.1615,
+      "step": 8900
+    },
+    {
+      "epoch": 0.23837551500882873,
+      "grad_norm": 1.6341334581375122,
+      "learning_rate": 9.958282426987503e-05,
+      "loss": 0.1631,
+      "step": 8910
+    },
+    {
+      "epoch": 0.2386430520627107,
+      "grad_norm": 1.9485529661178589,
+      "learning_rate": 9.95806900947504e-05,
+      "loss": 0.1,
+      "step": 8920
+    },
+    {
+      "epoch": 0.23891058911659266,
+      "grad_norm": 1.7049938440322876,
+      "learning_rate": 9.957855049755604e-05,
+      "loss": 0.1052,
+      "step": 8930
+    },
+    {
+      "epoch": 0.2391781261704746,
+      "grad_norm": 1.844305157661438,
+      "learning_rate": 9.957640547852593e-05,
+      "loss": 0.0824,
+      "step": 8940
+    },
+    {
+      "epoch": 0.23944566322435656,
+      "grad_norm": 0.764891505241394,
+      "learning_rate": 9.957425503789466e-05,
+      "loss": 0.1058,
+      "step": 8950
+    },
+    {
+      "epoch": 0.23971320027823853,
+      "grad_norm": 2.7427659034729004,
+      "learning_rate": 9.957209917589738e-05,
+      "loss": 0.1155,
+      "step": 8960
+    },
+    {
+      "epoch": 0.2399807373321205,
+      "grad_norm": 2.2665855884552,
+      "learning_rate": 9.956993789276987e-05,
+      "loss": 0.1317,
+      "step": 8970
+    },
+    {
+      "epoch": 0.24024827438600246,
+      "grad_norm": 2.2529265880584717,
+      "learning_rate": 9.956777118874847e-05,
+      "loss": 0.1273,
+      "step": 8980
+    },
+    {
+      "epoch": 0.24051581143988443,
+      "grad_norm": 1.9103055000305176,
+      "learning_rate": 9.956559906407016e-05,
+      "loss": 0.1185,
+      "step": 8990
+    },
+    {
+      "epoch": 0.2407833484937664,
+      "grad_norm": 1.159731388092041,
+      "learning_rate": 9.956342151897245e-05,
+      "loss": 0.1358,
+      "step": 9000
+    },
+    {
+      "epoch": 0.24105088554764834,
+      "grad_norm": 2.055572271347046,
+      "learning_rate": 9.956123855369346e-05,
+      "loss": 0.1024,
+      "step": 9010
+    },
+    {
+      "epoch": 0.2413184226015303,
+      "grad_norm": 3.1539649963378906,
+      "learning_rate": 9.955905016847196e-05,
+      "loss": 0.1784,
+      "step": 9020
+    },
+    {
+      "epoch": 0.24158595965541227,
+      "grad_norm": 1.6665563583374023,
+      "learning_rate": 9.955685636354723e-05,
+      "loss": 0.2291,
+      "step": 9030
+    },
+    {
+      "epoch": 0.24185349670929424,
+      "grad_norm": 6.818243980407715,
+      "learning_rate": 9.95546571391592e-05,
+      "loss": 0.1707,
+      "step": 9040
+    },
+    {
+      "epoch": 0.2421210337631762,
+      "grad_norm": 1.2849680185317993,
+      "learning_rate": 9.955245249554837e-05,
+      "loss": 0.1351,
+      "step": 9050
+    },
+    {
+      "epoch": 0.24238857081705817,
+      "grad_norm": 3.121690511703491,
+      "learning_rate": 9.955024243295582e-05,
+      "loss": 0.1453,
+      "step": 9060
+    },
+    {
+      "epoch": 0.24265610787094014,
+      "grad_norm": 2.174335479736328,
+      "learning_rate": 9.954802695162328e-05,
+      "loss": 0.2244,
+      "step": 9070
+    },
+    {
+      "epoch": 0.24292364492482207,
+      "grad_norm": 2.153876304626465,
+      "learning_rate": 9.954580605179302e-05,
+      "loss": 0.2159,
+      "step": 9080
+    },
+    {
+      "epoch": 0.24319118197870404,
+      "grad_norm": 1.9050780534744263,
+      "learning_rate": 9.954357973370788e-05,
+      "loss": 0.0977,
+      "step": 9090
+    },
+    {
+      "epoch": 0.243458719032586,
+      "grad_norm": 2.2903575897216797,
+      "learning_rate": 9.954134799761135e-05,
+      "loss": 0.1281,
+      "step": 9100
+    },
+    {
+      "epoch": 0.24372625608646797,
+      "grad_norm": 1.5738930702209473,
+      "learning_rate": 9.953911084374748e-05,
+      "loss": 0.2099,
+      "step": 9110
+    },
+    {
+      "epoch": 0.24399379314034994,
+      "grad_norm": 2.555774211883545,
+      "learning_rate": 9.953686827236093e-05,
+      "loss": 0.2307,
+      "step": 9120
+    },
+    {
+      "epoch": 0.2442613301942319,
+      "grad_norm": 1.7551316022872925,
+      "learning_rate": 9.953462028369695e-05,
+      "loss": 0.1527,
+      "step": 9130
+    },
+    {
+      "epoch": 0.24452886724811387,
+      "grad_norm": 1.5359466075897217,
+      "learning_rate": 9.953236687800136e-05,
+      "loss": 0.1744,
+      "step": 9140
+    },
+    {
+      "epoch": 0.24479640430199584,
+      "grad_norm": 4.662628650665283,
+      "learning_rate": 9.95301080555206e-05,
+      "loss": 0.1652,
+      "step": 9150
+    },
+    {
+      "epoch": 0.24506394135587778,
+      "grad_norm": 1.6426950693130493,
+      "learning_rate": 9.952784381650171e-05,
+      "loss": 0.2137,
+      "step": 9160
+    },
+    {
+      "epoch": 0.24533147840975975,
+      "grad_norm": 2.416210889816284,
+      "learning_rate": 9.952557416119226e-05,
+      "loss": 0.1223,
+      "step": 9170
+    },
+    {
+      "epoch": 0.2455990154636417,
+      "grad_norm": 2.99348783493042,
+      "learning_rate": 9.95232990898405e-05,
+      "loss": 0.223,
+      "step": 9180
+    },
+    {
+      "epoch": 0.24586655251752368,
+      "grad_norm": 1.5514556169509888,
+      "learning_rate": 9.95210186026952e-05,
+      "loss": 0.1857,
+      "step": 9190
+    },
+    {
+      "epoch": 0.24613408957140565,
+      "grad_norm": 2.6005101203918457,
+      "learning_rate": 9.951873270000576e-05,
+      "loss": 0.1226,
+      "step": 9200
+    },
+    {
+      "epoch": 0.2464016266252876,
+      "grad_norm": 4.305438041687012,
+      "learning_rate": 9.951644138202216e-05,
+      "loss": 0.1772,
+      "step": 9210
+    },
+    {
+      "epoch": 0.24666916367916958,
+      "grad_norm": 1.190798282623291,
+      "learning_rate": 9.951414464899498e-05,
+      "loss": 0.1498,
+      "step": 9220
+    },
+    {
+      "epoch": 0.24693670073305152,
+      "grad_norm": 2.50909686088562,
+      "learning_rate": 9.951184250117538e-05,
+      "loss": 0.0762,
+      "step": 9230
+    },
+    {
+      "epoch": 0.24720423778693348,
+      "grad_norm": 0.9732526540756226,
+      "learning_rate": 9.950953493881513e-05,
+      "loss": 0.1136,
+      "step": 9240
+    },
+    {
+      "epoch": 0.24747177484081545,
+      "grad_norm": 2.6547279357910156,
+      "learning_rate": 9.950722196216658e-05,
+      "loss": 0.2325,
+      "step": 9250
+    },
+    {
+      "epoch": 0.24773931189469742,
+      "grad_norm": 2.415470600128174,
+      "learning_rate": 9.950490357148265e-05,
+      "loss": 0.1413,
+      "step": 9260
+    },
+    {
+      "epoch": 0.24800684894857938,
+      "grad_norm": 2.437957525253296,
+      "learning_rate": 9.950257976701692e-05,
+      "loss": 0.1666,
+      "step": 9270
+    },
+    {
+      "epoch": 0.24827438600246135,
+      "grad_norm": 1.2392637729644775,
+      "learning_rate": 9.950025054902348e-05,
+      "loss": 0.0894,
+      "step": 9280
+    },
+    {
+      "epoch": 0.24854192305634332,
+      "grad_norm": 1.7520335912704468,
+      "learning_rate": 9.949791591775706e-05,
+      "loss": 0.1356,
+      "step": 9290
+    },
+    {
+      "epoch": 0.24880946011022526,
+      "grad_norm": 2.041790008544922,
+      "learning_rate": 9.949557587347298e-05,
+      "loss": 0.115,
+      "step": 9300
+    },
+    {
+      "epoch": 0.24907699716410722,
+      "grad_norm": 1.8079843521118164,
+      "learning_rate": 9.949323041642713e-05,
+      "loss": 0.1398,
+      "step": 9310
+    },
+    {
+      "epoch": 0.2493445342179892,
+      "grad_norm": 2.0146939754486084,
+      "learning_rate": 9.949087954687602e-05,
+      "loss": 0.13,
+      "step": 9320
+    },
+    {
+      "epoch": 0.24961207127187116,
+      "grad_norm": 1.08698308467865,
+      "learning_rate": 9.948852326507672e-05,
+      "loss": 0.1636,
+      "step": 9330
+    },
+    {
+      "epoch": 0.24987960832575312,
+      "grad_norm": 2.953688859939575,
+      "learning_rate": 9.948616157128694e-05,
+      "loss": 0.1239,
+      "step": 9340
+    },
+    {
+      "epoch": 0.25014714537963506,
+      "grad_norm": 2.9318881034851074,
+      "learning_rate": 9.948379446576493e-05,
+      "loss": 0.1516,
+      "step": 9350
+    },
+    {
+      "epoch": 0.25041468243351706,
+      "grad_norm": 1.9916322231292725,
+      "learning_rate": 9.948142194876952e-05,
+      "loss": 0.1642,
+      "step": 9360
+    },
+    {
+      "epoch": 0.250682219487399,
+      "grad_norm": 2.1268150806427,
+      "learning_rate": 9.947904402056024e-05,
+      "loss": 0.2322,
+      "step": 9370
+    },
+    {
+      "epoch": 0.250949756541281,
+      "grad_norm": 3.107719898223877,
+      "learning_rate": 9.947666068139708e-05,
+      "loss": 0.1282,
+      "step": 9380
+    },
+    {
+      "epoch": 0.2512172935951629,
+      "grad_norm": 3.0808701515197754,
+      "learning_rate": 9.947427193154071e-05,
+      "loss": 0.0873,
+      "step": 9390
+    },
+    {
+      "epoch": 0.25148483064904487,
+      "grad_norm": 1.809828519821167,
+      "learning_rate": 9.947187777125233e-05,
+      "loss": 0.1488,
+      "step": 9400
+    },
+    {
+      "epoch": 0.25175236770292686,
+      "grad_norm": 2.5936684608459473,
+      "learning_rate": 9.946947820079377e-05,
+      "loss": 0.107,
+      "step": 9410
+    },
+    {
+      "epoch": 0.2520199047568088,
+      "grad_norm": 2.826093912124634,
+      "learning_rate": 9.946707322042747e-05,
+      "loss": 0.2034,
+      "step": 9420
+    },
+    {
+      "epoch": 0.2522874418106908,
+      "grad_norm": 4.422873497009277,
+      "learning_rate": 9.94646628304164e-05,
+      "loss": 0.1187,
+      "step": 9430
+    },
+    {
+      "epoch": 0.25255497886457273,
+      "grad_norm": 0.9721447825431824,
+      "learning_rate": 9.946224703102418e-05,
+      "loss": 0.1015,
+      "step": 9440
+    },
+    {
+      "epoch": 0.2528225159184547,
+      "grad_norm": 1.8062744140625,
+      "learning_rate": 9.945982582251498e-05,
+      "loss": 0.1403,
+      "step": 9450
+    },
+    {
+      "epoch": 0.25309005297233667,
+      "grad_norm": 2.7548739910125732,
+      "learning_rate": 9.94573992051536e-05,
+      "loss": 0.1504,
+      "step": 9460
+    },
+    {
+      "epoch": 0.2533575900262186,
+      "grad_norm": 3.774507761001587,
+      "learning_rate": 9.94549671792054e-05,
+      "loss": 0.1479,
+      "step": 9470
+    },
+    {
+      "epoch": 0.2536251270801006,
+      "grad_norm": 2.7577037811279297,
+      "learning_rate": 9.945252974493635e-05,
+      "loss": 0.1606,
+      "step": 9480
+    },
+    {
+      "epoch": 0.25389266413398254,
+      "grad_norm": 1.0760810375213623,
+      "learning_rate": 9.9450086902613e-05,
+      "loss": 0.1778,
+      "step": 9490
+    },
+    {
+      "epoch": 0.25416020118786453,
+      "grad_norm": 1.6968427896499634,
+      "learning_rate": 9.944763865250248e-05,
+      "loss": 0.083,
+      "step": 9500
+    },
+    {
+      "epoch": 0.25442773824174647,
+      "grad_norm": 2.2191667556762695,
+      "learning_rate": 9.944518499487254e-05,
+      "loss": 0.1241,
+      "step": 9510
+    },
+    {
+      "epoch": 0.25469527529562846,
+      "grad_norm": 2.3702619075775146,
+      "learning_rate": 9.944272592999151e-05,
+      "loss": 0.208,
+      "step": 9520
+    },
+    {
+      "epoch": 0.2549628123495104,
+      "grad_norm": 2.654205560684204,
+      "learning_rate": 9.94402614581283e-05,
+      "loss": 0.1177,
+      "step": 9530
+    },
+    {
+      "epoch": 0.25523034940339234,
+      "grad_norm": 2.2863264083862305,
+      "learning_rate": 9.943779157955244e-05,
+      "loss": 0.1124,
+      "step": 9540
+    },
+    {
+      "epoch": 0.25549788645727434,
+      "grad_norm": 1.897598147392273,
+      "learning_rate": 9.943531629453403e-05,
+      "loss": 0.1046,
+      "step": 9550
+    },
+    {
+      "epoch": 0.2557654235111563,
+      "grad_norm": 1.5324757099151611,
+      "learning_rate": 9.943283560334375e-05,
+      "loss": 0.0783,
+      "step": 9560
+    },
+    {
+      "epoch": 0.25603296056503827,
+      "grad_norm": 2.3164193630218506,
+      "learning_rate": 9.943034950625288e-05,
+      "loss": 0.0868,
+      "step": 9570
+    },
+    {
+      "epoch": 0.2563004976189202,
+      "grad_norm": 1.761932373046875,
+      "learning_rate": 9.942785800353332e-05,
+      "loss": 0.0936,
+      "step": 9580
+    },
+    {
+      "epoch": 0.2565680346728022,
+      "grad_norm": 2.6939728260040283,
+      "learning_rate": 9.942536109545751e-05,
+      "loss": 0.1259,
+      "step": 9590
+    },
+    {
+      "epoch": 0.25683557172668414,
+      "grad_norm": 1.8386049270629883,
+      "learning_rate": 9.942285878229853e-05,
+      "loss": 0.1812,
+      "step": 9600
+    },
+    {
+      "epoch": 0.2571031087805661,
+      "grad_norm": 2.866825580596924,
+      "learning_rate": 9.942035106433001e-05,
+      "loss": 0.1252,
+      "step": 9610
+    },
+    {
+      "epoch": 0.2573706458344481,
+      "grad_norm": 2.841935396194458,
+      "learning_rate": 9.94178379418262e-05,
+      "loss": 0.2042,
+      "step": 9620
+    },
+    {
+      "epoch": 0.25763818288833,
+      "grad_norm": 1.6583850383758545,
+      "learning_rate": 9.941531941506194e-05,
+      "loss": 0.1136,
+      "step": 9630
+    },
+    {
+      "epoch": 0.257905719942212,
+      "grad_norm": 1.123732328414917,
+      "learning_rate": 9.941279548431263e-05,
+      "loss": 0.1381,
+      "step": 9640
+    },
+    {
+      "epoch": 0.25817325699609395,
+      "grad_norm": 1.0111355781555176,
+      "learning_rate": 9.941026614985431e-05,
+      "loss": 0.1385,
+      "step": 9650
+    },
+    {
+      "epoch": 0.25844079404997594,
+      "grad_norm": 1.3635011911392212,
+      "learning_rate": 9.940773141196357e-05,
+      "loss": 0.0943,
+      "step": 9660
+    },
+    {
+      "epoch": 0.2587083311038579,
+      "grad_norm": 2.9066855907440186,
+      "learning_rate": 9.94051912709176e-05,
+      "loss": 0.1275,
+      "step": 9670
+    },
+    {
+      "epoch": 0.2589758681577399,
+      "grad_norm": 3.8228259086608887,
+      "learning_rate": 9.940264572699421e-05,
+      "loss": 0.2593,
+      "step": 9680
+    },
+    {
+      "epoch": 0.2592434052116218,
+      "grad_norm": 3.156449317932129,
+      "learning_rate": 9.940009478047174e-05,
+      "loss": 0.1571,
+      "step": 9690
+    },
+    {
+      "epoch": 0.25951094226550375,
+      "grad_norm": 3.5937435626983643,
+      "learning_rate": 9.939753843162918e-05,
+      "loss": 0.1542,
+      "step": 9700
+    },
+    {
+      "epoch": 0.25977847931938575,
+      "grad_norm": 4.038090705871582,
+      "learning_rate": 9.939497668074609e-05,
+      "loss": 0.1779,
+      "step": 9710
+    },
+    {
+      "epoch": 0.2600460163732677,
+      "grad_norm": 3.501190662384033,
+      "learning_rate": 9.93924095281026e-05,
+      "loss": 0.1008,
+      "step": 9720
+    },
+    {
+      "epoch": 0.2603135534271497,
+      "grad_norm": 4.813151836395264,
+      "learning_rate": 9.938983697397948e-05,
+      "loss": 0.1404,
+      "step": 9730
+    },
+    {
+      "epoch": 0.2605810904810316,
+      "grad_norm": 1.4127782583236694,
+      "learning_rate": 9.938725901865805e-05,
+      "loss": 0.0893,
+      "step": 9740
+    },
+    {
+      "epoch": 0.2608486275349136,
+      "grad_norm": 1.5778734683990479,
+      "learning_rate": 9.93846756624202e-05,
+      "loss": 0.1089,
+      "step": 9750
+    },
+    {
+      "epoch": 0.26111616458879555,
+      "grad_norm": 4.731590747833252,
+      "learning_rate": 9.938208690554849e-05,
+      "loss": 0.351,
+      "step": 9760
+    },
+    {
+      "epoch": 0.2613837016426775,
+      "grad_norm": 2.675917387008667,
+      "learning_rate": 9.9379492748326e-05,
+      "loss": 0.2352,
+      "step": 9770
+    },
+    {
+      "epoch": 0.2616512386965595,
+      "grad_norm": 2.4547815322875977,
+      "learning_rate": 9.937689319103641e-05,
+      "loss": 0.1274,
+      "step": 9780
+    },
+    {
+      "epoch": 0.2619187757504414,
+      "grad_norm": 3.21757435798645,
+      "learning_rate": 9.937428823396404e-05,
+      "loss": 0.1762,
+      "step": 9790
+    },
+    {
+      "epoch": 0.2621863128043234,
+      "grad_norm": 2.2968409061431885,
+      "learning_rate": 9.937167787739372e-05,
+      "loss": 0.1581,
+      "step": 9800
+    },
+    {
+      "epoch": 0.26245384985820536,
+      "grad_norm": 1.708569049835205,
+      "learning_rate": 9.936906212161095e-05,
+      "loss": 0.0767,
+      "step": 9810
+    },
+    {
+      "epoch": 0.26272138691208735,
+      "grad_norm": 2.7155797481536865,
+      "learning_rate": 9.936644096690176e-05,
+      "loss": 0.1179,
+      "step": 9820
+    },
+    {
+      "epoch": 0.2629889239659693,
+      "grad_norm": 3.6176304817199707,
+      "learning_rate": 9.936381441355282e-05,
+      "loss": 0.187,
+      "step": 9830
+    },
+    {
+      "epoch": 0.26325646101985123,
+      "grad_norm": 0.8038590550422668,
+      "learning_rate": 9.936118246185136e-05,
+      "loss": 0.1407,
+      "step": 9840
+    },
+    {
+      "epoch": 0.2635239980737332,
+      "grad_norm": 2.298049211502075,
+      "learning_rate": 9.935854511208518e-05,
+      "loss": 0.1247,
+      "step": 9850
+    },
+    {
+      "epoch": 0.26379153512761516,
+      "grad_norm": 2.0069191455841064,
+      "learning_rate": 9.935590236454272e-05,
+      "loss": 0.1115,
+      "step": 9860
+    },
+    {
+      "epoch": 0.26405907218149716,
+      "grad_norm": 3.7639517784118652,
+      "learning_rate": 9.935325421951298e-05,
+      "loss": 0.1508,
+      "step": 9870
+    },
+    {
+      "epoch": 0.2643266092353791,
+      "grad_norm": 1.660439372062683,
+      "learning_rate": 9.935060067728557e-05,
+      "loss": 0.1693,
+      "step": 9880
+    },
+    {
+      "epoch": 0.2645941462892611,
+      "grad_norm": 2.1805105209350586,
+      "learning_rate": 9.934794173815067e-05,
+      "loss": 0.1647,
+      "step": 9890
+    },
+    {
+      "epoch": 0.26486168334314303,
+      "grad_norm": 2.4818639755249023,
+      "learning_rate": 9.934527740239906e-05,
+      "loss": 0.0755,
+      "step": 9900
+    },
+    {
+      "epoch": 0.26512922039702497,
+      "grad_norm": 1.1921519041061401,
+      "learning_rate": 9.934260767032209e-05,
+      "loss": 0.1106,
+      "step": 9910
+    },
+    {
+      "epoch": 0.26539675745090696,
+      "grad_norm": 1.1355395317077637,
+      "learning_rate": 9.933993254221172e-05,
+      "loss": 0.079,
+      "step": 9920
+    },
+    {
+      "epoch": 0.2656642945047889,
+      "grad_norm": 3.615402936935425,
+      "learning_rate": 9.933725201836053e-05,
+      "loss": 0.1445,
+      "step": 9930
+    },
+    {
+      "epoch": 0.2659318315586709,
+      "grad_norm": 4.150886535644531,
+      "learning_rate": 9.933456609906162e-05,
+      "loss": 0.1386,
+      "step": 9940
+    },
+    {
+      "epoch": 0.26619936861255283,
+      "grad_norm": 1.4971691370010376,
+      "learning_rate": 9.933187478460875e-05,
+      "loss": 0.1549,
+      "step": 9950
+    },
+    {
+      "epoch": 0.26646690566643483,
+      "grad_norm": 3.2159249782562256,
+      "learning_rate": 9.93291780752962e-05,
+      "loss": 0.1392,
+      "step": 9960
+    },
+    {
+      "epoch": 0.26673444272031677,
+      "grad_norm": 1.2645758390426636,
+      "learning_rate": 9.932647597141893e-05,
+      "loss": 0.0989,
+      "step": 9970
+    },
+    {
+      "epoch": 0.2670019797741987,
+      "grad_norm": 3.0027430057525635,
+      "learning_rate": 9.932376847327239e-05,
+      "loss": 0.1196,
+      "step": 9980
+    },
+    {
+      "epoch": 0.2672695168280807,
+      "grad_norm": 2.862985610961914,
+      "learning_rate": 9.932105558115268e-05,
+      "loss": 0.1013,
+      "step": 9990
+    },
+    {
+      "epoch": 0.26753705388196264,
+      "grad_norm": 2.511793851852417,
+      "learning_rate": 9.931833729535651e-05,
+      "loss": 0.2348,
+      "step": 10000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 100000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}