diff --git "a/checkpoint-30500/trainer_state.json" "b/checkpoint-30500/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-30500/trainer_state.json"
@@ -0,0 +1,21859 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.2103174603174602,
+  "eval_steps": 500,
+  "global_step": 30500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 23.495065689086914,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 8.6232,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 24.87381935119629,
+      "learning_rate": 9e-07,
+      "loss": 8.4514,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 22.63846778869629,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 8.3152,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 20.65507698059082,
+      "learning_rate": 1.9e-06,
+      "loss": 7.9732,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 16.147356033325195,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 7.6879,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 14.252687454223633,
+      "learning_rate": 2.9e-06,
+      "loss": 7.4152,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 12.011382102966309,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 7.2472,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.081536293029785,
+      "learning_rate": 3.9e-06,
+      "loss": 7.2433,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.65556526184082,
+      "learning_rate": 4.4e-06,
+      "loss": 7.2298,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.830799102783203,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 7.19,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.019427299499512,
+      "learning_rate": 5.4e-06,
+      "loss": 7.2362,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.13768482208252,
+      "learning_rate": 5.9e-06,
+      "loss": 7.2804,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.02840805053711,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 7.2345,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.178731918334961,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 7.3014,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.83882999420166,
+      "learning_rate": 7.4e-06,
+      "loss": 7.3875,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.3687162399292,
+      "learning_rate": 7.9e-06,
+      "loss": 7.4546,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.09121036529541,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 7.407,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 12.731450080871582,
+      "learning_rate": 8.9e-06,
+      "loss": 7.4321,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.601162910461426,
+      "learning_rate": 9.4e-06,
+      "loss": 7.4155,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.006213188171387,
+      "learning_rate": 9.900000000000002e-06,
+      "loss": 7.3705,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.79794454574585,
+      "learning_rate": 1.04e-05,
+      "loss": 7.3759,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.281346321105957,
+      "learning_rate": 1.09e-05,
+      "loss": 7.3798,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.605266094207764,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 7.3878,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.446412563323975,
+      "learning_rate": 1.19e-05,
+      "loss": 7.3098,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.6324334144592285,
+      "learning_rate": 1.24e-05,
+      "loss": 7.2995,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.330737113952637,
+      "learning_rate": 1.29e-05,
+      "loss": 7.1887,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.283708572387695,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 7.1184,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.950069904327393,
+      "learning_rate": 1.3900000000000002e-05,
+      "loss": 6.9876,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.844958782196045,
+      "learning_rate": 1.44e-05,
+      "loss": 6.9319,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.499362468719482,
+      "learning_rate": 1.49e-05,
+      "loss": 6.9253,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.45572280883789,
+      "learning_rate": 1.54e-05,
+      "loss": 6.7999,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.108625411987305,
+      "learning_rate": 1.59e-05,
+      "loss": 6.7667,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.131596565246582,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 6.7599,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.2159247398376465,
+      "learning_rate": 1.69e-05,
+      "loss": 6.7379,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.335048675537109,
+      "learning_rate": 1.74e-05,
+      "loss": 6.6553,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.361081600189209,
+      "learning_rate": 1.79e-05,
+      "loss": 6.6905,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.59442138671875,
+      "learning_rate": 1.84e-05,
+      "loss": 6.6356,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.31695032119751,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 6.6538,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.541380882263184,
+      "learning_rate": 1.94e-05,
+      "loss": 6.6873,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 8.056180000305176,
+      "learning_rate": 1.9900000000000003e-05,
+      "loss": 6.6332,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.74859094619751,
+      "learning_rate": 2.04e-05,
+      "loss": 6.5488,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.700525760650635,
+      "learning_rate": 2.09e-05,
+      "loss": 6.611,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.347009181976318,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 6.6621,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.607337951660156,
+      "learning_rate": 2.19e-05,
+      "loss": 6.568,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.816825866699219,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 6.6293,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.984443664550781,
+      "learning_rate": 2.29e-05,
+      "loss": 6.5667,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.995021820068359,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 6.5062,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.876286029815674,
+      "learning_rate": 2.39e-05,
+      "loss": 6.5034,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.953365802764893,
+      "learning_rate": 2.44e-05,
+      "loss": 6.459,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.246520519256592,
+      "learning_rate": 2.4900000000000002e-05,
+      "loss": 6.4359,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 6.133457660675049,
+      "eval_runtime": 422.8798,
+      "eval_samples_per_second": 50.182,
+      "eval_steps_per_second": 0.393,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.1657819747924805,
+      "learning_rate": 2.54e-05,
+      "loss": 6.4372,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.203813552856445,
+      "learning_rate": 2.5900000000000003e-05,
+      "loss": 6.3684,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.829228401184082,
+      "learning_rate": 2.64e-05,
+      "loss": 6.3275,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.641738414764404,
+      "learning_rate": 2.6900000000000003e-05,
+      "loss": 6.2923,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.824960708618164,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 6.3685,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.7754364013671875,
+      "learning_rate": 2.7900000000000004e-05,
+      "loss": 6.3069,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.3608903884887695,
+      "learning_rate": 2.84e-05,
+      "loss": 6.2915,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.551064968109131,
+      "learning_rate": 2.8899999999999998e-05,
+      "loss": 6.1676,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.588657379150391,
+      "learning_rate": 2.94e-05,
+      "loss": 6.2311,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.631275177001953,
+      "learning_rate": 2.9900000000000002e-05,
+      "loss": 6.1824,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.2990288734436035,
+      "learning_rate": 3.04e-05,
+      "loss": 6.1641,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.6439714431762695,
+      "learning_rate": 3.09e-05,
+      "loss": 6.0125,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.402431964874268,
+      "learning_rate": 3.1400000000000004e-05,
+      "loss": 6.1314,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.883937358856201,
+      "learning_rate": 3.19e-05,
+      "loss": 5.9858,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.988973140716553,
+      "learning_rate": 3.24e-05,
+      "loss": 5.9886,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.186301231384277,
+      "learning_rate": 3.29e-05,
+      "loss": 5.9732,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.2982354164123535,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 6.0391,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.037730693817139,
+      "learning_rate": 3.3900000000000004e-05,
+      "loss": 6.0547,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.280949592590332,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 5.8734,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.861902236938477,
+      "learning_rate": 3.49e-05,
+      "loss": 5.8997,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.4033589363098145,
+      "learning_rate": 3.54e-05,
+      "loss": 5.8111,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.376819133758545,
+      "learning_rate": 3.59e-05,
+      "loss": 5.9027,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.777069568634033,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 5.727,
+      "step": 730
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.183162689208984,
+      "learning_rate": 3.69e-05,
+      "loss": 5.733,
+      "step": 740
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.658604621887207,
+      "learning_rate": 3.74e-05,
+      "loss": 5.7018,
+      "step": 750
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.263491153717041,
+      "learning_rate": 3.79e-05,
+      "loss": 5.7315,
+      "step": 760
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.86756706237793,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 5.6637,
+      "step": 770
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.122657299041748,
+      "learning_rate": 3.8900000000000004e-05,
+      "loss": 5.6849,
+      "step": 780
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.6024580001831055,
+      "learning_rate": 3.94e-05,
+      "loss": 5.5835,
+      "step": 790
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.619604587554932,
+      "learning_rate": 3.99e-05,
+      "loss": 5.7,
+      "step": 800
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.7342352867126465,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 5.4825,
+      "step": 810
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.796237468719482,
+      "learning_rate": 4.09e-05,
+      "loss": 5.6496,
+      "step": 820
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.601135730743408,
+      "learning_rate": 4.14e-05,
+      "loss": 5.3526,
+      "step": 830
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.576679229736328,
+      "learning_rate": 4.19e-05,
+      "loss": 5.3757,
+      "step": 840
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.774380683898926,
+      "learning_rate": 4.24e-05,
+      "loss": 5.5164,
+      "step": 850
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.140038967132568,
+      "learning_rate": 4.29e-05,
+      "loss": 5.3124,
+      "step": 860
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.971868515014648,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 5.3387,
+      "step": 870
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.431824207305908,
+      "learning_rate": 4.39e-05,
+      "loss": 5.4363,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.174604892730713,
+      "learning_rate": 4.44e-05,
+      "loss": 5.282,
+      "step": 890
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.313474178314209,
+      "learning_rate": 4.49e-05,
+      "loss": 5.3754,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.066281318664551,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 5.3197,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 7.4433817863464355,
+      "learning_rate": 4.5900000000000004e-05,
+      "loss": 5.3277,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.884599208831787,
+      "learning_rate": 4.64e-05,
+      "loss": 5.0681,
+      "step": 930
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.17655611038208,
+      "learning_rate": 4.69e-05,
+      "loss": 5.1159,
+      "step": 940
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 7.090456485748291,
+      "learning_rate": 4.74e-05,
+      "loss": 5.2264,
+      "step": 950
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.508240699768066,
+      "learning_rate": 4.79e-05,
+      "loss": 5.2113,
+      "step": 960
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.5788445472717285,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 4.9996,
+      "step": 970
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.722221374511719,
+      "learning_rate": 4.89e-05,
+      "loss": 4.9688,
+      "step": 980
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.813933849334717,
+      "learning_rate": 4.94e-05,
+      "loss": 4.8691,
+      "step": 990
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.425748348236084,
+      "learning_rate": 4.99e-05,
+      "loss": 4.9883,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 4.425854682922363,
+      "eval_runtime": 422.6592,
+      "eval_samples_per_second": 50.208,
+      "eval_steps_per_second": 0.393,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.170848369598389,
+      "learning_rate": 4.9999996764541565e-05,
+      "loss": 4.8794,
+      "step": 1010
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.247808933258057,
+      "learning_rate": 4.99999836204931e-05,
+      "loss": 4.9664,
+      "step": 1020
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.27077054977417,
+      "learning_rate": 4.9999960365643774e-05,
+      "loss": 4.7753,
+      "step": 1030
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.284353733062744,
+      "learning_rate": 4.999992700000298e-05,
+      "loss": 4.8909,
+      "step": 1040
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.716318607330322,
+      "learning_rate": 4.999988352358421e-05,
+      "loss": 4.861,
+      "step": 1050
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.448729991912842,
+      "learning_rate": 4.999982993640506e-05,
+      "loss": 4.8776,
+      "step": 1060
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.401464939117432,
+      "learning_rate": 4.9999766238487204e-05,
+      "loss": 4.8507,
+      "step": 1070
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.219631195068359,
+      "learning_rate": 4.999969242985639e-05,
+      "loss": 4.8234,
+      "step": 1080
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.225848197937012,
+      "learning_rate": 4.9999608510542485e-05,
+      "loss": 4.6093,
+      "step": 1090
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.117152690887451,
+      "learning_rate": 4.999951448057941e-05,
+      "loss": 4.786,
+      "step": 1100
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.788968086242676,
+      "learning_rate": 4.999941034000521e-05,
+      "loss": 4.8389,
+      "step": 1110
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.037156581878662,
+      "learning_rate": 4.999929608886198e-05,
+      "loss": 4.7674,
+      "step": 1120
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.567563056945801,
+      "learning_rate": 4.999917172719596e-05,
+      "loss": 4.8247,
+      "step": 1130
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.557987213134766,
+      "learning_rate": 4.999903725505742e-05,
+      "loss": 4.7517,
+      "step": 1140
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.490599632263184,
+      "learning_rate": 4.999889267250075e-05,
+      "loss": 4.6037,
+      "step": 1150
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.000855922698975,
+      "learning_rate": 4.999873797958443e-05,
+      "loss": 4.6077,
+      "step": 1160
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.378462314605713,
+      "learning_rate": 4.999857317637102e-05,
+      "loss": 4.3662,
+      "step": 1170
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.4290924072265625,
+      "learning_rate": 4.999839826292718e-05,
+      "loss": 4.5339,
+      "step": 1180
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.727224826812744,
+      "learning_rate": 4.999821323932363e-05,
+      "loss": 4.5744,
+      "step": 1190
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.893082618713379,
+      "learning_rate": 4.9998018105635214e-05,
+      "loss": 4.5247,
+      "step": 1200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.759618759155273,
+      "learning_rate": 4.999781286194085e-05,
+      "loss": 4.5744,
+      "step": 1210
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.082376003265381,
+      "learning_rate": 4.999759750832355e-05,
+      "loss": 4.6011,
+      "step": 1220
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.305749416351318,
+      "learning_rate": 4.999737204487039e-05,
+      "loss": 4.4587,
+      "step": 1230
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.710574626922607,
+      "learning_rate": 4.999713647167258e-05,
+      "loss": 4.46,
+      "step": 1240
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.753530502319336,
+      "learning_rate": 4.999689078882537e-05,
+      "loss": 4.5674,
+      "step": 1250
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.949398040771484,
+      "learning_rate": 4.999663499642814e-05,
+      "loss": 4.3999,
+      "step": 1260
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.530541896820068,
+      "learning_rate": 4.999636909458434e-05,
+      "loss": 4.533,
+      "step": 1270
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.9646453857421875,
+      "learning_rate": 4.9996093083401495e-05,
+      "loss": 4.374,
+      "step": 1280
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.346386432647705,
+      "learning_rate": 4.9995806962991246e-05,
+      "loss": 4.4758,
+      "step": 1290
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.797152519226074,
+      "learning_rate": 4.9995510733469295e-05,
+      "loss": 4.6275,
+      "step": 1300
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.426751136779785,
+      "learning_rate": 4.999520439495547e-05,
+      "loss": 4.3561,
+      "step": 1310
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.3839030265808105,
+      "learning_rate": 4.9994887947573646e-05,
+      "loss": 4.3134,
+      "step": 1320
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.01127290725708,
+      "learning_rate": 4.9994594501953555e-05,
+      "loss": 4.4336,
+      "step": 1330
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.228607654571533,
+      "learning_rate": 4.999425884807851e-05,
+      "loss": 4.1543,
+      "step": 1340
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.958486080169678,
+      "learning_rate": 4.999391308571788e-05,
+      "loss": 4.3806,
+      "step": 1350
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.816386699676514,
+      "learning_rate": 4.99935572150115e-05,
+      "loss": 4.3759,
+      "step": 1360
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.345638751983643,
+      "learning_rate": 4.9993191236103295e-05,
+      "loss": 4.5759,
+      "step": 1370
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.738880634307861,
+      "learning_rate": 4.9992815149141294e-05,
+      "loss": 4.3274,
+      "step": 1380
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.694160461425781,
+      "learning_rate": 4.9992428954277573e-05,
+      "loss": 4.376,
+      "step": 1390
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.050902843475342,
+      "learning_rate": 4.999203265166834e-05,
+      "loss": 4.1629,
+      "step": 1400
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 7.09380578994751,
+      "learning_rate": 4.9991626241473876e-05,
+      "loss": 4.3638,
+      "step": 1410
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.538166046142578,
+      "learning_rate": 4.999120972385853e-05,
+      "loss": 4.1318,
+      "step": 1420
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.052664279937744,
+      "learning_rate": 4.9990783098990765e-05,
+      "loss": 4.1856,
+      "step": 1430
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.707901477813721,
+      "learning_rate": 4.9990346367043114e-05,
+      "loss": 4.3361,
+      "step": 1440
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.3822150230407715,
+      "learning_rate": 4.998989952819223e-05,
+      "loss": 4.301,
+      "step": 1450
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.586170196533203,
+      "learning_rate": 4.998944258261879e-05,
+      "loss": 4.3131,
+      "step": 1460
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.62904691696167,
+      "learning_rate": 4.998897553050763e-05,
+      "loss": 4.3357,
+      "step": 1470
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.669075965881348,
+      "learning_rate": 4.9988498372047633e-05,
+      "loss": 4.2434,
+      "step": 1480
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.776549339294434,
+      "learning_rate": 4.998801110743177e-05,
+      "loss": 4.0441,
+      "step": 1490
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.442574977874756,
+      "learning_rate": 4.9987513736857106e-05,
+      "loss": 4.0931,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 3.676598310470581,
+      "eval_runtime": 423.4536,
+      "eval_samples_per_second": 50.114,
+      "eval_steps_per_second": 0.392,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.529382705688477,
+      "learning_rate": 4.998700626052481e-05,
+      "loss": 4.1739,
+      "step": 1510
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.5605387687683105,
+      "learning_rate": 4.99864886786401e-05,
+      "loss": 4.1861,
+      "step": 1520
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.168206214904785,
+      "learning_rate": 4.998596099141231e-05,
+      "loss": 4.0245,
+      "step": 1530
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.634309768676758,
+      "learning_rate": 4.998542319905486e-05,
+      "loss": 4.2131,
+      "step": 1540
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.191194534301758,
+      "learning_rate": 4.998487530178525e-05,
+      "loss": 4.163,
+      "step": 1550
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.52882719039917,
+      "learning_rate": 4.998431729982506e-05,
+      "loss": 4.0668,
+      "step": 1560
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.614863395690918,
+      "learning_rate": 4.998374919339998e-05,
+      "loss": 4.2501,
+      "step": 1570
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.7852373123168945,
+      "learning_rate": 4.9983170982739745e-05,
+      "loss": 4.207,
+      "step": 1580
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 7.439071178436279,
+      "learning_rate": 4.998258266807822e-05,
+      "loss": 4.0462,
+      "step": 1590
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.510477542877197,
+      "learning_rate": 4.998198424965333e-05,
+      "loss": 4.277,
+      "step": 1600
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.421988487243652,
+      "learning_rate": 4.9981375727707114e-05,
+      "loss": 4.0289,
+      "step": 1610
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.632848739624023,
+      "learning_rate": 4.998075710248565e-05,
+      "loss": 4.2922,
+      "step": 1620
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.031602382659912,
+      "learning_rate": 4.9980128374239156e-05,
+      "loss": 3.9845,
+      "step": 1630
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.3867669105529785,
+      "learning_rate": 4.997948954322189e-05,
+      "loss": 4.1095,
+      "step": 1640
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.150663375854492,
+      "learning_rate": 4.997884060969222e-05,
+      "loss": 3.9247,
+      "step": 1650
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.717851638793945,
+      "learning_rate": 4.99781815739126e-05,
+      "loss": 4.1911,
+      "step": 1660
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.766101837158203,
+      "learning_rate": 4.9977512436149563e-05,
+      "loss": 3.9194,
+      "step": 1670
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.316621780395508,
+      "learning_rate": 4.9976833196673735e-05,
+      "loss": 4.101,
+      "step": 1680
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.107331275939941,
+      "learning_rate": 4.9976143855759814e-05,
+      "loss": 4.0706,
+      "step": 1690
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.533346652984619,
+      "learning_rate": 4.99754444136866e-05,
+      "loss": 4.0164,
+      "step": 1700
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.967229843139648,
+      "learning_rate": 4.9974734870736975e-05,
+      "loss": 4.1229,
+      "step": 1710
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.502392768859863,
+      "learning_rate": 4.997401522719788e-05,
+      "loss": 4.0456,
+      "step": 1720
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.6358962059021,
+      "learning_rate": 4.997328548336038e-05,
+      "loss": 3.9432,
+      "step": 1730
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.525423049926758,
+      "learning_rate": 4.99725456395196e-05,
+      "loss": 4.05,
+      "step": 1740
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.675175666809082,
+      "learning_rate": 4.997179569597475e-05,
+      "loss": 3.8145,
+      "step": 1750
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.409067630767822,
+      "learning_rate": 4.9971035653029153e-05,
+      "loss": 3.914,
+      "step": 1760
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.812373161315918,
+      "learning_rate": 4.997026551099017e-05,
+      "loss": 3.9817,
+      "step": 1770
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.165591239929199,
+      "learning_rate": 4.996948527016929e-05,
+      "loss": 3.9414,
+      "step": 1780
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.0605034828186035,
+      "learning_rate": 4.996869493088205e-05,
+      "loss": 3.8139,
+      "step": 1790
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.348527908325195,
+      "learning_rate": 4.996789449344812e-05,
+      "loss": 3.7905,
+      "step": 1800
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.135634422302246,
+      "learning_rate": 4.9967083958191187e-05,
+      "loss": 3.7834,
+      "step": 1810
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 8.48004150390625,
+      "learning_rate": 4.996626332543907e-05,
+      "loss": 3.7551,
+      "step": 1820
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.518482208251953,
+      "learning_rate": 4.996543259552367e-05,
+      "loss": 3.8619,
+      "step": 1830
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.7253737449646,
+      "learning_rate": 4.996459176878095e-05,
+      "loss": 3.7978,
+      "step": 1840
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.030267715454102,
+      "learning_rate": 4.9963740845550965e-05,
+      "loss": 3.9309,
+      "step": 1850
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.561445713043213,
+      "learning_rate": 4.996287982617787e-05,
+      "loss": 3.9761,
+      "step": 1860
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.555076599121094,
+      "learning_rate": 4.996200871100988e-05,
+      "loss": 3.8587,
+      "step": 1870
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.502940654754639,
+      "learning_rate": 4.9961127500399295e-05,
+      "loss": 3.847,
+      "step": 1880
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.557964324951172,
+      "learning_rate": 4.996023619470252e-05,
+      "loss": 3.6117,
+      "step": 1890
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.37682580947876,
+      "learning_rate": 4.9959334794280014e-05,
+      "loss": 3.7581,
+      "step": 1900
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.152722358703613,
+      "learning_rate": 4.9958423299496344e-05,
+      "loss": 3.8454,
+      "step": 1910
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.4067301750183105,
+      "learning_rate": 4.995750171072014e-05,
+      "loss": 3.9353,
+      "step": 1920
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.564638614654541,
+      "learning_rate": 4.9956570028324124e-05,
+      "loss": 3.738,
+      "step": 1930
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.320713520050049,
+      "learning_rate": 4.99556282526851e-05,
+      "loss": 3.5365,
+      "step": 1940
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.384575843811035,
+      "learning_rate": 4.995467638418395e-05,
+      "loss": 3.8153,
+      "step": 1950
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.424959659576416,
+      "learning_rate": 4.995371442320565e-05,
+      "loss": 3.779,
+      "step": 1960
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.928106784820557,
+      "learning_rate": 4.9952742370139224e-05,
+      "loss": 3.7395,
+      "step": 1970
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.09404182434082,
+      "learning_rate": 4.995176022537784e-05,
+      "loss": 3.757,
+      "step": 1980
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 8.332275390625,
+      "learning_rate": 4.995076798931867e-05,
+      "loss": 3.8866,
+      "step": 1990
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.44631814956665,
+      "learning_rate": 4.994976566236303e-05,
+      "loss": 3.8395,
+      "step": 2000
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 3.2655727863311768,
+      "eval_runtime": 423.1622,
+      "eval_samples_per_second": 50.149,
+      "eval_steps_per_second": 0.392,
+      "step": 2000
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.428518295288086,
+      "learning_rate": 4.994875324491629e-05,
+      "loss": 3.7103,
+      "step": 2010
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.11751651763916,
+      "learning_rate": 4.9947730737387885e-05,
+      "loss": 3.8224,
+      "step": 2020
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.404431343078613,
+      "learning_rate": 4.994669814019138e-05,
+      "loss": 3.9208,
+      "step": 2030
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.487029075622559,
+      "learning_rate": 4.994565545374438e-05,
+      "loss": 3.7412,
+      "step": 2040
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.979548454284668,
+      "learning_rate": 4.9944602678468575e-05,
+      "loss": 3.6008,
+      "step": 2050
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.379942417144775,
+      "learning_rate": 4.994353981478974e-05,
+      "loss": 3.6077,
+      "step": 2060
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.666346549987793,
+      "learning_rate": 4.994246686313774e-05,
+      "loss": 3.8157,
+      "step": 2070
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.252011299133301,
+      "learning_rate": 4.99413838239465e-05,
+      "loss": 3.6929,
+      "step": 2080
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.579212188720703,
+      "learning_rate": 4.9940290697654045e-05,
+      "loss": 3.9075,
+      "step": 2090
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.747458457946777,
+      "learning_rate": 4.993918748470246e-05,
+      "loss": 3.5528,
+      "step": 2100
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.356452465057373,
+      "learning_rate": 4.993818596932118e-05,
+      "loss": 3.4987,
+      "step": 2110
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.761812686920166,
+      "learning_rate": 4.993706359294986e-05,
+      "loss": 3.7627,
+      "step": 2120
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.732122898101807,
+      "learning_rate": 4.993593113122456e-05,
+      "loss": 3.6106,
+      "step": 2130
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 11.450267791748047,
+      "learning_rate": 4.993478858460329e-05,
+      "loss": 3.4716,
+      "step": 2140
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.774083614349365,
+      "learning_rate": 4.993363595354813e-05,
+      "loss": 3.5741,
+      "step": 2150
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.652318000793457,
+      "learning_rate": 4.9932473238525247e-05,
+      "loss": 3.676,
+      "step": 2160
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.903106689453125,
+      "learning_rate": 4.993130044000487e-05,
+      "loss": 3.6334,
+      "step": 2170
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.358557224273682,
+      "learning_rate": 4.993011755846132e-05,
+      "loss": 3.5592,
+      "step": 2180
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.509594440460205,
+      "learning_rate": 4.9928924594372995e-05,
+      "loss": 3.6512,
+      "step": 2190
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.318192958831787,
+      "learning_rate": 4.9927721548222374e-05,
+      "loss": 3.3691,
+      "step": 2200
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.848221778869629,
+      "learning_rate": 4.9926508420496e-05,
+      "loss": 3.3721,
+      "step": 2210
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.9782490730285645,
+      "learning_rate": 4.992528521168449e-05,
+      "loss": 3.5859,
+      "step": 2220
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.406031608581543,
+      "learning_rate": 4.992405192228258e-05,
+      "loss": 3.6206,
+      "step": 2230
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.102032661437988,
+      "learning_rate": 4.992280855278903e-05,
+      "loss": 3.4615,
+      "step": 2240
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.5005035400390625,
+      "learning_rate": 4.992155510370669e-05,
+      "loss": 3.507,
+      "step": 2250
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 8.336762428283691,
+      "learning_rate": 4.9920291575542515e-05,
+      "loss": 3.3897,
+      "step": 2260
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 7.851747989654541,
+      "learning_rate": 4.9919017968807514e-05,
+      "loss": 3.5232,
+      "step": 2270
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.816250324249268,
+      "learning_rate": 4.991773428401676e-05,
+      "loss": 3.5638,
+      "step": 2280
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.746279716491699,
+      "learning_rate": 4.9916440521689433e-05,
+      "loss": 3.5446,
+      "step": 2290
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.8899922370910645,
+      "learning_rate": 4.9915136682348774e-05,
+      "loss": 3.3533,
+      "step": 2300
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.160922050476074,
+      "learning_rate": 4.9913822766522076e-05,
+      "loss": 3.4222,
+      "step": 2310
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.936460494995117,
+      "learning_rate": 4.991249877474075e-05,
+      "loss": 3.5763,
+      "step": 2320
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 16.517559051513672,
+      "learning_rate": 4.991116470754025e-05,
+      "loss": 3.5324,
+      "step": 2330
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.840503692626953,
+      "learning_rate": 4.9909820565460116e-05,
+      "loss": 3.5764,
+      "step": 2340
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.952908515930176,
+      "learning_rate": 4.990846634904398e-05,
+      "loss": 3.4369,
+      "step": 2350
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.108431339263916,
+      "learning_rate": 4.9907102058839504e-05,
+      "loss": 3.4518,
+      "step": 2360
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.068904399871826,
+      "learning_rate": 4.990572769539848e-05,
+      "loss": 3.5594,
+      "step": 2370
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.512117385864258,
+      "learning_rate": 4.990434325927671e-05,
+      "loss": 3.6366,
+      "step": 2380
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.916763782501221,
+      "learning_rate": 4.990294875103413e-05,
+      "loss": 3.4344,
+      "step": 2390
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.244757175445557,
+      "learning_rate": 4.990154417123472e-05,
+      "loss": 3.5858,
+      "step": 2400
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.071867942810059,
+      "learning_rate": 4.990012952044654e-05,
+      "loss": 3.5888,
+      "step": 2410
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.984373092651367,
+      "learning_rate": 4.98987047992417e-05,
+      "loss": 3.5084,
+      "step": 2420
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.062913417816162,
+      "learning_rate": 4.989727000819644e-05,
+      "loss": 3.4169,
+      "step": 2430
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.671230792999268,
+      "learning_rate": 4.9895825147891e-05,
+      "loss": 3.3354,
+      "step": 2440
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.235222339630127,
+      "learning_rate": 4.989437021890975e-05,
+      "loss": 3.4756,
+      "step": 2450
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.888786792755127,
+      "learning_rate": 4.98929052218411e-05,
+      "loss": 3.3409,
+      "step": 2460
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.014183044433594,
+      "learning_rate": 4.989143015727754e-05,
+      "loss": 3.4246,
+      "step": 2470
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 8.75373649597168,
+      "learning_rate": 4.988994502581565e-05,
+      "loss": 3.2431,
+      "step": 2480
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.800814628601074,
+      "learning_rate": 4.988844982805605e-05,
+      "loss": 3.4702,
+      "step": 2490
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.22312068939209,
+      "learning_rate": 4.9886944564603444e-05,
+      "loss": 3.2994,
+      "step": 2500
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 2.9097418785095215,
+      "eval_runtime": 423.0547,
+      "eval_samples_per_second": 50.161,
+      "eval_steps_per_second": 0.392,
+      "step": 2500
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.41413688659668,
+      "learning_rate": 4.988542923606663e-05,
+      "loss": 3.5138,
+      "step": 2510
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.5836968421936035,
+      "learning_rate": 4.988390384305843e-05,
+      "loss": 3.4161,
+      "step": 2520
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.957060813903809,
+      "learning_rate": 4.9882368386195776e-05,
+      "loss": 3.4227,
+      "step": 2530
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.718446731567383,
+      "learning_rate": 4.988082286609965e-05,
+      "loss": 3.3632,
+      "step": 2540
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.273780822753906,
+      "learning_rate": 4.9879267283395125e-05,
+      "loss": 3.4018,
+      "step": 2550
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.000144004821777,
+      "learning_rate": 4.9877701638711314e-05,
+      "loss": 3.2865,
+      "step": 2560
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.369386672973633,
+      "learning_rate": 4.987612593268142e-05,
+      "loss": 3.2916,
+      "step": 2570
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.275691032409668,
+      "learning_rate": 4.98745401659427e-05,
+      "loss": 3.3678,
+      "step": 2580
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.77783727645874,
+      "learning_rate": 4.9872944339136503e-05,
+      "loss": 3.2567,
+      "step": 2590
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.3765435218811035,
+      "learning_rate": 4.987133845290822e-05,
+      "loss": 3.3649,
+      "step": 2600
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.956243991851807,
+      "learning_rate": 4.9869722507907343e-05,
+      "loss": 3.428,
+      "step": 2610
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.933529376983643,
+      "learning_rate": 4.9868096504787395e-05,
+      "loss": 3.0448,
+      "step": 2620
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.319052696228027,
+      "learning_rate": 4.9866460444205983e-05,
+      "loss": 3.3518,
+      "step": 2630
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.592201232910156,
+      "learning_rate": 4.986481432682479e-05,
+      "loss": 3.1236,
+      "step": 2640
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.243892192840576,
+      "learning_rate": 4.986315815330956e-05,
+      "loss": 3.2447,
+      "step": 2650
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.197099208831787,
+      "learning_rate": 4.98614919243301e-05,
+      "loss": 3.3031,
+      "step": 2660
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.438327312469482,
+      "learning_rate": 4.98598156405603e-05,
+      "loss": 3.5433,
+      "step": 2670
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.134957313537598,
+      "learning_rate": 4.985812930267808e-05,
+      "loss": 3.3124,
+      "step": 2680
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.727309703826904,
+      "learning_rate": 4.985643291136546e-05,
+      "loss": 3.2169,
+      "step": 2690
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.680230617523193,
+      "learning_rate": 4.985472646730852e-05,
+      "loss": 3.0445,
+      "step": 2700
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.9644598960876465,
+      "learning_rate": 4.985300997119739e-05,
+      "loss": 3.1588,
+      "step": 2710
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.430914402008057,
+      "learning_rate": 4.985128342372629e-05,
+      "loss": 3.2385,
+      "step": 2720
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.565753936767578,
+      "learning_rate": 4.9849546825593476e-05,
+      "loss": 3.2654,
+      "step": 2730
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.126991271972656,
+      "learning_rate": 4.9847800177501305e-05,
+      "loss": 3.3024,
+      "step": 2740
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.406001567840576,
+      "learning_rate": 4.984604348015616e-05,
+      "loss": 3.219,
+      "step": 2750
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.55892276763916,
+      "learning_rate": 4.984427673426851e-05,
+      "loss": 3.3882,
+      "step": 2760
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.547699451446533,
+      "learning_rate": 4.9842499940552886e-05,
+      "loss": 3.2334,
+      "step": 2770
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.514155864715576,
+      "learning_rate": 4.9840713099727885e-05,
+      "loss": 3.2041,
+      "step": 2780
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.4210991859436035,
+      "learning_rate": 4.983891621251616e-05,
+      "loss": 3.1746,
+      "step": 2790
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.854736328125,
+      "learning_rate": 4.983710927964442e-05,
+      "loss": 3.1961,
+      "step": 2800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.277487754821777,
+      "learning_rate": 4.9835292301843476e-05,
+      "loss": 2.9688,
+      "step": 2810
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.925424098968506,
+      "learning_rate": 4.983346527984813e-05,
+      "loss": 3.2315,
+      "step": 2820
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 7.818734169006348,
+      "learning_rate": 4.983162821439733e-05,
+      "loss": 3.325,
+      "step": 2830
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.312294006347656,
+      "learning_rate": 4.982978110623402e-05,
+      "loss": 3.2141,
+      "step": 2840
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.804452419281006,
+      "learning_rate": 4.982792395610524e-05,
+      "loss": 3.1214,
+      "step": 2850
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.256209850311279,
+      "learning_rate": 4.9826056764762074e-05,
+      "loss": 3.2357,
+      "step": 2860
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.713292121887207,
+      "learning_rate": 4.982417953295968e-05,
+      "loss": 3.0448,
+      "step": 2870
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.762032985687256,
+      "learning_rate": 4.982229226145727e-05,
+      "loss": 3.1855,
+      "step": 2880
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.245543956756592,
+      "learning_rate": 4.982039495101812e-05,
+      "loss": 3.1024,
+      "step": 2890
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.456753730773926,
+      "learning_rate": 4.9818487602409556e-05,
+      "loss": 3.2703,
+      "step": 2900
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.831927299499512,
+      "learning_rate": 4.981657021640298e-05,
+      "loss": 3.2172,
+      "step": 2910
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.4401350021362305,
+      "learning_rate": 4.981464279377384e-05,
+      "loss": 3.2972,
+      "step": 2920
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.1732892990112305,
+      "learning_rate": 4.981270533530164e-05,
+      "loss": 3.1305,
+      "step": 2930
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.929070472717285,
+      "learning_rate": 4.9810757841769976e-05,
+      "loss": 3.0932,
+      "step": 2940
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.50985860824585,
+      "learning_rate": 4.980880031396645e-05,
+      "loss": 3.1417,
+      "step": 2950
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.071355819702148,
+      "learning_rate": 4.980683275268277e-05,
+      "loss": 3.1796,
+      "step": 2960
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.182791709899902,
+      "learning_rate": 4.980485515871467e-05,
+      "loss": 3.268,
+      "step": 2970
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.4410786628723145,
+      "learning_rate": 4.980286753286195e-05,
+      "loss": 3.2922,
+      "step": 2980
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 22.151193618774414,
+      "learning_rate": 4.9800869875928475e-05,
+      "loss": 3.0482,
+      "step": 2990
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.808634281158447,
+      "learning_rate": 4.9798862188722164e-05,
+      "loss": 3.1515,
+      "step": 3000
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 2.6660187244415283,
+      "eval_runtime": 423.2518,
+      "eval_samples_per_second": 50.138,
+      "eval_steps_per_second": 0.392,
+      "step": 3000
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 9.806035995483398,
+      "learning_rate": 4.979684447205499e-05,
+      "loss": 3.0872,
+      "step": 3010
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.456057071685791,
+      "learning_rate": 4.9794816726742976e-05,
+      "loss": 2.882,
+      "step": 3020
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.348060607910156,
+      "learning_rate": 4.979277895360622e-05,
+      "loss": 3.1984,
+      "step": 3030
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.867720603942871,
+      "learning_rate": 4.979073115346885e-05,
+      "loss": 3.0328,
+      "step": 3040
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.720703601837158,
+      "learning_rate": 4.9788673327159076e-05,
+      "loss": 3.1852,
+      "step": 3050
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.027235984802246,
+      "learning_rate": 4.978660547550914e-05,
+      "loss": 3.1082,
+      "step": 3060
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.131529808044434,
+      "learning_rate": 4.978452759935535e-05,
+      "loss": 3.0318,
+      "step": 3070
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.773056983947754,
+      "learning_rate": 4.978243969953806e-05,
+      "loss": 3.3087,
+      "step": 3080
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.788400650024414,
+      "learning_rate": 4.9780341776901696e-05,
+      "loss": 2.9589,
+      "step": 3090
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.815254211425781,
+      "learning_rate": 4.9778233832294724e-05,
+      "loss": 3.1083,
+      "step": 3100
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.483798027038574,
+      "learning_rate": 4.9776115866569654e-05,
+      "loss": 3.1023,
+      "step": 3110
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.394237995147705,
+      "learning_rate": 4.9773987880583075e-05,
+      "loss": 3.0322,
+      "step": 3120
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.961243152618408,
+      "learning_rate": 4.97718498751956e-05,
+      "loss": 3.0204,
+      "step": 3130
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.827451229095459,
+      "learning_rate": 4.976970185127191e-05,
+      "loss": 3.1435,
+      "step": 3140
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.503066539764404,
+      "learning_rate": 4.9767543809680744e-05,
+      "loss": 3.0807,
+      "step": 3150
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.9387288093566895,
+      "learning_rate": 4.9765375751294874e-05,
+      "loss": 2.7987,
+      "step": 3160
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.066097259521484,
+      "learning_rate": 4.9763197676991134e-05,
+      "loss": 3.0099,
+      "step": 3170
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 7.227752685546875,
+      "learning_rate": 4.976100958765042e-05,
+      "loss": 2.9368,
+      "step": 3180
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.03649377822876,
+      "learning_rate": 4.975881148415765e-05,
+      "loss": 2.9237,
+      "step": 3190
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.1522979736328125,
+      "learning_rate": 4.9756603367401814e-05,
+      "loss": 2.96,
+      "step": 3200
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.068181037902832,
+      "learning_rate": 4.975438523827594e-05,
+      "loss": 2.9398,
+      "step": 3210
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.104649066925049,
+      "learning_rate": 4.975215709767712e-05,
+      "loss": 2.781,
+      "step": 3220
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 7.2377471923828125,
+      "learning_rate": 4.974991894650649e-05,
+      "loss": 3.2773,
+      "step": 3230
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.730634689331055,
+      "learning_rate": 4.974767078566922e-05,
+      "loss": 3.0184,
+      "step": 3240
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.900280952453613,
+      "learning_rate": 4.974541261607454e-05,
+      "loss": 2.9432,
+      "step": 3250
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.755638122558594,
+      "learning_rate": 4.974314443863573e-05,
+      "loss": 2.9988,
+      "step": 3260
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.039461612701416,
+      "learning_rate": 4.974086625427011e-05,
+      "loss": 3.0655,
+      "step": 3270
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.955098628997803,
+      "learning_rate": 4.9738578063899066e-05,
+      "loss": 2.822,
+      "step": 3280
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.892226696014404,
+      "learning_rate": 4.9736279868447996e-05,
+      "loss": 2.9688,
+      "step": 3290
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.9840497970581055,
+      "learning_rate": 4.9733971668846376e-05,
+      "loss": 2.9008,
+      "step": 3300
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.418862819671631,
+      "learning_rate": 4.973165346602771e-05,
+      "loss": 2.9117,
+      "step": 3310
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.666804790496826,
+      "learning_rate": 4.972932526092956e-05,
+      "loss": 3.0392,
+      "step": 3320
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.902047634124756,
+      "learning_rate": 4.972698705449353e-05,
+      "loss": 3.0601,
+      "step": 3330
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.236441612243652,
+      "learning_rate": 4.972463884766525e-05,
+      "loss": 2.961,
+      "step": 3340
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.548509120941162,
+      "learning_rate": 4.972228064139444e-05,
+      "loss": 2.8966,
+      "step": 3350
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.943577766418457,
+      "learning_rate": 4.9719912436634796e-05,
+      "loss": 3.0408,
+      "step": 3360
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.531796455383301,
+      "learning_rate": 4.971753423434413e-05,
+      "loss": 2.8598,
+      "step": 3370
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.425514221191406,
+      "learning_rate": 4.971514603548425e-05,
+      "loss": 3.0003,
+      "step": 3380
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.9626593589782715,
+      "learning_rate": 4.9712747841021024e-05,
+      "loss": 3.0137,
+      "step": 3390
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.935604572296143,
+      "learning_rate": 4.971033965192435e-05,
+      "loss": 2.9509,
+      "step": 3400
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.610740661621094,
+      "learning_rate": 4.9708163737130644e-05,
+      "loss": 3.0231,
+      "step": 3410
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.248142242431641,
+      "learning_rate": 4.970573656091701e-05,
+      "loss": 2.7345,
+      "step": 3420
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.92423677444458,
+      "learning_rate": 4.9703299392905516e-05,
+      "loss": 2.9733,
+      "step": 3430
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.833972454071045,
+      "learning_rate": 4.9700852234081843e-05,
+      "loss": 2.8882,
+      "step": 3440
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.712262153625488,
+      "learning_rate": 4.96983950854357e-05,
+      "loss": 2.9849,
+      "step": 3450
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.70693826675415,
+      "learning_rate": 4.969592794796083e-05,
+      "loss": 2.7034,
+      "step": 3460
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.603397369384766,
+      "learning_rate": 4.969345082265503e-05,
+      "loss": 2.9441,
+      "step": 3470
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.455355644226074,
+      "learning_rate": 4.9690963710520125e-05,
+      "loss": 3.0818,
+      "step": 3480
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.410728454589844,
+      "learning_rate": 4.968846661256199e-05,
+      "loss": 2.7837,
+      "step": 3490
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.591252326965332,
+      "learning_rate": 4.9685959529790526e-05,
+      "loss": 2.8103,
+      "step": 3500
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 2.493670701980591,
+      "eval_runtime": 423.4603,
+      "eval_samples_per_second": 50.113,
+      "eval_steps_per_second": 0.392,
+      "step": 3500
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.1827216148376465,
+      "learning_rate": 4.9683442463219674e-05,
+      "loss": 2.9208,
+      "step": 3510
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.817008018493652,
+      "learning_rate": 4.968091541386742e-05,
+      "loss": 2.9904,
+      "step": 3520
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.470141887664795,
+      "learning_rate": 4.96783783827558e-05,
+      "loss": 2.7532,
+      "step": 3530
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.129313945770264,
+      "learning_rate": 4.967583137091085e-05,
+      "loss": 3.1164,
+      "step": 3540
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.951664924621582,
+      "learning_rate": 4.967327437936268e-05,
+      "loss": 3.0038,
+      "step": 3550
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.184429168701172,
+      "learning_rate": 4.96707074091454e-05,
+      "loss": 2.8789,
+      "step": 3560
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.133409023284912,
+      "learning_rate": 4.9668130461297194e-05,
+      "loss": 2.8137,
+      "step": 3570
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.30552339553833,
+      "learning_rate": 4.9665543536860255e-05,
+      "loss": 2.764,
+      "step": 3580
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.916990280151367,
+      "learning_rate": 4.966294663688083e-05,
+      "loss": 2.8359,
+      "step": 3590
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.573614597320557,
+      "learning_rate": 4.966033976240916e-05,
+      "loss": 2.9216,
+      "step": 3600
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.318258285522461,
+      "learning_rate": 4.965772291449958e-05,
+      "loss": 2.8818,
+      "step": 3610
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.087459564208984,
+      "learning_rate": 4.965509609421042e-05,
+      "loss": 2.6802,
+      "step": 3620
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.408172607421875,
+      "learning_rate": 4.9652459302604046e-05,
+      "loss": 2.7713,
+      "step": 3630
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.441077709197998,
+      "learning_rate": 4.964981254074686e-05,
+      "loss": 2.8727,
+      "step": 3640
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.799238681793213,
+      "learning_rate": 4.9647155809709306e-05,
+      "loss": 2.9911,
+      "step": 3650
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.362485885620117,
+      "learning_rate": 4.9644489110565845e-05,
+      "loss": 2.8088,
+      "step": 3660
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 6.419576168060303,
+      "learning_rate": 4.964181244439498e-05,
+      "loss": 2.8133,
+      "step": 3670
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 6.19252347946167,
+      "learning_rate": 4.963912581227924e-05,
+      "loss": 2.8429,
+      "step": 3680
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 6.425354480743408,
+      "learning_rate": 4.963642921530518e-05,
+      "loss": 2.8272,
+      "step": 3690
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.866875171661377,
+      "learning_rate": 4.9633722654563405e-05,
+      "loss": 2.816,
+      "step": 3700
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.425843238830566,
+      "learning_rate": 4.963100613114852e-05,
+      "loss": 2.8552,
+      "step": 3710
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 6.268444538116455,
+      "learning_rate": 4.962827964615919e-05,
+      "loss": 2.8642,
+      "step": 3720
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.511022090911865,
+      "learning_rate": 4.962554320069808e-05,
+      "loss": 2.8388,
+      "step": 3730
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 8.639039993286133,
+      "learning_rate": 4.9622796795871904e-05,
+      "loss": 2.809,
+      "step": 3740
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 6.79989767074585,
+      "learning_rate": 4.9620040432791395e-05,
+      "loss": 3.0316,
+      "step": 3750
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.568778991699219,
+      "learning_rate": 4.961727411257131e-05,
+      "loss": 2.7867,
+      "step": 3760
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.364297389984131,
+      "learning_rate": 4.961449783633045e-05,
+      "loss": 3.0213,
+      "step": 3770
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.523791790008545,
+      "learning_rate": 4.9611711605191625e-05,
+      "loss": 2.7875,
+      "step": 3780
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.39138650894165,
+      "learning_rate": 4.9608915420281675e-05,
+      "loss": 2.7144,
+      "step": 3790
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.800009250640869,
+      "learning_rate": 4.9606109282731463e-05,
+      "loss": 2.8091,
+      "step": 3800
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.470329284667969,
+      "learning_rate": 4.96032931936759e-05,
+      "loss": 2.8055,
+      "step": 3810
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.461717128753662,
+      "learning_rate": 4.9600467154253885e-05,
+      "loss": 2.8633,
+      "step": 3820
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.095701217651367,
+      "learning_rate": 4.959763116560836e-05,
+      "loss": 2.9004,
+      "step": 3830
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.429464340209961,
+      "learning_rate": 4.95947852288863e-05,
+      "loss": 2.6428,
+      "step": 3840
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.361551761627197,
+      "learning_rate": 4.95919293452387e-05,
+      "loss": 2.867,
+      "step": 3850
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.917137622833252,
+      "learning_rate": 4.958906351582054e-05,
+      "loss": 2.7266,
+      "step": 3860
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.074246406555176,
+      "learning_rate": 4.958618774179089e-05,
+      "loss": 2.8333,
+      "step": 3870
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 6.167778968811035,
+      "learning_rate": 4.9583302024312796e-05,
+      "loss": 2.7009,
+      "step": 3880
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.73693323135376,
+      "learning_rate": 4.9580406364553325e-05,
+      "loss": 2.5967,
+      "step": 3890
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.558803558349609,
+      "learning_rate": 4.957750076368357e-05,
+      "loss": 2.6763,
+      "step": 3900
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 7.1173481941223145,
+      "learning_rate": 4.957458522287868e-05,
+      "loss": 2.8593,
+      "step": 3910
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 7.378798484802246,
+      "learning_rate": 4.9571659743317754e-05,
+      "loss": 2.7876,
+      "step": 3920
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.375428199768066,
+      "learning_rate": 4.956872432618399e-05,
+      "loss": 2.7443,
+      "step": 3930
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.509357452392578,
+      "learning_rate": 4.9565778972664534e-05,
+      "loss": 2.8787,
+      "step": 3940
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.430800437927246,
+      "learning_rate": 4.95628236839506e-05,
+      "loss": 2.609,
+      "step": 3950
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.8563618659973145,
+      "learning_rate": 4.95598584612374e-05,
+      "loss": 2.7912,
+      "step": 3960
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.738502502441406,
+      "learning_rate": 4.9556883305724155e-05,
+      "loss": 2.8222,
+      "step": 3970
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.658450603485107,
+      "learning_rate": 4.955389821861412e-05,
+      "loss": 2.6731,
+      "step": 3980
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.250921726226807,
+      "learning_rate": 4.955090320111457e-05,
+      "loss": 2.6865,
+      "step": 3990
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.725744247436523,
+      "learning_rate": 4.954789825443678e-05,
+      "loss": 2.6249,
+      "step": 4000
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 2.342557668685913,
+      "eval_runtime": 423.1443,
+      "eval_samples_per_second": 50.151,
+      "eval_steps_per_second": 0.392,
+      "step": 4000
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.008103847503662,
+      "learning_rate": 4.954488337979604e-05,
+      "loss": 2.6645,
+      "step": 4010
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.027274131774902,
+      "learning_rate": 4.9541858578411674e-05,
+      "loss": 2.8588,
+      "step": 4020
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.368979454040527,
+      "learning_rate": 4.9538823851507006e-05,
+      "loss": 2.7103,
+      "step": 4030
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.687747001647949,
+      "learning_rate": 4.953577920030937e-05,
+      "loss": 2.5958,
+      "step": 4040
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.669676780700684,
+      "learning_rate": 4.953272462605013e-05,
+      "loss": 2.7199,
+      "step": 4050
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.0332112312316895,
+      "learning_rate": 4.952966012996466e-05,
+      "loss": 2.62,
+      "step": 4060
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.005265235900879,
+      "learning_rate": 4.952658571329233e-05,
+      "loss": 2.5974,
+      "step": 4070
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.806674003601074,
+      "learning_rate": 4.952350137727653e-05,
+      "loss": 2.9317,
+      "step": 4080
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.117702484130859,
+      "learning_rate": 4.9520407123164676e-05,
+      "loss": 2.6551,
+      "step": 4090
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.5031023025512695,
+      "learning_rate": 4.9517302952208174e-05,
+      "loss": 2.7362,
+      "step": 4100
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.63228702545166,
+      "learning_rate": 4.951418886566247e-05,
+      "loss": 2.6743,
+      "step": 4110
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.421300888061523,
+      "learning_rate": 4.951106486478698e-05,
+      "loss": 2.6827,
+      "step": 4120
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.456374168395996,
+      "learning_rate": 4.950793095084516e-05,
+      "loss": 2.7266,
+      "step": 4130
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.39730978012085,
+      "learning_rate": 4.950478712510446e-05,
+      "loss": 2.8165,
+      "step": 4140
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.817816734313965,
+      "learning_rate": 4.9501633388836354e-05,
+      "loss": 2.6275,
+      "step": 4150
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 7.556663990020752,
+      "learning_rate": 4.949846974331631e-05,
+      "loss": 2.6368,
+      "step": 4160
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.171590805053711,
+      "learning_rate": 4.949529618982381e-05,
+      "loss": 2.8336,
+      "step": 4170
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.013977527618408,
+      "learning_rate": 4.9492112729642346e-05,
+      "loss": 2.6988,
+      "step": 4180
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.861149311065674,
+      "learning_rate": 4.948891936405941e-05,
+      "loss": 2.8883,
+      "step": 4190
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.240298271179199,
+      "learning_rate": 4.948571609436649e-05,
+      "loss": 2.8353,
+      "step": 4200
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.673245429992676,
+      "learning_rate": 4.94825029218591e-05,
+      "loss": 2.6835,
+      "step": 4210
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.680504322052002,
+      "learning_rate": 4.947927984783674e-05,
+      "loss": 2.594,
+      "step": 4220
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.967173099517822,
+      "learning_rate": 4.947604687360295e-05,
+      "loss": 2.5797,
+      "step": 4230
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.29856538772583,
+      "learning_rate": 4.947280400046525e-05,
+      "loss": 2.6197,
+      "step": 4240
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.518858432769775,
+      "learning_rate": 4.9469551229735126e-05,
+      "loss": 2.6727,
+      "step": 4250
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.493609428405762,
+      "learning_rate": 4.946628856272814e-05,
+      "loss": 2.521,
+      "step": 4260
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.581683158874512,
+      "learning_rate": 4.9463016000763794e-05,
+      "loss": 2.4878,
+      "step": 4270
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.769138336181641,
+      "learning_rate": 4.945973354516564e-05,
+      "loss": 2.6396,
+      "step": 4280
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.217662811279297,
+      "learning_rate": 4.9456441197261195e-05,
+      "loss": 2.3943,
+      "step": 4290
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.986959457397461,
+      "learning_rate": 4.9453138958381994e-05,
+      "loss": 2.9015,
+      "step": 4300
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.872792720794678,
+      "learning_rate": 4.944982682986357e-05,
+      "loss": 2.7326,
+      "step": 4310
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.723078727722168,
+      "learning_rate": 4.944650481304545e-05,
+      "loss": 2.7212,
+      "step": 4320
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.189296722412109,
+      "learning_rate": 4.944317290927117e-05,
+      "loss": 2.8079,
+      "step": 4330
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.358461856842041,
+      "learning_rate": 4.943983111988827e-05,
+      "loss": 2.6931,
+      "step": 4340
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.47238826751709,
+      "learning_rate": 4.943647944624826e-05,
+      "loss": 2.5772,
+      "step": 4350
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.701056957244873,
+      "learning_rate": 4.943311788970667e-05,
+      "loss": 2.5407,
+      "step": 4360
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.032464504241943,
+      "learning_rate": 4.9429746451623024e-05,
+      "loss": 2.6685,
+      "step": 4370
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.141815185546875,
+      "learning_rate": 4.9426365133360834e-05,
+      "loss": 2.7409,
+      "step": 4380
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.483497619628906,
+      "learning_rate": 4.942297393628763e-05,
+      "loss": 2.5305,
+      "step": 4390
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.223423957824707,
+      "learning_rate": 4.941957286177491e-05,
+      "loss": 2.8228,
+      "step": 4400
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.8881402015686035,
+      "learning_rate": 4.9416161911198176e-05,
+      "loss": 2.5842,
+      "step": 4410
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.5558929443359375,
+      "learning_rate": 4.941274108593693e-05,
+      "loss": 2.5049,
+      "step": 4420
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 7.3317694664001465,
+      "learning_rate": 4.940931038737468e-05,
+      "loss": 2.5361,
+      "step": 4430
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.913518905639648,
+      "learning_rate": 4.940586981689888e-05,
+      "loss": 2.7733,
+      "step": 4440
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 8.628571510314941,
+      "learning_rate": 4.940241937590102e-05,
+      "loss": 2.642,
+      "step": 4450
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.870291233062744,
+      "learning_rate": 4.939895906577658e-05,
+      "loss": 2.9244,
+      "step": 4460
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.8803253173828125,
+      "learning_rate": 4.939548888792502e-05,
+      "loss": 2.6324,
+      "step": 4470
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.114381790161133,
+      "learning_rate": 4.939200884374979e-05,
+      "loss": 2.674,
+      "step": 4480
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 7.145431041717529,
+      "learning_rate": 4.938851893465832e-05,
+      "loss": 2.6255,
+      "step": 4490
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.35526704788208,
+      "learning_rate": 4.938501916206206e-05,
+      "loss": 2.8467,
+      "step": 4500
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 2.259162664413452,
+      "eval_runtime": 421.9557,
+      "eval_samples_per_second": 50.292,
+      "eval_steps_per_second": 0.393,
+      "step": 4500
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.096010208129883,
+      "learning_rate": 4.938150952737642e-05,
+      "loss": 2.6732,
+      "step": 4510
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.902897357940674,
+      "learning_rate": 4.937799003202081e-05,
+      "loss": 2.5127,
+      "step": 4520
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.106553077697754,
+      "learning_rate": 4.9374460677418635e-05,
+      "loss": 2.5394,
+      "step": 4530
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.985318183898926,
+      "learning_rate": 4.937092146499727e-05,
+      "loss": 2.6342,
+      "step": 4540
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.647487163543701,
+      "learning_rate": 4.9367372396188095e-05,
+      "loss": 2.7253,
+      "step": 4550
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.041297435760498,
+      "learning_rate": 4.9363813472426465e-05,
+      "loss": 2.7264,
+      "step": 4560
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.295587062835693,
+      "learning_rate": 4.9360244695151716e-05,
+      "loss": 2.6379,
+      "step": 4570
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.751054286956787,
+      "learning_rate": 4.935666606580719e-05,
+      "loss": 2.6497,
+      "step": 4580
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.367273330688477,
+      "learning_rate": 4.935307758584019e-05,
+      "loss": 2.3892,
+      "step": 4590
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.491433143615723,
+      "learning_rate": 4.9349479256702023e-05,
+      "loss": 2.4803,
+      "step": 4600
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.676270484924316,
+      "learning_rate": 4.934587107984796e-05,
+      "loss": 2.4972,
+      "step": 4610
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.060196876525879,
+      "learning_rate": 4.934225305673728e-05,
+      "loss": 2.4878,
+      "step": 4620
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.352468967437744,
+      "learning_rate": 4.93386251888332e-05,
+      "loss": 2.4713,
+      "step": 4630
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 8.105636596679688,
+      "learning_rate": 4.9334987477602975e-05,
+      "loss": 2.7637,
+      "step": 4640
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.352108478546143,
+      "learning_rate": 4.933133992451779e-05,
+      "loss": 2.5642,
+      "step": 4650
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.43104362487793,
+      "learning_rate": 4.932768253105286e-05,
+      "loss": 2.6106,
+      "step": 4660
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.641717433929443,
+      "learning_rate": 4.932401529868733e-05,
+      "loss": 2.6049,
+      "step": 4670
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.041056156158447,
+      "learning_rate": 4.932033822890436e-05,
+      "loss": 2.512,
+      "step": 4680
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.870728015899658,
+      "learning_rate": 4.9316651323191067e-05,
+      "loss": 2.4649,
+      "step": 4690
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.347024917602539,
+      "learning_rate": 4.931295458303856e-05,
+      "loss": 2.5996,
+      "step": 4700
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.286810398101807,
+      "learning_rate": 4.9309248009941914e-05,
+      "loss": 2.5899,
+      "step": 4710
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.839949131011963,
+      "learning_rate": 4.93055316054002e-05,
+      "loss": 2.5646,
+      "step": 4720
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.071991920471191,
+      "learning_rate": 4.930180537091644e-05,
+      "loss": 2.7168,
+      "step": 4730
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.042398452758789,
+      "learning_rate": 4.929806930799765e-05,
+      "loss": 2.7169,
+      "step": 4740
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.924988269805908,
+      "learning_rate": 4.9294323418154805e-05,
+      "loss": 2.4587,
+      "step": 4750
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.021846294403076,
+      "learning_rate": 4.929056770290288e-05,
+      "loss": 2.5327,
+      "step": 4760
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.043169975280762,
+      "learning_rate": 4.928680216376079e-05,
+      "loss": 2.4268,
+      "step": 4770
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 7.106345176696777,
+      "learning_rate": 4.9283026802251454e-05,
+      "loss": 2.5116,
+      "step": 4780
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.8474650382995605,
+      "learning_rate": 4.927924161990175e-05,
+      "loss": 2.5163,
+      "step": 4790
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.736364841461182,
+      "learning_rate": 4.9275446618242516e-05,
+      "loss": 2.4918,
+      "step": 4800
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.512024879455566,
+      "learning_rate": 4.927164179880858e-05,
+      "loss": 2.5756,
+      "step": 4810
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.911081790924072,
+      "learning_rate": 4.926782716313874e-05,
+      "loss": 2.4661,
+      "step": 4820
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.932026386260986,
+      "learning_rate": 4.9264002712775745e-05,
+      "loss": 2.5059,
+      "step": 4830
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.116228103637695,
+      "learning_rate": 4.9260168449266335e-05,
+      "loss": 2.4183,
+      "step": 4840
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.023027420043945,
+      "learning_rate": 4.925632437416121e-05,
+      "loss": 2.4144,
+      "step": 4850
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.4858832359313965,
+      "learning_rate": 4.925247048901504e-05,
+      "loss": 2.4708,
+      "step": 4860
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.690423965454102,
+      "learning_rate": 4.924860679538645e-05,
+      "loss": 2.7061,
+      "step": 4870
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.818970203399658,
+      "learning_rate": 4.924473329483806e-05,
+      "loss": 2.6152,
+      "step": 4880
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.238198280334473,
+      "learning_rate": 4.924084998893642e-05,
+      "loss": 2.3184,
+      "step": 4890
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.587952613830566,
+      "learning_rate": 4.923695687925207e-05,
+      "loss": 2.6381,
+      "step": 4900
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.2837066650390625,
+      "learning_rate": 4.923305396735952e-05,
+      "loss": 2.52,
+      "step": 4910
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.908483028411865,
+      "learning_rate": 4.9229141254837226e-05,
+      "loss": 2.675,
+      "step": 4920
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.803688049316406,
+      "learning_rate": 4.922521874326761e-05,
+      "loss": 2.6054,
+      "step": 4930
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.124865531921387,
+      "learning_rate": 4.922128643423707e-05,
+      "loss": 2.6076,
+      "step": 4940
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.534984111785889,
+      "learning_rate": 4.921734432933596e-05,
+      "loss": 2.4958,
+      "step": 4950
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 7.013035297393799,
+      "learning_rate": 4.921339243015858e-05,
+      "loss": 2.6099,
+      "step": 4960
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.598790645599365,
+      "learning_rate": 4.920943073830322e-05,
+      "loss": 2.493,
+      "step": 4970
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.958088397979736,
+      "learning_rate": 4.920545925537212e-05,
+      "loss": 2.5689,
+      "step": 4980
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.180896282196045,
+      "learning_rate": 4.9201477982971466e-05,
+      "loss": 2.6138,
+      "step": 4990
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.9859209060668945,
+      "learning_rate": 4.919748692271141e-05,
+      "loss": 2.5003,
+      "step": 5000
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 2.1625306606292725,
+      "eval_runtime": 422.5397,
+      "eval_samples_per_second": 50.223,
+      "eval_steps_per_second": 0.393,
+      "step": 5000
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.54530143737793,
+      "learning_rate": 4.9193486076206075e-05,
+      "loss": 2.446,
+      "step": 5010
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.814847469329834,
+      "learning_rate": 4.918947544507353e-05,
+      "loss": 2.3595,
+      "step": 5020
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.315622806549072,
+      "learning_rate": 4.91854550309358e-05,
+      "loss": 2.6138,
+      "step": 5030
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.739955425262451,
+      "learning_rate": 4.9181424835418874e-05,
+      "loss": 2.5347,
+      "step": 5040
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.838977336883545,
+      "learning_rate": 4.91773848601527e-05,
+      "loss": 2.3811,
+      "step": 5050
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.2752766609191895,
+      "learning_rate": 4.9173335106771154e-05,
+      "loss": 2.3222,
+      "step": 5060
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 7.31117057800293,
+      "learning_rate": 4.916927557691211e-05,
+      "loss": 2.3912,
+      "step": 5070
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.800614833831787,
+      "learning_rate": 4.916520627221736e-05,
+      "loss": 2.5183,
+      "step": 5080
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.574156284332275,
+      "learning_rate": 4.9161127194332655e-05,
+      "loss": 2.4722,
+      "step": 5090
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.964874744415283,
+      "learning_rate": 4.915703834490773e-05,
+      "loss": 2.6562,
+      "step": 5100
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.67961311340332,
+      "learning_rate": 4.915293972559623e-05,
+      "loss": 2.6234,
+      "step": 5110
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.129438877105713,
+      "learning_rate": 4.9148831338055766e-05,
+      "loss": 2.3889,
+      "step": 5120
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.661647796630859,
+      "learning_rate": 4.9144713183947915e-05,
+      "loss": 2.337,
+      "step": 5130
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.4586567878723145,
+      "learning_rate": 4.914058526493819e-05,
+      "loss": 2.525,
+      "step": 5140
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.5740532875061035,
+      "learning_rate": 4.913644758269604e-05,
+      "loss": 2.4817,
+      "step": 5150
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.7009124755859375,
+      "learning_rate": 4.9132300138894895e-05,
+      "loss": 2.3586,
+      "step": 5160
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.846322536468506,
+      "learning_rate": 4.912814293521211e-05,
+      "loss": 2.4756,
+      "step": 5170
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.497846603393555,
+      "learning_rate": 4.9123975973328986e-05,
+      "loss": 2.3421,
+      "step": 5180
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.7938971519470215,
+      "learning_rate": 4.911979925493079e-05,
+      "loss": 2.609,
+      "step": 5190
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.914577960968018,
+      "learning_rate": 4.91156127817067e-05,
+      "loss": 2.5344,
+      "step": 5200
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.262577533721924,
+      "learning_rate": 4.911141655534989e-05,
+      "loss": 2.4138,
+      "step": 5210
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.629760265350342,
+      "learning_rate": 4.9107210577557415e-05,
+      "loss": 2.4292,
+      "step": 5220
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 18.421449661254883,
+      "learning_rate": 4.9102994850030344e-05,
+      "loss": 2.594,
+      "step": 5230
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.440621376037598,
+      "learning_rate": 4.909876937447363e-05,
+      "loss": 2.338,
+      "step": 5240
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.139665603637695,
+      "learning_rate": 4.90945341525962e-05,
+      "loss": 2.4571,
+      "step": 5250
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.897266864776611,
+      "learning_rate": 4.909028918611091e-05,
+      "loss": 2.4774,
+      "step": 5260
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.971827507019043,
+      "learning_rate": 4.908603447673457e-05,
+      "loss": 2.5588,
+      "step": 5270
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.420186996459961,
+      "learning_rate": 4.9081770026187914e-05,
+      "loss": 2.4299,
+      "step": 5280
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.321455478668213,
+      "learning_rate": 4.907749583619562e-05,
+      "loss": 2.4522,
+      "step": 5290
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.8206400871276855,
+      "learning_rate": 4.907321190848632e-05,
+      "loss": 2.5081,
+      "step": 5300
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.017814636230469,
+      "learning_rate": 4.9068918244792565e-05,
+      "loss": 2.2797,
+      "step": 5310
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.493027210235596,
+      "learning_rate": 4.906461484685085e-05,
+      "loss": 2.4124,
+      "step": 5320
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.415233612060547,
+      "learning_rate": 4.906030171640163e-05,
+      "loss": 2.4818,
+      "step": 5330
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.612936496734619,
+      "learning_rate": 4.905597885518923e-05,
+      "loss": 2.5769,
+      "step": 5340
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.570766448974609,
+      "learning_rate": 4.9051646264962005e-05,
+      "loss": 2.3302,
+      "step": 5350
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.15294885635376,
+      "learning_rate": 4.904730394747216e-05,
+      "loss": 2.3181,
+      "step": 5360
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.568897247314453,
+      "learning_rate": 4.904295190447587e-05,
+      "loss": 2.4157,
+      "step": 5370
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.8973236083984375,
+      "learning_rate": 4.903859013773325e-05,
+      "loss": 2.5564,
+      "step": 5380
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.484323024749756,
+      "learning_rate": 4.903421864900835e-05,
+      "loss": 2.3377,
+      "step": 5390
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.30608606338501,
+      "learning_rate": 4.902983744006913e-05,
+      "loss": 2.3119,
+      "step": 5400
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.372623920440674,
+      "learning_rate": 4.902544651268749e-05,
+      "loss": 2.3764,
+      "step": 5410
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.923567771911621,
+      "learning_rate": 4.902104586863927e-05,
+      "loss": 2.452,
+      "step": 5420
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.1613664627075195,
+      "learning_rate": 4.901663550970422e-05,
+      "loss": 2.5677,
+      "step": 5430
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.322874069213867,
+      "learning_rate": 4.901221543766604e-05,
+      "loss": 2.3442,
+      "step": 5440
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.805524826049805,
+      "learning_rate": 4.900778565431236e-05,
+      "loss": 2.377,
+      "step": 5450
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.61544942855835,
+      "learning_rate": 4.900334616143471e-05,
+      "loss": 2.2881,
+      "step": 5460
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 9.518655776977539,
+      "learning_rate": 4.8998896960828576e-05,
+      "loss": 2.5395,
+      "step": 5470
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.4181108474731445,
+      "learning_rate": 4.899443805429336e-05,
+      "loss": 2.5815,
+      "step": 5480
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 7.65767765045166,
+      "learning_rate": 4.8989969443632366e-05,
+      "loss": 2.383,
+      "step": 5490
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.252597332000732,
+      "learning_rate": 4.898549113065287e-05,
+      "loss": 2.3579,
+      "step": 5500
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 2.0875422954559326,
+      "eval_runtime": 423.4806,
+      "eval_samples_per_second": 50.111,
+      "eval_steps_per_second": 0.392,
+      "step": 5500
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.8088860511779785,
+      "learning_rate": 4.898100311716604e-05,
+      "loss": 2.4613,
+      "step": 5510
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.02274227142334,
+      "learning_rate": 4.897650540498697e-05,
+      "loss": 2.2701,
+      "step": 5520
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.915895938873291,
+      "learning_rate": 4.8971997995934675e-05,
+      "loss": 2.3393,
+      "step": 5530
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.784215450286865,
+      "learning_rate": 4.896748089183211e-05,
+      "loss": 2.5954,
+      "step": 5540
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.802945613861084,
+      "learning_rate": 4.896295409450612e-05,
+      "loss": 2.2532,
+      "step": 5550
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.0984649658203125,
+      "learning_rate": 4.895841760578751e-05,
+      "loss": 2.3553,
+      "step": 5560
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.783207893371582,
+      "learning_rate": 4.8953871427510965e-05,
+      "loss": 2.445,
+      "step": 5570
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.820375919342041,
+      "learning_rate": 4.894931556151512e-05,
+      "loss": 2.4465,
+      "step": 5580
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 54.05234146118164,
+      "learning_rate": 4.894475000964249e-05,
+      "loss": 2.4557,
+      "step": 5590
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.997950077056885,
+      "learning_rate": 4.8940174773739564e-05,
+      "loss": 2.2733,
+      "step": 5600
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.4290642738342285,
+      "learning_rate": 4.8935589855656694e-05,
+      "loss": 2.3262,
+      "step": 5610
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.6273064613342285,
+      "learning_rate": 4.893099525724818e-05,
+      "loss": 2.4704,
+      "step": 5620
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.293818473815918,
+      "learning_rate": 4.892639098037221e-05,
+      "loss": 2.525,
+      "step": 5630
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.767632484436035,
+      "learning_rate": 4.892177702689091e-05,
+      "loss": 2.2745,
+      "step": 5640
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.506222248077393,
+      "learning_rate": 4.891715339867033e-05,
+      "loss": 2.4635,
+      "step": 5650
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.971251010894775,
+      "learning_rate": 4.8912520097580395e-05,
+      "loss": 2.5934,
+      "step": 5660
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.757556438446045,
+      "learning_rate": 4.890787712549496e-05,
+      "loss": 2.4497,
+      "step": 5670
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 8.545730590820312,
+      "learning_rate": 4.890369018346882e-05,
+      "loss": 2.3384,
+      "step": 5680
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.717365264892578,
+      "learning_rate": 4.889902884166843e-05,
+      "loss": 2.3127,
+      "step": 5690
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.706165790557861,
+      "learning_rate": 4.889435783432884e-05,
+      "loss": 2.1936,
+      "step": 5700
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.551875591278076,
+      "learning_rate": 4.888967716333917e-05,
+      "loss": 2.5815,
+      "step": 5710
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.260705471038818,
+      "learning_rate": 4.888498683059243e-05,
+      "loss": 2.3263,
+      "step": 5720
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.164261341094971,
+      "learning_rate": 4.8880286837985526e-05,
+      "loss": 2.276,
+      "step": 5730
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.247165679931641,
+      "learning_rate": 4.887557718741931e-05,
+      "loss": 2.4281,
+      "step": 5740
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.835640907287598,
+      "learning_rate": 4.88708578807985e-05,
+      "loss": 2.2763,
+      "step": 5750
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.9643354415893555,
+      "learning_rate": 4.886612892003174e-05,
+      "loss": 2.4745,
+      "step": 5760
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 7.18971061706543,
+      "learning_rate": 4.886139030703159e-05,
+      "loss": 2.4685,
+      "step": 5770
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.121529579162598,
+      "learning_rate": 4.885664204371446e-05,
+      "loss": 2.5183,
+      "step": 5780
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.581070423126221,
+      "learning_rate": 4.885188413200075e-05,
+      "loss": 2.5167,
+      "step": 5790
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.491427898406982,
+      "learning_rate": 4.8847116573814676e-05,
+      "loss": 2.383,
+      "step": 5800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.595372676849365,
+      "learning_rate": 4.8842339371084414e-05,
+      "loss": 2.0989,
+      "step": 5810
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.8517165184021,
+      "learning_rate": 4.8837552525742004e-05,
+      "loss": 2.4912,
+      "step": 5820
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.275095462799072,
+      "learning_rate": 4.88327560397234e-05,
+      "loss": 2.3524,
+      "step": 5830
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.616918563842773,
+      "learning_rate": 4.8827949914968474e-05,
+      "loss": 1.9738,
+      "step": 5840
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 7.1793999671936035,
+      "learning_rate": 4.882313415342097e-05,
+      "loss": 2.2815,
+      "step": 5850
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.783013343811035,
+      "learning_rate": 4.881830875702852e-05,
+      "loss": 2.4371,
+      "step": 5860
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 18.31147003173828,
+      "learning_rate": 4.88134737277427e-05,
+      "loss": 2.5314,
+      "step": 5870
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.92976713180542,
+      "learning_rate": 4.880862906751893e-05,
+      "loss": 2.2746,
+      "step": 5880
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.150814056396484,
+      "learning_rate": 4.8803774778316557e-05,
+      "loss": 2.277,
+      "step": 5890
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.579446792602539,
+      "learning_rate": 4.87989108620988e-05,
+      "loss": 2.2136,
+      "step": 5900
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.303000450134277,
+      "learning_rate": 4.8794037320832806e-05,
+      "loss": 2.258,
+      "step": 5910
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.469798564910889,
+      "learning_rate": 4.878915415648957e-05,
+      "loss": 2.4185,
+      "step": 5920
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.8365960121154785,
+      "learning_rate": 4.8784261371044016e-05,
+      "loss": 2.3902,
+      "step": 5930
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.7992987632751465,
+      "learning_rate": 4.877984963973593e-05,
+      "loss": 2.3468,
+      "step": 5940
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.936422348022461,
+      "learning_rate": 4.877493857965077e-05,
+      "loss": 2.3446,
+      "step": 5950
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.544583797454834,
+      "learning_rate": 4.8770017904212525e-05,
+      "loss": 2.2751,
+      "step": 5960
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.835214138031006,
+      "learning_rate": 4.876508761541129e-05,
+      "loss": 2.3919,
+      "step": 5970
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.438685417175293,
+      "learning_rate": 4.8760147715241e-05,
+      "loss": 2.3636,
+      "step": 5980
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.63602876663208,
+      "learning_rate": 4.875519820569954e-05,
+      "loss": 2.4277,
+      "step": 5990
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.563483238220215,
+      "learning_rate": 4.8750239088788636e-05,
+      "loss": 2.2586,
+      "step": 6000
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 2.014089584350586,
+      "eval_runtime": 422.9188,
+      "eval_samples_per_second": 50.177,
+      "eval_steps_per_second": 0.393,
+      "step": 6000
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.133357524871826,
+      "learning_rate": 4.8745270366513914e-05,
+      "loss": 2.2622,
+      "step": 6010
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 7.025454521179199,
+      "learning_rate": 4.87402920408849e-05,
+      "loss": 2.2786,
+      "step": 6020
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.6243181228637695,
+      "learning_rate": 4.873530411391498e-05,
+      "loss": 2.3901,
+      "step": 6030
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.4480109214782715,
+      "learning_rate": 4.873030658762143e-05,
+      "loss": 2.1692,
+      "step": 6040
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.633693695068359,
+      "learning_rate": 4.8725299464025414e-05,
+      "loss": 2.198,
+      "step": 6050
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.928077220916748,
+      "learning_rate": 4.872028274515198e-05,
+      "loss": 2.0226,
+      "step": 6060
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.940571308135986,
+      "learning_rate": 4.871525643303004e-05,
+      "loss": 2.5249,
+      "step": 6070
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.45697021484375,
+      "learning_rate": 4.871022052969241e-05,
+      "loss": 2.184,
+      "step": 6080
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.87355899810791,
+      "learning_rate": 4.870517503717576e-05,
+      "loss": 2.4252,
+      "step": 6090
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.142642021179199,
+      "learning_rate": 4.8700119957520653e-05,
+      "loss": 2.2134,
+      "step": 6100
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.03863000869751,
+      "learning_rate": 4.869505529277154e-05,
+      "loss": 2.4,
+      "step": 6110
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.4546003341674805,
+      "learning_rate": 4.8689981044976715e-05,
+      "loss": 2.2927,
+      "step": 6120
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 10.309746742248535,
+      "learning_rate": 4.868489721618838e-05,
+      "loss": 2.3127,
+      "step": 6130
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.640440464019775,
+      "learning_rate": 4.867980380846259e-05,
+      "loss": 2.1445,
+      "step": 6140
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 7.428249835968018,
+      "learning_rate": 4.86747008238593e-05,
+      "loss": 2.4585,
+      "step": 6150
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.020784854888916,
+      "learning_rate": 4.86695882644423e-05,
+      "loss": 2.2498,
+      "step": 6160
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.098503112792969,
+      "learning_rate": 4.8664466132279294e-05,
+      "loss": 2.1231,
+      "step": 6170
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.907447814941406,
+      "learning_rate": 4.8659334429441825e-05,
+      "loss": 2.2518,
+      "step": 6180
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 38.992706298828125,
+      "learning_rate": 4.8654193158005336e-05,
+      "loss": 2.4063,
+      "step": 6190
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.094198226928711,
+      "learning_rate": 4.86490423200491e-05,
+      "loss": 2.1258,
+      "step": 6200
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.0221686363220215,
+      "learning_rate": 4.864388191765629e-05,
+      "loss": 2.5631,
+      "step": 6210
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.385717391967773,
+      "learning_rate": 4.863871195291395e-05,
+      "loss": 2.4827,
+      "step": 6220
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.672618389129639,
+      "learning_rate": 4.863353242791297e-05,
+      "loss": 2.2295,
+      "step": 6230
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.568421840667725,
+      "learning_rate": 4.8628343344748116e-05,
+      "loss": 2.2235,
+      "step": 6240
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.92976713180542,
+      "learning_rate": 4.8623144705518034e-05,
+      "loss": 2.2561,
+      "step": 6250
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 7.64150857925415,
+      "learning_rate": 4.8617936512325214e-05,
+      "loss": 2.5747,
+      "step": 6260
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.804096221923828,
+      "learning_rate": 4.861271876727601e-05,
+      "loss": 2.1542,
+      "step": 6270
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.928376197814941,
+      "learning_rate": 4.860749147248066e-05,
+      "loss": 2.2101,
+      "step": 6280
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.408993244171143,
+      "learning_rate": 4.8602254630053235e-05,
+      "loss": 2.3248,
+      "step": 6290
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.5954813957214355,
+      "learning_rate": 4.8597008242111694e-05,
+      "loss": 2.3062,
+      "step": 6300
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.591384410858154,
+      "learning_rate": 4.859175231077785e-05,
+      "loss": 2.2668,
+      "step": 6310
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.448415279388428,
+      "learning_rate": 4.858648683817736e-05,
+      "loss": 2.2404,
+      "step": 6320
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.797817707061768,
+      "learning_rate": 4.8581211826439765e-05,
+      "loss": 2.3924,
+      "step": 6330
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.555346965789795,
+      "learning_rate": 4.8575927277698435e-05,
+      "loss": 2.2588,
+      "step": 6340
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.0208821296691895,
+      "learning_rate": 4.857063319409062e-05,
+      "loss": 2.2423,
+      "step": 6350
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.1246466636657715,
+      "learning_rate": 4.8565329577757426e-05,
+      "loss": 2.1666,
+      "step": 6360
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 12.975321769714355,
+      "learning_rate": 4.856001643084378e-05,
+      "loss": 2.253,
+      "step": 6370
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.758956432342529,
+      "learning_rate": 4.855469375549853e-05,
+      "loss": 2.061,
+      "step": 6380
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.913463115692139,
+      "learning_rate": 4.8549361553874295e-05,
+      "loss": 2.2273,
+      "step": 6390
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 8.23082160949707,
+      "learning_rate": 4.854401982812762e-05,
+      "loss": 2.599,
+      "step": 6400
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 8.17214298248291,
+      "learning_rate": 4.853866858041887e-05,
+      "loss": 2.5361,
+      "step": 6410
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 9.37897777557373,
+      "learning_rate": 4.853330781291224e-05,
+      "loss": 2.3414,
+      "step": 6420
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.9388427734375,
+      "learning_rate": 4.852793752777582e-05,
+      "loss": 2.3078,
+      "step": 6430
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.1688055992126465,
+      "learning_rate": 4.8522557727181517e-05,
+      "loss": 2.2146,
+      "step": 6440
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 13.751556396484375,
+      "learning_rate": 4.85171684133051e-05,
+      "loss": 2.3457,
+      "step": 6450
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.383643627166748,
+      "learning_rate": 4.851176958832618e-05,
+      "loss": 2.4478,
+      "step": 6460
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.072232723236084,
+      "learning_rate": 4.850636125442821e-05,
+      "loss": 2.0503,
+      "step": 6470
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.105084419250488,
+      "learning_rate": 4.850094341379851e-05,
+      "loss": 2.4092,
+      "step": 6480
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.889256477355957,
+      "learning_rate": 4.8495516068628216e-05,
+      "loss": 2.1762,
+      "step": 6490
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 10.305624008178711,
+      "learning_rate": 4.849007922111233e-05,
+      "loss": 2.2757,
+      "step": 6500
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 1.962647795677185,
+      "eval_runtime": 422.8287,
+      "eval_samples_per_second": 50.188,
+      "eval_steps_per_second": 0.393,
+      "step": 6500
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 12.288636207580566,
+      "learning_rate": 4.8484632873449676e-05,
+      "loss": 2.3152,
+      "step": 6510
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 18.38760757446289,
+      "learning_rate": 4.847917702784295e-05,
+      "loss": 2.3602,
+      "step": 6520
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 7.56703519821167,
+      "learning_rate": 4.8473711686498655e-05,
+      "loss": 2.3064,
+      "step": 6530
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.525967121124268,
+      "learning_rate": 4.8468236851627176e-05,
+      "loss": 2.2774,
+      "step": 6540
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.542328357696533,
+      "learning_rate": 4.846275252544269e-05,
+      "loss": 2.2453,
+      "step": 6550
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 9.50683879852295,
+      "learning_rate": 4.845725871016324e-05,
+      "loss": 2.43,
+      "step": 6560
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 7.898688793182373,
+      "learning_rate": 4.84517554080107e-05,
+      "loss": 2.1673,
+      "step": 6570
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.77755069732666,
+      "learning_rate": 4.8446242621210804e-05,
+      "loss": 2.2338,
+      "step": 6580
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.6925225257873535,
+      "learning_rate": 4.844072035199307e-05,
+      "loss": 2.2819,
+      "step": 6590
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.21223258972168,
+      "learning_rate": 4.843518860259091e-05,
+      "loss": 2.3132,
+      "step": 6600
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.520556926727295,
+      "learning_rate": 4.842964737524153e-05,
+      "loss": 2.3449,
+      "step": 6610
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 21.28665542602539,
+      "learning_rate": 4.842409667218598e-05,
+      "loss": 2.315,
+      "step": 6620
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.1759033203125,
+      "learning_rate": 4.841853649566914e-05,
+      "loss": 2.3547,
+      "step": 6630
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 19.148984909057617,
+      "learning_rate": 4.841296684793973e-05,
+      "loss": 2.5135,
+      "step": 6640
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 9.910473823547363,
+      "learning_rate": 4.84073877312503e-05,
+      "loss": 2.204,
+      "step": 6650
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 27.07410430908203,
+      "learning_rate": 4.8401799147857216e-05,
+      "loss": 2.4857,
+      "step": 6660
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 62.342281341552734,
+      "learning_rate": 4.839620110002069e-05,
+      "loss": 2.3909,
+      "step": 6670
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 103.58372497558594,
+      "learning_rate": 4.8391154766739814e-05,
+      "loss": 2.5908,
+      "step": 6680
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 22.150035858154297,
+      "learning_rate": 4.83855387427013e-05,
+      "loss": 2.3127,
+      "step": 6690
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 14.04951000213623,
+      "learning_rate": 4.8379913260795586e-05,
+      "loss": 2.4046,
+      "step": 6700
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 22.72818946838379,
+      "learning_rate": 4.837427832329779e-05,
+      "loss": 2.8313,
+      "step": 6710
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 13.296052932739258,
+      "learning_rate": 4.836863393248684e-05,
+      "loss": 2.4321,
+      "step": 6720
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.5498833656311035,
+      "learning_rate": 4.836298009064554e-05,
+      "loss": 2.3137,
+      "step": 6730
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 13.326327323913574,
+      "learning_rate": 4.835731680006047e-05,
+      "loss": 2.291,
+      "step": 6740
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 32.48208999633789,
+      "learning_rate": 4.835164406302205e-05,
+      "loss": 2.6136,
+      "step": 6750
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 11.908808708190918,
+      "learning_rate": 4.834596188182453e-05,
+      "loss": 2.1887,
+      "step": 6760
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 10.075331687927246,
+      "learning_rate": 4.834027025876595e-05,
+      "loss": 2.3643,
+      "step": 6770
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.72551155090332,
+      "learning_rate": 4.83345691961482e-05,
+      "loss": 2.2839,
+      "step": 6780
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.926360130310059,
+      "learning_rate": 4.8328858696276955e-05,
+      "loss": 2.3022,
+      "step": 6790
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 9.422292709350586,
+      "learning_rate": 4.832313876146175e-05,
+      "loss": 2.4454,
+      "step": 6800
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 13.58336353302002,
+      "learning_rate": 4.83174093940159e-05,
+      "loss": 2.176,
+      "step": 6810
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 8.902631759643555,
+      "learning_rate": 4.831167059625654e-05,
+      "loss": 2.3952,
+      "step": 6820
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 16.641265869140625,
+      "learning_rate": 4.830592237050464e-05,
+      "loss": 2.2495,
+      "step": 6830
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 11.762859344482422,
+      "learning_rate": 4.830016471908496e-05,
+      "loss": 2.4345,
+      "step": 6840
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.909867763519287,
+      "learning_rate": 4.829439764432607e-05,
+      "loss": 2.207,
+      "step": 6850
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 13.684906959533691,
+      "learning_rate": 4.828862114856038e-05,
+      "loss": 2.3951,
+      "step": 6860
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.779490947723389,
+      "learning_rate": 4.8282835234124074e-05,
+      "loss": 2.3883,
+      "step": 6870
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 7.6955366134643555,
+      "learning_rate": 4.827703990335718e-05,
+      "loss": 2.3341,
+      "step": 6880
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 7.92070198059082,
+      "learning_rate": 4.82712351586035e-05,
+      "loss": 2.4001,
+      "step": 6890
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 12.462532997131348,
+      "learning_rate": 4.826542100221067e-05,
+      "loss": 2.2261,
+      "step": 6900
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 18.708791732788086,
+      "learning_rate": 4.8259597436530125e-05,
+      "loss": 2.1191,
+      "step": 6910
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.56512975692749,
+      "learning_rate": 4.8253764463917096e-05,
+      "loss": 2.2219,
+      "step": 6920
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.9286699295043945,
+      "learning_rate": 4.8247922086730634e-05,
+      "loss": 2.3153,
+      "step": 6930
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 7.228487491607666,
+      "learning_rate": 4.8242070307333574e-05,
+      "loss": 2.4929,
+      "step": 6940
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 11.886446952819824,
+      "learning_rate": 4.8236209128092566e-05,
+      "loss": 2.3487,
+      "step": 6950
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.152067184448242,
+      "learning_rate": 4.823033855137807e-05,
+      "loss": 2.2251,
+      "step": 6960
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 14.04540729522705,
+      "learning_rate": 4.8224458579564334e-05,
+      "loss": 2.3503,
+      "step": 6970
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 9.333481788635254,
+      "learning_rate": 4.8218569215029405e-05,
+      "loss": 2.3221,
+      "step": 6980
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 35.72920227050781,
+      "learning_rate": 4.821267046015513e-05,
+      "loss": 2.3805,
+      "step": 6990
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 18.05036163330078,
+      "learning_rate": 4.820676231732716e-05,
+      "loss": 2.2986,
+      "step": 7000
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 1.956457495689392,
+      "eval_runtime": 423.6428,
+      "eval_samples_per_second": 50.092,
+      "eval_steps_per_second": 0.392,
+      "step": 7000
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 18.98715591430664,
+      "learning_rate": 4.820084478893494e-05,
+      "loss": 2.3742,
+      "step": 7010
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 7.488858699798584,
+      "learning_rate": 4.819491787737171e-05,
+      "loss": 2.3052,
+      "step": 7020
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 7.39491605758667,
+      "learning_rate": 4.81889815850345e-05,
+      "loss": 2.54,
+      "step": 7030
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 7.2730865478515625,
+      "learning_rate": 4.8183035914324136e-05,
+      "loss": 2.2487,
+      "step": 7040
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 7.395474433898926,
+      "learning_rate": 4.8177080867645245e-05,
+      "loss": 2.3872,
+      "step": 7050
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 13.659703254699707,
+      "learning_rate": 4.817111644740624e-05,
+      "loss": 2.3713,
+      "step": 7060
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 21.210573196411133,
+      "learning_rate": 4.8165142656019325e-05,
+      "loss": 2.2776,
+      "step": 7070
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 35.836185455322266,
+      "learning_rate": 4.81591594959005e-05,
+      "loss": 2.4872,
+      "step": 7080
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 13.00560188293457,
+      "learning_rate": 4.815316696946953e-05,
+      "loss": 2.2952,
+      "step": 7090
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 7.423733234405518,
+      "learning_rate": 4.814716507915e-05,
+      "loss": 2.3123,
+      "step": 7100
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 29.948333740234375,
+      "learning_rate": 4.8141153827369264e-05,
+      "loss": 2.0931,
+      "step": 7110
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 7.296632289886475,
+      "learning_rate": 4.813513321655848e-05,
+      "loss": 2.2626,
+      "step": 7120
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.078789710998535,
+      "learning_rate": 4.8129103249152554e-05,
+      "loss": 2.193,
+      "step": 7130
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.776204586029053,
+      "learning_rate": 4.812306392759022e-05,
+      "loss": 2.2646,
+      "step": 7140
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.253006458282471,
+      "learning_rate": 4.811701525431396e-05,
+      "loss": 2.2343,
+      "step": 7150
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 41.79158401489258,
+      "learning_rate": 4.811095723177006e-05,
+      "loss": 2.232,
+      "step": 7160
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 8.95724105834961,
+      "learning_rate": 4.810488986240858e-05,
+      "loss": 2.2008,
+      "step": 7170
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 9.038932800292969,
+      "learning_rate": 4.8098813148683356e-05,
+      "loss": 2.1931,
+      "step": 7180
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.18593692779541,
+      "learning_rate": 4.809272709305201e-05,
+      "loss": 2.2143,
+      "step": 7190
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 7.01367712020874,
+      "learning_rate": 4.808663169797594e-05,
+      "loss": 2.1562,
+      "step": 7200
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.437080383300781,
+      "learning_rate": 4.808052696592032e-05,
+      "loss": 2.1143,
+      "step": 7210
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 7.393974304199219,
+      "learning_rate": 4.807441289935411e-05,
+      "loss": 2.4122,
+      "step": 7220
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.011038780212402,
+      "learning_rate": 4.806828950075002e-05,
+      "loss": 2.1467,
+      "step": 7230
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 9.512205123901367,
+      "learning_rate": 4.806215677258456e-05,
+      "loss": 2.2061,
+      "step": 7240
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.728775978088379,
+      "learning_rate": 4.805601471733801e-05,
+      "loss": 2.234,
+      "step": 7250
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 10.132803916931152,
+      "learning_rate": 4.8049863337494383e-05,
+      "loss": 2.249,
+      "step": 7260
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 11.986092567443848,
+      "learning_rate": 4.8043702635541534e-05,
+      "loss": 2.2069,
+      "step": 7270
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.403532981872559,
+      "learning_rate": 4.8037532613971035e-05,
+      "loss": 2.1335,
+      "step": 7280
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 28.95196533203125,
+      "learning_rate": 4.803135327527825e-05,
+      "loss": 2.3068,
+      "step": 7290
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 7.692062854766846,
+      "learning_rate": 4.8025164621962284e-05,
+      "loss": 2.4062,
+      "step": 7300
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.148292541503906,
+      "learning_rate": 4.801896665652603e-05,
+      "loss": 2.2943,
+      "step": 7310
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 7.389751434326172,
+      "learning_rate": 4.801275938147617e-05,
+      "loss": 2.2432,
+      "step": 7320
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 9.887344360351562,
+      "learning_rate": 4.800654279932311e-05,
+      "loss": 2.3278,
+      "step": 7330
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.3506927490234375,
+      "learning_rate": 4.800031691258103e-05,
+      "loss": 2.1808,
+      "step": 7340
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 7.447039604187012,
+      "learning_rate": 4.79940817237679e-05,
+      "loss": 2.3918,
+      "step": 7350
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 7.131879806518555,
+      "learning_rate": 4.798783723540541e-05,
+      "loss": 2.3571,
+      "step": 7360
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 9.371427536010742,
+      "learning_rate": 4.7981583450019045e-05,
+      "loss": 2.3176,
+      "step": 7370
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.0981903076171875,
+      "learning_rate": 4.797532037013803e-05,
+      "loss": 2.2672,
+      "step": 7380
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.475356578826904,
+      "learning_rate": 4.796904799829537e-05,
+      "loss": 2.1954,
+      "step": 7390
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.3766937255859375,
+      "learning_rate": 4.796276633702781e-05,
+      "loss": 2.0606,
+      "step": 7400
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.111029624938965,
+      "learning_rate": 4.795647538887584e-05,
+      "loss": 2.148,
+      "step": 7410
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.0170817375183105,
+      "learning_rate": 4.795017515638375e-05,
+      "loss": 1.9902,
+      "step": 7420
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.29054594039917,
+      "learning_rate": 4.794386564209953e-05,
+      "loss": 2.0952,
+      "step": 7430
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.003581523895264,
+      "learning_rate": 4.793754684857497e-05,
+      "loss": 1.987,
+      "step": 7440
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.239002227783203,
+      "learning_rate": 4.7931218778365594e-05,
+      "loss": 2.1665,
+      "step": 7450
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.7367377281188965,
+      "learning_rate": 4.792488143403067e-05,
+      "loss": 2.1673,
+      "step": 7460
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.604746341705322,
+      "learning_rate": 4.791853481813322e-05,
+      "loss": 2.1562,
+      "step": 7470
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.525786876678467,
+      "learning_rate": 4.7912178933240035e-05,
+      "loss": 2.1059,
+      "step": 7480
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.215798377990723,
+      "learning_rate": 4.790581378192164e-05,
+      "loss": 2.1964,
+      "step": 7490
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 21.55533790588379,
+      "learning_rate": 4.7899439366752294e-05,
+      "loss": 2.2484,
+      "step": 7500
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 1.8438102006912231,
+      "eval_runtime": 423.8305,
+      "eval_samples_per_second": 50.07,
+      "eval_steps_per_second": 0.392,
+      "step": 7500
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.74520206451416,
+      "learning_rate": 4.789305569031002e-05,
+      "loss": 2.0724,
+      "step": 7510
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.096541404724121,
+      "learning_rate": 4.7886662755176594e-05,
+      "loss": 2.142,
+      "step": 7520
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.870784282684326,
+      "learning_rate": 4.7880260563937515e-05,
+      "loss": 2.0857,
+      "step": 7530
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.653884410858154,
+      "learning_rate": 4.787384911918204e-05,
+      "loss": 2.1568,
+      "step": 7540
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.95023775100708,
+      "learning_rate": 4.786742842350316e-05,
+      "loss": 2.1114,
+      "step": 7550
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.527095317840576,
+      "learning_rate": 4.786099847949761e-05,
+      "loss": 2.1395,
+      "step": 7560
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.1902899742126465,
+      "learning_rate": 4.785455928976588e-05,
+      "loss": 1.9845,
+      "step": 7570
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.819202899932861,
+      "learning_rate": 4.784811085691218e-05,
+      "loss": 2.0774,
+      "step": 7580
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.495123386383057,
+      "learning_rate": 4.7841653183544455e-05,
+      "loss": 2.2617,
+      "step": 7590
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.861876487731934,
+      "learning_rate": 4.783518627227441e-05,
+      "loss": 2.313,
+      "step": 7600
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.845401763916016,
+      "learning_rate": 4.782871012571747e-05,
+      "loss": 2.1452,
+      "step": 7610
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.459070205688477,
+      "learning_rate": 4.782222474649279e-05,
+      "loss": 1.91,
+      "step": 7620
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.476640224456787,
+      "learning_rate": 4.7815730137223273e-05,
+      "loss": 2.1809,
+      "step": 7630
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 8.694378852844238,
+      "learning_rate": 4.780922630053555e-05,
+      "loss": 2.0834,
+      "step": 7640
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 7.8879923820495605,
+      "learning_rate": 4.780271323905998e-05,
+      "loss": 2.2303,
+      "step": 7650
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 7.245183944702148,
+      "learning_rate": 4.7796190955430655e-05,
+      "loss": 1.9748,
+      "step": 7660
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.349466323852539,
+      "learning_rate": 4.7789659452285394e-05,
+      "loss": 2.1068,
+      "step": 7670
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 11.410516738891602,
+      "learning_rate": 4.778311873226575e-05,
+      "loss": 2.2332,
+      "step": 7680
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.053547382354736,
+      "learning_rate": 4.777656879801701e-05,
+      "loss": 2.0781,
+      "step": 7690
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.239130973815918,
+      "learning_rate": 4.777000965218817e-05,
+      "loss": 2.1827,
+      "step": 7700
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.953700542449951,
+      "learning_rate": 4.776344129743196e-05,
+      "loss": 2.2209,
+      "step": 7710
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 7.110509872436523,
+      "learning_rate": 4.775686373640484e-05,
+      "loss": 2.3088,
+      "step": 7720
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.350749969482422,
+      "learning_rate": 4.7750276971766996e-05,
+      "loss": 1.9362,
+      "step": 7730
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.501804828643799,
+      "learning_rate": 4.774368100618231e-05,
+      "loss": 1.9958,
+      "step": 7740
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.133676052093506,
+      "learning_rate": 4.77370758423184e-05,
+      "loss": 2.0108,
+      "step": 7750
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.481894016265869,
+      "learning_rate": 4.773046148284663e-05,
+      "loss": 2.0169,
+      "step": 7760
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 21.402868270874023,
+      "learning_rate": 4.772383793044205e-05,
+      "loss": 1.9482,
+      "step": 7770
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.794774532318115,
+      "learning_rate": 4.771720518778344e-05,
+      "loss": 2.1169,
+      "step": 7780
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.211555480957031,
+      "learning_rate": 4.7710563257553286e-05,
+      "loss": 2.1805,
+      "step": 7790
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.304569244384766,
+      "learning_rate": 4.7703912142437815e-05,
+      "loss": 2.0067,
+      "step": 7800
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.812686920166016,
+      "learning_rate": 4.769725184512694e-05,
+      "loss": 2.0295,
+      "step": 7810
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 7.2158308029174805,
+      "learning_rate": 4.7690582368314304e-05,
+      "loss": 2.1737,
+      "step": 7820
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.527587890625,
+      "learning_rate": 4.7683903714697264e-05,
+      "loss": 2.099,
+      "step": 7830
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.601354122161865,
+      "learning_rate": 4.767721588697688e-05,
+      "loss": 2.308,
+      "step": 7840
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 7.4081711769104,
+      "learning_rate": 4.7670518887857916e-05,
+      "loss": 2.1246,
+      "step": 7850
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.526096820831299,
+      "learning_rate": 4.766381272004887e-05,
+      "loss": 2.1721,
+      "step": 7860
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.237331867218018,
+      "learning_rate": 4.765709738626192e-05,
+      "loss": 2.0706,
+      "step": 7870
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 11.353988647460938,
+      "learning_rate": 4.7650372889212976e-05,
+      "loss": 1.929,
+      "step": 7880
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 9.323110580444336,
+      "learning_rate": 4.7643639231621626e-05,
+      "loss": 2.0205,
+      "step": 7890
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.578587055206299,
+      "learning_rate": 4.76368964162112e-05,
+      "loss": 1.9712,
+      "step": 7900
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.492863178253174,
+      "learning_rate": 4.76301444457087e-05,
+      "loss": 2.1388,
+      "step": 7910
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 8.908137321472168,
+      "learning_rate": 4.7623383322844835e-05,
+      "loss": 1.9039,
+      "step": 7920
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.375561714172363,
+      "learning_rate": 4.7616613050354025e-05,
+      "loss": 2.0121,
+      "step": 7930
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.633610248565674,
+      "learning_rate": 4.760983363097439e-05,
+      "loss": 2.0154,
+      "step": 7940
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.779325008392334,
+      "learning_rate": 4.760304506744774e-05,
+      "loss": 2.0507,
+      "step": 7950
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.787586688995361,
+      "learning_rate": 4.75962473625196e-05,
+      "loss": 2.078,
+      "step": 7960
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.715172290802002,
+      "learning_rate": 4.758944051893917e-05,
+      "loss": 2.063,
+      "step": 7970
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.478941440582275,
+      "learning_rate": 4.7582624539459366e-05,
+      "loss": 2.0065,
+      "step": 7980
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.507533073425293,
+      "learning_rate": 4.7575799426836785e-05,
+      "loss": 1.9767,
+      "step": 7990
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.369671821594238,
+      "learning_rate": 4.7568965183831726e-05,
+      "loss": 1.9402,
+      "step": 8000
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 1.7692859172821045,
+      "eval_runtime": 423.4176,
+      "eval_samples_per_second": 50.118,
+      "eval_steps_per_second": 0.392,
+      "step": 8000
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.809439659118652,
+      "learning_rate": 4.756212181320817e-05,
+      "loss": 1.9837,
+      "step": 8010
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.8267292976379395,
+      "learning_rate": 4.7555269317733806e-05,
+      "loss": 1.9497,
+      "step": 8020
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.9060163497924805,
+      "learning_rate": 4.754840770018001e-05,
+      "loss": 2.0145,
+      "step": 8030
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.141025066375732,
+      "learning_rate": 4.7541536963321834e-05,
+      "loss": 1.8596,
+      "step": 8040
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 7.8809123039245605,
+      "learning_rate": 4.753465710993801e-05,
+      "loss": 2.0378,
+      "step": 8050
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.924752235412598,
+      "learning_rate": 4.7527768142811003e-05,
+      "loss": 2.1096,
+      "step": 8060
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.651288986206055,
+      "learning_rate": 4.7520870064726916e-05,
+      "loss": 1.8447,
+      "step": 8070
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.8274922370910645,
+      "learning_rate": 4.751396287847556e-05,
+      "loss": 1.9829,
+      "step": 8080
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.125004291534424,
+      "learning_rate": 4.750704658685041e-05,
+      "loss": 2.1189,
+      "step": 8090
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 88.02458953857422,
+      "learning_rate": 4.750012119264866e-05,
+      "loss": 2.101,
+      "step": 8100
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 10.099868774414062,
+      "learning_rate": 4.749318669867115e-05,
+      "loss": 2.1791,
+      "step": 8110
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 7.033977508544922,
+      "learning_rate": 4.748624310772242e-05,
+      "loss": 2.0594,
+      "step": 8120
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.70294189453125,
+      "learning_rate": 4.747929042261068e-05,
+      "loss": 1.9056,
+      "step": 8130
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.636380672454834,
+      "learning_rate": 4.7472328646147814e-05,
+      "loss": 2.0761,
+      "step": 8140
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.38163948059082,
+      "learning_rate": 4.74653577811494e-05,
+      "loss": 1.9754,
+      "step": 8150
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.055973529815674,
+      "learning_rate": 4.7458377830434676e-05,
+      "loss": 1.8625,
+      "step": 8160
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.357255458831787,
+      "learning_rate": 4.7451388796826545e-05,
+      "loss": 1.8988,
+      "step": 8170
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.224727153778076,
+      "learning_rate": 4.744439068315163e-05,
+      "loss": 1.9152,
+      "step": 8180
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.7499165534973145,
+      "learning_rate": 4.743738349224016e-05,
+      "loss": 2.3241,
+      "step": 8190
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.902726650238037,
+      "learning_rate": 4.74303672269261e-05,
+      "loss": 2.0803,
+      "step": 8200
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 9.815139770507812,
+      "learning_rate": 4.742334189004704e-05,
+      "loss": 2.0549,
+      "step": 8210
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.603668689727783,
+      "learning_rate": 4.741630748444424e-05,
+      "loss": 2.1398,
+      "step": 8220
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.725936412811279,
+      "learning_rate": 4.740926401296266e-05,
+      "loss": 1.97,
+      "step": 8230
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.910879611968994,
+      "learning_rate": 4.74022114784509e-05,
+      "loss": 1.8817,
+      "step": 8240
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.222038745880127,
+      "learning_rate": 4.7395149883761235e-05,
+      "loss": 1.7971,
+      "step": 8250
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.179903507232666,
+      "learning_rate": 4.73880792317496e-05,
+      "loss": 1.8004,
+      "step": 8260
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.983460903167725,
+      "learning_rate": 4.738099952527559e-05,
+      "loss": 2.1015,
+      "step": 8270
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.896063804626465,
+      "learning_rate": 4.737391076720249e-05,
+      "loss": 2.1134,
+      "step": 8280
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.984912872314453,
+      "learning_rate": 4.736681296039719e-05,
+      "loss": 2.1292,
+      "step": 8290
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.538473606109619,
+      "learning_rate": 4.7359706107730295e-05,
+      "loss": 1.9702,
+      "step": 8300
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.9707231521606445,
+      "learning_rate": 4.7352590212076034e-05,
+      "loss": 1.771,
+      "step": 8310
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.303897380828857,
+      "learning_rate": 4.734546527631232e-05,
+      "loss": 2.0974,
+      "step": 8320
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.653863906860352,
+      "learning_rate": 4.733833130332068e-05,
+      "loss": 1.8675,
+      "step": 8330
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 7.186466217041016,
+      "learning_rate": 4.733118829598635e-05,
+      "loss": 2.1016,
+      "step": 8340
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.940579414367676,
+      "learning_rate": 4.732403625719818e-05,
+      "loss": 2.1889,
+      "step": 8350
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.717761516571045,
+      "learning_rate": 4.73168751898487e-05,
+      "loss": 1.7199,
+      "step": 8360
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.8094892501831055,
+      "learning_rate": 4.730970509683406e-05,
+      "loss": 1.958,
+      "step": 8370
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.522322177886963,
+      "learning_rate": 4.730252598105407e-05,
+      "loss": 1.9033,
+      "step": 8380
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.876448631286621,
+      "learning_rate": 4.7295337845412216e-05,
+      "loss": 1.9084,
+      "step": 8390
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.507818222045898,
+      "learning_rate": 4.7288140692815605e-05,
+      "loss": 1.8546,
+      "step": 8400
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.461522102355957,
+      "learning_rate": 4.7280934526175e-05,
+      "loss": 1.9473,
+      "step": 8410
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.854064464569092,
+      "learning_rate": 4.727371934840481e-05,
+      "loss": 1.8768,
+      "step": 8420
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.8991923332214355,
+      "learning_rate": 4.726649516242307e-05,
+      "loss": 2.0273,
+      "step": 8430
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.412707805633545,
+      "learning_rate": 4.7259261971151494e-05,
+      "loss": 2.057,
+      "step": 8440
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.540420055389404,
+      "learning_rate": 4.725201977751541e-05,
+      "loss": 1.7884,
+      "step": 8450
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.3580241203308105,
+      "learning_rate": 4.724476858444379e-05,
+      "loss": 1.9676,
+      "step": 8460
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.679504871368408,
+      "learning_rate": 4.723750839486926e-05,
+      "loss": 2.0706,
+      "step": 8470
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.0920538902282715,
+      "learning_rate": 4.723023921172807e-05,
+      "loss": 1.9423,
+      "step": 8480
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.097662448883057,
+      "learning_rate": 4.722296103796011e-05,
+      "loss": 1.8774,
+      "step": 8490
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 8.337467193603516,
+      "learning_rate": 4.721567387650892e-05,
+      "loss": 2.0885,
+      "step": 8500
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 1.678551435470581,
+      "eval_runtime": 423.6293,
+      "eval_samples_per_second": 50.093,
+      "eval_steps_per_second": 0.392,
+      "step": 8500
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.902133941650391,
+      "learning_rate": 4.7208377730321643e-05,
+      "loss": 2.2287,
+      "step": 8510
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 7.351017951965332,
+      "learning_rate": 4.72010726023491e-05,
+      "loss": 2.1598,
+      "step": 8520
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.693913459777832,
+      "learning_rate": 4.71937584955457e-05,
+      "loss": 1.8859,
+      "step": 8530
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.01932954788208,
+      "learning_rate": 4.7186435412869515e-05,
+      "loss": 1.9043,
+      "step": 8540
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 24.060230255126953,
+      "learning_rate": 4.7179103357282236e-05,
+      "loss": 1.9134,
+      "step": 8550
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.253117561340332,
+      "learning_rate": 4.717176233174918e-05,
+      "loss": 2.0946,
+      "step": 8560
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.719861030578613,
+      "learning_rate": 4.71644123392393e-05,
+      "loss": 1.9406,
+      "step": 8570
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 11.881241798400879,
+      "learning_rate": 4.7157053382725164e-05,
+      "loss": 2.2027,
+      "step": 8580
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.93878173828125,
+      "learning_rate": 4.714968546518297e-05,
+      "loss": 1.914,
+      "step": 8590
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.93536376953125,
+      "learning_rate": 4.7142308589592556e-05,
+      "loss": 2.1651,
+      "step": 8600
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.014473915100098,
+      "learning_rate": 4.7134922758937355e-05,
+      "loss": 1.9013,
+      "step": 8610
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 25.88618278503418,
+      "learning_rate": 4.712752797620444e-05,
+      "loss": 2.01,
+      "step": 8620
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.001585483551025,
+      "learning_rate": 4.71201242443845e-05,
+      "loss": 1.7434,
+      "step": 8630
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 7.327158451080322,
+      "learning_rate": 4.7112711566471834e-05,
+      "loss": 2.107,
+      "step": 8640
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.623456954956055,
+      "learning_rate": 4.7105289945464384e-05,
+      "loss": 1.953,
+      "step": 8650
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.4558868408203125,
+      "learning_rate": 4.7097859384363686e-05,
+      "loss": 1.8178,
+      "step": 8660
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.775446891784668,
+      "learning_rate": 4.70904198861749e-05,
+      "loss": 1.9418,
+      "step": 8670
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 10.967681884765625,
+      "learning_rate": 4.708297145390679e-05,
+      "loss": 1.993,
+      "step": 8680
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.730310916900635,
+      "learning_rate": 4.707551409057176e-05,
+      "loss": 2.0054,
+      "step": 8690
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 9.371408462524414,
+      "learning_rate": 4.70680477991858e-05,
+      "loss": 2.0018,
+      "step": 8700
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 6.070098876953125,
+      "learning_rate": 4.706057258276851e-05,
+      "loss": 2.0201,
+      "step": 8710
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 9.502942085266113,
+      "learning_rate": 4.705308844434313e-05,
+      "loss": 2.1858,
+      "step": 8720
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 10.906458854675293,
+      "learning_rate": 4.704559538693647e-05,
+      "loss": 1.9434,
+      "step": 8730
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 6.904813289642334,
+      "learning_rate": 4.7038093413578975e-05,
+      "loss": 1.8493,
+      "step": 8740
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.065293312072754,
+      "learning_rate": 4.7030582527304675e-05,
+      "loss": 1.7872,
+      "step": 8750
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 11.148314476013184,
+      "learning_rate": 4.702306273115122e-05,
+      "loss": 1.8762,
+      "step": 8760
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 6.202971458435059,
+      "learning_rate": 4.701553402815986e-05,
+      "loss": 2.0413,
+      "step": 8770
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 10.41578197479248,
+      "learning_rate": 4.700799642137544e-05,
+      "loss": 1.9312,
+      "step": 8780
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.611337184906006,
+      "learning_rate": 4.7000449913846424e-05,
+      "loss": 1.9724,
+      "step": 8790
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 49.853824615478516,
+      "learning_rate": 4.699289450862485e-05,
+      "loss": 2.0025,
+      "step": 8800
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 6.44148063659668,
+      "learning_rate": 4.6985330208766375e-05,
+      "loss": 1.8576,
+      "step": 8810
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 6.226993083953857,
+      "learning_rate": 4.6977757017330245e-05,
+      "loss": 1.8268,
+      "step": 8820
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 7.233063220977783,
+      "learning_rate": 4.69701749373793e-05,
+      "loss": 1.8525,
+      "step": 8830
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 7.900531768798828,
+      "learning_rate": 4.696258397197998e-05,
+      "loss": 2.1743,
+      "step": 8840
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 6.996172904968262,
+      "learning_rate": 4.695498412420232e-05,
+      "loss": 1.7183,
+      "step": 8850
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.938333034515381,
+      "learning_rate": 4.694737539711994e-05,
+      "loss": 1.9403,
+      "step": 8860
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.985813140869141,
+      "learning_rate": 4.6939757793810055e-05,
+      "loss": 1.9518,
+      "step": 8870
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 7.172812461853027,
+      "learning_rate": 4.693213131735348e-05,
+      "loss": 2.0953,
+      "step": 8880
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 41.72245407104492,
+      "learning_rate": 4.692449597083459e-05,
+      "loss": 2.0219,
+      "step": 8890
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.960461616516113,
+      "learning_rate": 4.691685175734138e-05,
+      "loss": 1.9091,
+      "step": 8900
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.79845666885376,
+      "learning_rate": 4.6909198679965405e-05,
+      "loss": 1.9832,
+      "step": 8910
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 9.317214965820312,
+      "learning_rate": 4.690153674180183e-05,
+      "loss": 2.2419,
+      "step": 8920
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 6.582641124725342,
+      "learning_rate": 4.689386594594938e-05,
+      "loss": 1.9132,
+      "step": 8930
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.3012471199035645,
+      "learning_rate": 4.688618629551038e-05,
+      "loss": 1.9446,
+      "step": 8940
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 7.338431358337402,
+      "learning_rate": 4.6878497793590725e-05,
+      "loss": 1.8855,
+      "step": 8950
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 6.319138526916504,
+      "learning_rate": 4.6870800443299896e-05,
+      "loss": 1.8556,
+      "step": 8960
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 7.105714321136475,
+      "learning_rate": 4.686309424775094e-05,
+      "loss": 2.1338,
+      "step": 8970
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.138299942016602,
+      "learning_rate": 4.6855379210060506e-05,
+      "loss": 1.8654,
+      "step": 8980
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.727604866027832,
+      "learning_rate": 4.684765533334879e-05,
+      "loss": 2.1185,
+      "step": 8990
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.990983009338379,
+      "learning_rate": 4.68399226207396e-05,
+      "loss": 1.8292,
+      "step": 9000
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 1.639684796333313,
+      "eval_runtime": 422.9949,
+      "eval_samples_per_second": 50.168,
+      "eval_steps_per_second": 0.392,
+      "step": 9000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.5144429206848145,
+      "learning_rate": 4.6832181075360274e-05,
+      "loss": 2.1007,
+      "step": 9010
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 6.92350959777832,
+      "learning_rate": 4.6824430700341736e-05,
+      "loss": 1.9391,
+      "step": 9020
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.39557409286499,
+      "learning_rate": 4.681667149881852e-05,
+      "loss": 1.8487,
+      "step": 9030
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.977539539337158,
+      "learning_rate": 4.680890347392867e-05,
+      "loss": 1.8729,
+      "step": 9040
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 6.647794723510742,
+      "learning_rate": 4.680112662881383e-05,
+      "loss": 2.031,
+      "step": 9050
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.776614189147949,
+      "learning_rate": 4.6793340966619214e-05,
+      "loss": 1.8169,
+      "step": 9060
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 6.00407600402832,
+      "learning_rate": 4.678554649049359e-05,
+      "loss": 1.8746,
+      "step": 9070
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.949479103088379,
+      "learning_rate": 4.677774320358931e-05,
+      "loss": 1.7223,
+      "step": 9080
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 6.128554344177246,
+      "learning_rate": 4.676993110906225e-05,
+      "loss": 1.7069,
+      "step": 9090
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 6.835212230682373,
+      "learning_rate": 4.6762110210071894e-05,
+      "loss": 2.1278,
+      "step": 9100
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.878398895263672,
+      "learning_rate": 4.675428050978127e-05,
+      "loss": 1.9742,
+      "step": 9110
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 6.689033508300781,
+      "learning_rate": 4.674644201135694e-05,
+      "loss": 2.0122,
+      "step": 9120
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.20786190032959,
+      "learning_rate": 4.673859471796905e-05,
+      "loss": 1.7227,
+      "step": 9130
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.527268409729004,
+      "learning_rate": 4.6730738632791325e-05,
+      "loss": 1.9288,
+      "step": 9140
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 10.17988109588623,
+      "learning_rate": 4.672287375900099e-05,
+      "loss": 1.9036,
+      "step": 9150
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 6.694543361663818,
+      "learning_rate": 4.6715000099778864e-05,
+      "loss": 1.9696,
+      "step": 9160
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 9.240964889526367,
+      "learning_rate": 4.6707117658309316e-05,
+      "loss": 1.9683,
+      "step": 9170
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 10.630289077758789,
+      "learning_rate": 4.669922643778025e-05,
+      "loss": 1.8037,
+      "step": 9180
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 6.575771331787109,
+      "learning_rate": 4.669132644138314e-05,
+      "loss": 1.9856,
+      "step": 9190
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.165228843688965,
+      "learning_rate": 4.668341767231299e-05,
+      "loss": 2.033,
+      "step": 9200
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 29.935487747192383,
+      "learning_rate": 4.6675500133768367e-05,
+      "loss": 1.9919,
+      "step": 9210
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.627548694610596,
+      "learning_rate": 4.6667573828951375e-05,
+      "loss": 2.0417,
+      "step": 9220
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 7.543898582458496,
+      "learning_rate": 4.665963876106767e-05,
+      "loss": 1.9762,
+      "step": 9230
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 12.477391242980957,
+      "learning_rate": 4.665169493332646e-05,
+      "loss": 1.8202,
+      "step": 9240
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 8.293655395507812,
+      "learning_rate": 4.6643742348940464e-05,
+      "loss": 1.9062,
+      "step": 9250
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.157840251922607,
+      "learning_rate": 4.663578101112599e-05,
+      "loss": 1.8978,
+      "step": 9260
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.024789810180664,
+      "learning_rate": 4.662781092310283e-05,
+      "loss": 1.658,
+      "step": 9270
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 8.7192964553833,
+      "learning_rate": 4.661983208809437e-05,
+      "loss": 1.8223,
+      "step": 9280
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.595676898956299,
+      "learning_rate": 4.6611844509327505e-05,
+      "loss": 1.8821,
+      "step": 9290
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.086737155914307,
+      "learning_rate": 4.660384819003266e-05,
+      "loss": 1.8998,
+      "step": 9300
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.518343448638916,
+      "learning_rate": 4.65958431334438e-05,
+      "loss": 1.8404,
+      "step": 9310
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.972059726715088,
+      "learning_rate": 4.6587829342798456e-05,
+      "loss": 1.951,
+      "step": 9320
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.087925910949707,
+      "learning_rate": 4.6579806821337634e-05,
+      "loss": 1.7567,
+      "step": 9330
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 7.334329128265381,
+      "learning_rate": 4.657177557230592e-05,
+      "loss": 1.8196,
+      "step": 9340
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.589081287384033,
+      "learning_rate": 4.6563735598951395e-05,
+      "loss": 1.9163,
+      "step": 9350
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.94594144821167,
+      "learning_rate": 4.6555686904525695e-05,
+      "loss": 1.9189,
+      "step": 9360
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 7.6889119148254395,
+      "learning_rate": 4.6547629492283976e-05,
+      "loss": 1.9279,
+      "step": 9370
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.893225193023682,
+      "learning_rate": 4.65395633654849e-05,
+      "loss": 1.8149,
+      "step": 9380
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 7.492513179779053,
+      "learning_rate": 4.653148852739068e-05,
+      "loss": 1.8264,
+      "step": 9390
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.651477813720703,
+      "learning_rate": 4.6523404981267036e-05,
+      "loss": 1.7099,
+      "step": 9400
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.677443504333496,
+      "learning_rate": 4.651531273038323e-05,
+      "loss": 2.0826,
+      "step": 9410
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.65981388092041,
+      "learning_rate": 4.6507211778012004e-05,
+      "loss": 1.8778,
+      "step": 9420
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.33742094039917,
+      "learning_rate": 4.6499102127429665e-05,
+      "loss": 1.976,
+      "step": 9430
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 7.806815147399902,
+      "learning_rate": 4.6490983781916015e-05,
+      "loss": 1.8781,
+      "step": 9440
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 7.299323081970215,
+      "learning_rate": 4.6482856744754364e-05,
+      "loss": 1.8524,
+      "step": 9450
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.100677490234375,
+      "learning_rate": 4.6474721019231566e-05,
+      "loss": 2.0033,
+      "step": 9460
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.778229236602783,
+      "learning_rate": 4.646657660863796e-05,
+      "loss": 1.8675,
+      "step": 9470
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.6370930671691895,
+      "learning_rate": 4.64584235162674e-05,
+      "loss": 1.8061,
+      "step": 9480
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 8.557727813720703,
+      "learning_rate": 4.6450261745417276e-05,
+      "loss": 2.016,
+      "step": 9490
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 7.485627174377441,
+      "learning_rate": 4.6442091299388467e-05,
+      "loss": 1.7278,
+      "step": 9500
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 1.5960439443588257,
+      "eval_runtime": 423.3373,
+      "eval_samples_per_second": 50.128,
+      "eval_steps_per_second": 0.392,
+      "step": 9500
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 7.613509654998779,
+      "learning_rate": 4.643391218148536e-05,
+      "loss": 1.9552,
+      "step": 9510
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.298041820526123,
+      "learning_rate": 4.6425724395015865e-05,
+      "loss": 2.0533,
+      "step": 9520
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.571812629699707,
+      "learning_rate": 4.641752794329137e-05,
+      "loss": 2.0098,
+      "step": 9530
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 7.494356155395508,
+      "learning_rate": 4.64093228296268e-05,
+      "loss": 1.902,
+      "step": 9540
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.5204315185546875,
+      "learning_rate": 4.640110905734057e-05,
+      "loss": 1.9357,
+      "step": 9550
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.17507791519165,
+      "learning_rate": 4.6392886629754586e-05,
+      "loss": 1.6532,
+      "step": 9560
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.883286476135254,
+      "learning_rate": 4.6384655550194256e-05,
+      "loss": 1.8173,
+      "step": 9570
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.756780624389648,
+      "learning_rate": 4.6376415821988514e-05,
+      "loss": 1.6706,
+      "step": 9580
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.555366516113281,
+      "learning_rate": 4.636816744846976e-05,
+      "loss": 1.8445,
+      "step": 9590
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.893411159515381,
+      "learning_rate": 4.6359910432973905e-05,
+      "loss": 1.8346,
+      "step": 9600
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.471080780029297,
+      "learning_rate": 4.635164477884035e-05,
+      "loss": 1.6428,
+      "step": 9610
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.882155895233154,
+      "learning_rate": 4.6343370489412e-05,
+      "loss": 1.7322,
+      "step": 9620
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 18.208683013916016,
+      "learning_rate": 4.633508756803523e-05,
+      "loss": 1.8269,
+      "step": 9630
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 14.348465919494629,
+      "learning_rate": 4.6326796018059936e-05,
+      "loss": 1.8359,
+      "step": 9640
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 7.5844831466674805,
+      "learning_rate": 4.631849584283949e-05,
+      "loss": 1.9765,
+      "step": 9650
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.831000328063965,
+      "learning_rate": 4.631018704573074e-05,
+      "loss": 1.7907,
+      "step": 9660
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.3236188888549805,
+      "learning_rate": 4.630186963009403e-05,
+      "loss": 1.7845,
+      "step": 9670
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 8.6161527633667,
+      "learning_rate": 4.629354359929321e-05,
+      "loss": 1.8121,
+      "step": 9680
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 7.705484390258789,
+      "learning_rate": 4.628520895669558e-05,
+      "loss": 2.0315,
+      "step": 9690
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.727534294128418,
+      "learning_rate": 4.627686570567194e-05,
+      "loss": 1.8766,
+      "step": 9700
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 6.244938850402832,
+      "learning_rate": 4.626851384959657e-05,
+      "loss": 1.7736,
+      "step": 9710
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.889083385467529,
+      "learning_rate": 4.626015339184724e-05,
+      "loss": 1.7259,
+      "step": 9720
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.227889537811279,
+      "learning_rate": 4.625178433580518e-05,
+      "loss": 1.704,
+      "step": 9730
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 6.0999250411987305,
+      "learning_rate": 4.624340668485511e-05,
+      "loss": 1.9576,
+      "step": 9740
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 14.676617622375488,
+      "learning_rate": 4.623502044238521e-05,
+      "loss": 1.8773,
+      "step": 9750
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.836556911468506,
+      "learning_rate": 4.6226625611787175e-05,
+      "loss": 1.8428,
+      "step": 9760
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 7.364138126373291,
+      "learning_rate": 4.621822219645612e-05,
+      "loss": 1.8232,
+      "step": 9770
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 12.476197242736816,
+      "learning_rate": 4.620981019979067e-05,
+      "loss": 2.0284,
+      "step": 9780
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 33.210792541503906,
+      "learning_rate": 4.62013896251929e-05,
+      "loss": 1.7304,
+      "step": 9790
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 7.111621856689453,
+      "learning_rate": 4.619296047606837e-05,
+      "loss": 1.9921,
+      "step": 9800
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 8.426152229309082,
+      "learning_rate": 4.6184522755826105e-05,
+      "loss": 1.6146,
+      "step": 9810
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 19.042428970336914,
+      "learning_rate": 4.617607646787858e-05,
+      "loss": 1.788,
+      "step": 9820
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 34.727783203125,
+      "learning_rate": 4.6167621615641746e-05,
+      "loss": 1.8662,
+      "step": 9830
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 11.465919494628906,
+      "learning_rate": 4.6159158202535026e-05,
+      "loss": 1.8507,
+      "step": 9840
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.54009485244751,
+      "learning_rate": 4.61506862319813e-05,
+      "loss": 1.6138,
+      "step": 9850
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 8.663006782531738,
+      "learning_rate": 4.614220570740689e-05,
+      "loss": 1.7006,
+      "step": 9860
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 12.138349533081055,
+      "learning_rate": 4.613371663224162e-05,
+      "loss": 2.0698,
+      "step": 9870
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.74115514755249,
+      "learning_rate": 4.612521900991872e-05,
+      "loss": 1.7282,
+      "step": 9880
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 11.925437927246094,
+      "learning_rate": 4.611671284387492e-05,
+      "loss": 1.6966,
+      "step": 9890
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 8.422083854675293,
+      "learning_rate": 4.610819813755038e-05,
+      "loss": 1.9501,
+      "step": 9900
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.11015510559082,
+      "learning_rate": 4.609967489438873e-05,
+      "loss": 1.6498,
+      "step": 9910
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.888397693634033,
+      "learning_rate": 4.609114311783703e-05,
+      "loss": 1.6546,
+      "step": 9920
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 17.405601501464844,
+      "learning_rate": 4.608260281134582e-05,
+      "loss": 1.7784,
+      "step": 9930
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 10.067267417907715,
+      "learning_rate": 4.607405397836908e-05,
+      "loss": 1.8552,
+      "step": 9940
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.969694137573242,
+      "learning_rate": 4.6065496622364224e-05,
+      "loss": 1.8743,
+      "step": 9950
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 6.426912784576416,
+      "learning_rate": 4.605693074679211e-05,
+      "loss": 1.8538,
+      "step": 9960
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 7.665450096130371,
+      "learning_rate": 4.604835635511709e-05,
+      "loss": 1.8622,
+      "step": 9970
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 6.462819576263428,
+      "learning_rate": 4.60397734508069e-05,
+      "loss": 2.0098,
+      "step": 9980
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 6.3902082443237305,
+      "learning_rate": 4.6031182037332745e-05,
+      "loss": 1.7431,
+      "step": 9990
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.557443618774414,
+      "learning_rate": 4.602258211816927e-05,
+      "loss": 1.6484,
+      "step": 10000
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 1.5718907117843628,
+      "eval_runtime": 423.3203,
+      "eval_samples_per_second": 50.13,
+      "eval_steps_per_second": 0.392,
+      "step": 10000
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.951988697052002,
+      "learning_rate": 4.601397369679457e-05,
+      "loss": 1.8044,
+      "step": 10010
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 7.188987731933594,
+      "learning_rate": 4.600535677669016e-05,
+      "loss": 1.8985,
+      "step": 10020
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 6.467220783233643,
+      "learning_rate": 4.5996731361340994e-05,
+      "loss": 1.7658,
+      "step": 10030
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.105921268463135,
+      "learning_rate": 4.598809745423549e-05,
+      "loss": 1.8672,
+      "step": 10040
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.458418846130371,
+      "learning_rate": 4.597945505886546e-05,
+      "loss": 2.0975,
+      "step": 10050
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.801342010498047,
+      "learning_rate": 4.5970804178726176e-05,
+      "loss": 1.9141,
+      "step": 10060
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.686854839324951,
+      "learning_rate": 4.596214481731634e-05,
+      "loss": 1.7676,
+      "step": 10070
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 7.215950012207031,
+      "learning_rate": 4.595347697813806e-05,
+      "loss": 1.775,
+      "step": 10080
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 7.41530179977417,
+      "learning_rate": 4.59448006646969e-05,
+      "loss": 1.79,
+      "step": 10090
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.15520715713501,
+      "learning_rate": 4.5936115880501845e-05,
+      "loss": 1.8232,
+      "step": 10100
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.490520000457764,
+      "learning_rate": 4.59274226290653e-05,
+      "loss": 1.7322,
+      "step": 10110
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 6.287522792816162,
+      "learning_rate": 4.591872091390309e-05,
+      "loss": 1.8003,
+      "step": 10120
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 20.15962791442871,
+      "learning_rate": 4.591001073853448e-05,
+      "loss": 1.8694,
+      "step": 10130
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.3742175102233887,
+      "learning_rate": 4.590129210648213e-05,
+      "loss": 1.6749,
+      "step": 10140
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 6.802846908569336,
+      "learning_rate": 4.5892565021272154e-05,
+      "loss": 1.8907,
+      "step": 10150
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 11.191157341003418,
+      "learning_rate": 4.588382948643406e-05,
+      "loss": 1.996,
+      "step": 10160
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 6.665444850921631,
+      "learning_rate": 4.587508550550078e-05,
+      "loss": 1.8376,
+      "step": 10170
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.446608543395996,
+      "learning_rate": 4.586633308200866e-05,
+      "loss": 1.8742,
+      "step": 10180
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.561806678771973,
+      "learning_rate": 4.585757221949747e-05,
+      "loss": 1.9791,
+      "step": 10190
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.850062847137451,
+      "learning_rate": 4.584880292151038e-05,
+      "loss": 1.786,
+      "step": 10200
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 6.454400539398193,
+      "learning_rate": 4.5840025191593984e-05,
+      "loss": 1.755,
+      "step": 10210
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.614055633544922,
+      "learning_rate": 4.583123903329827e-05,
+      "loss": 1.9708,
+      "step": 10220
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.978786468505859,
+      "learning_rate": 4.582244445017665e-05,
+      "loss": 1.6388,
+      "step": 10230
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.441361427307129,
+      "learning_rate": 4.581364144578593e-05,
+      "loss": 1.7346,
+      "step": 10240
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.76068115234375,
+      "learning_rate": 4.580483002368634e-05,
+      "loss": 1.9962,
+      "step": 10250
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 6.45198392868042,
+      "learning_rate": 4.57960101874415e-05,
+      "loss": 1.7792,
+      "step": 10260
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.766374111175537,
+      "learning_rate": 4.5787181940618435e-05,
+      "loss": 1.5923,
+      "step": 10270
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.983777046203613,
+      "learning_rate": 4.577834528678757e-05,
+      "loss": 1.8081,
+      "step": 10280
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.470437526702881,
+      "learning_rate": 4.576950022952274e-05,
+      "loss": 1.7369,
+      "step": 10290
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.393801212310791,
+      "learning_rate": 4.5760646772401175e-05,
+      "loss": 1.8261,
+      "step": 10300
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 15.60527229309082,
+      "learning_rate": 4.575178491900349e-05,
+      "loss": 1.6928,
+      "step": 10310
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 6.9888916015625,
+      "learning_rate": 4.574291467291371e-05,
+      "loss": 1.9217,
+      "step": 10320
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.985001564025879,
+      "learning_rate": 4.5734036037719236e-05,
+      "loss": 1.7271,
+      "step": 10330
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.762869834899902,
+      "learning_rate": 4.5725149017010896e-05,
+      "loss": 1.686,
+      "step": 10340
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.406029224395752,
+      "learning_rate": 4.5716253614382866e-05,
+      "loss": 1.7915,
+      "step": 10350
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.7803168296813965,
+      "learning_rate": 4.570734983343276e-05,
+      "loss": 1.9539,
+      "step": 10360
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 7.472147464752197,
+      "learning_rate": 4.569843767776153e-05,
+      "loss": 1.7136,
+      "step": 10370
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 7.438866138458252,
+      "learning_rate": 4.568951715097356e-05,
+      "loss": 1.7875,
+      "step": 10380
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 6.272189140319824,
+      "learning_rate": 4.568058825667658e-05,
+      "loss": 1.8589,
+      "step": 10390
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.803892612457275,
+      "learning_rate": 4.5671650998481735e-05,
+      "loss": 1.5895,
+      "step": 10400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 7.044991493225098,
+      "learning_rate": 4.566270538000354e-05,
+      "loss": 1.7368,
+      "step": 10410
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 6.217586040496826,
+      "learning_rate": 4.565375140485989e-05,
+      "loss": 1.6122,
+      "step": 10420
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.539994239807129,
+      "learning_rate": 4.564478907667207e-05,
+      "loss": 1.5242,
+      "step": 10430
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.166835784912109,
+      "learning_rate": 4.563581839906471e-05,
+      "loss": 1.8427,
+      "step": 10440
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 6.602473258972168,
+      "learning_rate": 4.5626839375665876e-05,
+      "loss": 1.8977,
+      "step": 10450
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.6593120098114014,
+      "learning_rate": 4.561785201010695e-05,
+      "loss": 1.5374,
+      "step": 10460
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.412270545959473,
+      "learning_rate": 4.560885630602272e-05,
+      "loss": 1.6548,
+      "step": 10470
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 7.576125621795654,
+      "learning_rate": 4.559985226705135e-05,
+      "loss": 1.6514,
+      "step": 10480
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 6.029335975646973,
+      "learning_rate": 4.559083989683434e-05,
+      "loss": 1.7833,
+      "step": 10490
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 6.206794738769531,
+      "learning_rate": 4.558181919901661e-05,
+      "loss": 1.8478,
+      "step": 10500
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 1.5044246912002563,
+      "eval_runtime": 423.6381,
+      "eval_samples_per_second": 50.092,
+      "eval_steps_per_second": 0.392,
+      "step": 10500
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 6.662048816680908,
+      "learning_rate": 4.5572790177246406e-05,
+      "loss": 1.6828,
+      "step": 10510
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.3925862312316895,
+      "learning_rate": 4.5563752835175354e-05,
+      "loss": 1.7448,
+      "step": 10520
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 6.040633201599121,
+      "learning_rate": 4.555470717645846e-05,
+      "loss": 1.6673,
+      "step": 10530
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.8295183181762695,
+      "learning_rate": 4.5545653204754065e-05,
+      "loss": 1.5644,
+      "step": 10540
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 6.955106735229492,
+      "learning_rate": 4.5536590923723906e-05,
+      "loss": 1.819,
+      "step": 10550
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 6.5708818435668945,
+      "learning_rate": 4.552752033703305e-05,
+      "loss": 1.7287,
+      "step": 10560
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.586352825164795,
+      "learning_rate": 4.5518441448349935e-05,
+      "loss": 1.5666,
+      "step": 10570
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 6.7832932472229,
+      "learning_rate": 4.550935426134636e-05,
+      "loss": 1.9005,
+      "step": 10580
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.757806777954102,
+      "learning_rate": 4.550025877969748e-05,
+      "loss": 1.7659,
+      "step": 10590
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.320318698883057,
+      "learning_rate": 4.54911550070818e-05,
+      "loss": 1.6198,
+      "step": 10600
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.772813320159912,
+      "learning_rate": 4.548204294718117e-05,
+      "loss": 2.1057,
+      "step": 10610
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.337530612945557,
+      "learning_rate": 4.5472922603680805e-05,
+      "loss": 1.8698,
+      "step": 10620
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.934597015380859,
+      "learning_rate": 4.5463793980269276e-05,
+      "loss": 1.7916,
+      "step": 10630
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 6.132730007171631,
+      "learning_rate": 4.5454657080638485e-05,
+      "loss": 1.6805,
+      "step": 10640
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 6.090786933898926,
+      "learning_rate": 4.54455119084837e-05,
+      "loss": 1.8776,
+      "step": 10650
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 6.216485023498535,
+      "learning_rate": 4.54363584675035e-05,
+      "loss": 1.6573,
+      "step": 10660
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.11155891418457,
+      "learning_rate": 4.542719676139987e-05,
+      "loss": 1.6054,
+      "step": 10670
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.956602096557617,
+      "learning_rate": 4.541802679387806e-05,
+      "loss": 1.6549,
+      "step": 10680
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.46177339553833,
+      "learning_rate": 4.540884856864672e-05,
+      "loss": 1.6097,
+      "step": 10690
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 6.2392578125,
+      "learning_rate": 4.539966208941783e-05,
+      "loss": 1.9215,
+      "step": 10700
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.628352165222168,
+      "learning_rate": 4.539046735990667e-05,
+      "loss": 1.7484,
+      "step": 10710
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 8.802276611328125,
+      "learning_rate": 4.538126438383192e-05,
+      "loss": 1.5186,
+      "step": 10720
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 6.957813262939453,
+      "learning_rate": 4.537205316491554e-05,
+      "loss": 1.6704,
+      "step": 10730
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 6.49714994430542,
+      "learning_rate": 4.536283370688286e-05,
+      "loss": 1.7561,
+      "step": 10740
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.772355556488037,
+      "learning_rate": 4.53536060134625e-05,
+      "loss": 1.6856,
+      "step": 10750
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.173638343811035,
+      "learning_rate": 4.5344370088386455e-05,
+      "loss": 1.6953,
+      "step": 10760
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.821722030639648,
+      "learning_rate": 4.533512593539004e-05,
+      "loss": 2.0994,
+      "step": 10770
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.338065147399902,
+      "learning_rate": 4.5325873558211875e-05,
+      "loss": 1.7117,
+      "step": 10780
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.492509365081787,
+      "learning_rate": 4.531661296059392e-05,
+      "loss": 1.7399,
+      "step": 10790
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.744643688201904,
+      "learning_rate": 4.530734414628146e-05,
+      "loss": 1.7788,
+      "step": 10800
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.065397262573242,
+      "learning_rate": 4.5298067119023114e-05,
+      "loss": 1.603,
+      "step": 10810
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 6.232301235198975,
+      "learning_rate": 4.5288781882570805e-05,
+      "loss": 1.6918,
+      "step": 10820
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 6.795167446136475,
+      "learning_rate": 4.527948844067977e-05,
+      "loss": 1.8678,
+      "step": 10830
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.472856521606445,
+      "learning_rate": 4.52701867971086e-05,
+      "loss": 1.6887,
+      "step": 10840
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.1734843254089355,
+      "learning_rate": 4.526087695561917e-05,
+      "loss": 1.5962,
+      "step": 10850
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.166112422943115,
+      "learning_rate": 4.525155891997668e-05,
+      "loss": 1.5192,
+      "step": 10860
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 6.10178279876709,
+      "learning_rate": 4.524223269394963e-05,
+      "loss": 1.6182,
+      "step": 10870
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 7.099591255187988,
+      "learning_rate": 4.5232898281309874e-05,
+      "loss": 1.862,
+      "step": 10880
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.071378231048584,
+      "learning_rate": 4.522355568583253e-05,
+      "loss": 1.5438,
+      "step": 10890
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.862138748168945,
+      "learning_rate": 4.521420491129605e-05,
+      "loss": 1.7539,
+      "step": 10900
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.930329322814941,
+      "learning_rate": 4.5204845961482204e-05,
+      "loss": 1.7749,
+      "step": 10910
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.047465801239014,
+      "learning_rate": 4.519547884017603e-05,
+      "loss": 1.8615,
+      "step": 10920
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.920933723449707,
+      "learning_rate": 4.518610355116592e-05,
+      "loss": 1.5318,
+      "step": 10930
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.903204441070557,
+      "learning_rate": 4.517672009824351e-05,
+      "loss": 1.7197,
+      "step": 10940
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.1994242668151855,
+      "learning_rate": 4.5167328485203796e-05,
+      "loss": 1.623,
+      "step": 10950
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.404425621032715,
+      "learning_rate": 4.515792871584505e-05,
+      "loss": 1.6837,
+      "step": 10960
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.51246976852417,
+      "learning_rate": 4.514852079396884e-05,
+      "loss": 1.5511,
+      "step": 10970
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.787316799163818,
+      "learning_rate": 4.513910472338002e-05,
+      "loss": 1.6386,
+      "step": 10980
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 6.080197811126709,
+      "learning_rate": 4.5129680507886764e-05,
+      "loss": 1.5743,
+      "step": 10990
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 6.592625141143799,
+      "learning_rate": 4.5120248151300535e-05,
+      "loss": 1.8016,
+      "step": 11000
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 1.4656002521514893,
+      "eval_runtime": 423.3562,
+      "eval_samples_per_second": 50.126,
+      "eval_steps_per_second": 0.392,
+      "step": 11000
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 6.005305290222168,
+      "learning_rate": 4.5110807657436075e-05,
+      "loss": 1.6975,
+      "step": 11010
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.094872951507568,
+      "learning_rate": 4.510135903011142e-05,
+      "loss": 1.6844,
+      "step": 11020
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.709961414337158,
+      "learning_rate": 4.509190227314791e-05,
+      "loss": 1.7298,
+      "step": 11030
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.193809509277344,
+      "learning_rate": 4.508243739037016e-05,
+      "loss": 1.7358,
+      "step": 11040
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.351953983306885,
+      "learning_rate": 4.507296438560607e-05,
+      "loss": 1.4251,
+      "step": 11050
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.901798248291016,
+      "learning_rate": 4.506348326268683e-05,
+      "loss": 1.6316,
+      "step": 11060
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.888655185699463,
+      "learning_rate": 4.505399402544692e-05,
+      "loss": 1.6291,
+      "step": 11070
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.988673210144043,
+      "learning_rate": 4.5044496677724086e-05,
+      "loss": 1.7475,
+      "step": 11080
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 6.504331588745117,
+      "learning_rate": 4.503499122335937e-05,
+      "loss": 1.6634,
+      "step": 11090
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 6.307519912719727,
+      "learning_rate": 4.5025477666197066e-05,
+      "loss": 1.8217,
+      "step": 11100
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.95925760269165,
+      "learning_rate": 4.501595601008479e-05,
+      "loss": 1.7216,
+      "step": 11110
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 49.420379638671875,
+      "learning_rate": 4.500642625887339e-05,
+      "loss": 1.8224,
+      "step": 11120
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.760485649108887,
+      "learning_rate": 4.499688841641701e-05,
+      "loss": 1.6322,
+      "step": 11130
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.579456806182861,
+      "learning_rate": 4.498734248657306e-05,
+      "loss": 1.6197,
+      "step": 11140
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 6.184917449951172,
+      "learning_rate": 4.497778847320223e-05,
+      "loss": 1.6196,
+      "step": 11150
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.741893768310547,
+      "learning_rate": 4.496822638016846e-05,
+      "loss": 1.6867,
+      "step": 11160
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.965624809265137,
+      "learning_rate": 4.495865621133897e-05,
+      "loss": 1.3808,
+      "step": 11170
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 6.604211807250977,
+      "learning_rate": 4.494907797058425e-05,
+      "loss": 1.5266,
+      "step": 11180
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.604640483856201,
+      "learning_rate": 4.493949166177806e-05,
+      "loss": 1.6457,
+      "step": 11190
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.649612903594971,
+      "learning_rate": 4.49298972887974e-05,
+      "loss": 1.6705,
+      "step": 11200
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 6.1628031730651855,
+      "learning_rate": 4.492029485552255e-05,
+      "loss": 1.6294,
+      "step": 11210
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.706023216247559,
+      "learning_rate": 4.491068436583703e-05,
+      "loss": 1.5566,
+      "step": 11220
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.297560214996338,
+      "learning_rate": 4.490106582362765e-05,
+      "loss": 1.6313,
+      "step": 11230
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.6740803718566895,
+      "learning_rate": 4.489143923278445e-05,
+      "loss": 1.4474,
+      "step": 11240
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.918575286865234,
+      "learning_rate": 4.488180459720074e-05,
+      "loss": 1.5736,
+      "step": 11250
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.530802249908447,
+      "learning_rate": 4.4872161920773075e-05,
+      "loss": 1.683,
+      "step": 11260
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.266594409942627,
+      "learning_rate": 4.486251120740127e-05,
+      "loss": 1.6557,
+      "step": 11270
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.779808521270752,
+      "learning_rate": 4.485285246098837e-05,
+      "loss": 1.9857,
+      "step": 11280
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.6034088134765625,
+      "learning_rate": 4.48431856854407e-05,
+      "loss": 1.6863,
+      "step": 11290
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.2785186767578125,
+      "learning_rate": 4.4833510884667805e-05,
+      "loss": 1.6183,
+      "step": 11300
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.0368452072143555,
+      "learning_rate": 4.482382806258249e-05,
+      "loss": 1.8994,
+      "step": 11310
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.603665351867676,
+      "learning_rate": 4.481413722310082e-05,
+      "loss": 1.6505,
+      "step": 11320
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.930195331573486,
+      "learning_rate": 4.480443837014205e-05,
+      "loss": 1.7675,
+      "step": 11330
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.467337608337402,
+      "learning_rate": 4.4794731507628734e-05,
+      "loss": 1.6801,
+      "step": 11340
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 6.546854496002197,
+      "learning_rate": 4.4785016639486635e-05,
+      "loss": 1.7569,
+      "step": 11350
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.412653923034668,
+      "learning_rate": 4.477529376964475e-05,
+      "loss": 1.824,
+      "step": 11360
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.92755126953125,
+      "learning_rate": 4.476556290203533e-05,
+      "loss": 1.6851,
+      "step": 11370
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.570308208465576,
+      "learning_rate": 4.475582404059385e-05,
+      "loss": 1.7238,
+      "step": 11380
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.924283027648926,
+      "learning_rate": 4.474607718925903e-05,
+      "loss": 1.7165,
+      "step": 11390
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.563042640686035,
+      "learning_rate": 4.473632235197279e-05,
+      "loss": 1.4755,
+      "step": 11400
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 6.890353679656982,
+      "learning_rate": 4.4726559532680326e-05,
+      "loss": 1.7279,
+      "step": 11410
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.909914493560791,
+      "learning_rate": 4.4716788735330016e-05,
+      "loss": 1.6262,
+      "step": 11420
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.578766345977783,
+      "learning_rate": 4.47070099638735e-05,
+      "loss": 1.3737,
+      "step": 11430
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 10.624035835266113,
+      "learning_rate": 4.4697223222265625e-05,
+      "loss": 1.8623,
+      "step": 11440
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.210607051849365,
+      "learning_rate": 4.4687428514464466e-05,
+      "loss": 1.5453,
+      "step": 11450
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.816832065582275,
+      "learning_rate": 4.467762584443131e-05,
+      "loss": 1.6239,
+      "step": 11460
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 6.2711405754089355,
+      "learning_rate": 4.4667815216130696e-05,
+      "loss": 1.7399,
+      "step": 11470
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 6.340164661407471,
+      "learning_rate": 4.465799663353034e-05,
+      "loss": 1.8794,
+      "step": 11480
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 6.173604965209961,
+      "learning_rate": 4.46481701006012e-05,
+      "loss": 1.7089,
+      "step": 11490
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.420295238494873,
+      "learning_rate": 4.463833562131743e-05,
+      "loss": 1.4616,
+      "step": 11500
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 1.3871608972549438,
+      "eval_runtime": 423.5,
+      "eval_samples_per_second": 50.109,
+      "eval_steps_per_second": 0.392,
+      "step": 11500
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 6.479365348815918,
+      "learning_rate": 4.462849319965643e-05,
+      "loss": 1.6149,
+      "step": 11510
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 7.298757553100586,
+      "learning_rate": 4.461864283959878e-05,
+      "loss": 1.7906,
+      "step": 11520
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.9603776931762695,
+      "learning_rate": 4.46087845451283e-05,
+      "loss": 1.7409,
+      "step": 11530
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 6.1247758865356445,
+      "learning_rate": 4.459891832023199e-05,
+      "loss": 1.5607,
+      "step": 11540
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 6.38535737991333,
+      "learning_rate": 4.458904416890006e-05,
+      "loss": 1.5341,
+      "step": 11550
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.868686676025391,
+      "learning_rate": 4.4579162095125965e-05,
+      "loss": 1.7445,
+      "step": 11560
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.29506778717041,
+      "learning_rate": 4.4569272102906314e-05,
+      "loss": 1.5512,
+      "step": 11570
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.732799530029297,
+      "learning_rate": 4.455937419624093e-05,
+      "loss": 1.7647,
+      "step": 11580
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.685189723968506,
+      "learning_rate": 4.454946837913287e-05,
+      "loss": 1.7306,
+      "step": 11590
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 6.369485855102539,
+      "learning_rate": 4.453955465558837e-05,
+      "loss": 1.7788,
+      "step": 11600
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.7086992263793945,
+      "learning_rate": 4.4529633029616824e-05,
+      "loss": 1.673,
+      "step": 11610
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.129878997802734,
+      "learning_rate": 4.451970350523089e-05,
+      "loss": 1.7806,
+      "step": 11620
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.230447292327881,
+      "learning_rate": 4.450976608644637e-05,
+      "loss": 1.7927,
+      "step": 11630
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.560041904449463,
+      "learning_rate": 4.4499820777282296e-05,
+      "loss": 1.7338,
+      "step": 11640
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.671140670776367,
+      "learning_rate": 4.4489867581760855e-05,
+      "loss": 1.5663,
+      "step": 11650
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.239694595336914,
+      "learning_rate": 4.447990650390745e-05,
+      "loss": 1.6173,
+      "step": 11660
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.434264659881592,
+      "learning_rate": 4.446993754775066e-05,
+      "loss": 1.4899,
+      "step": 11670
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 6.034209251403809,
+      "learning_rate": 4.445996071732226e-05,
+      "loss": 1.818,
+      "step": 11680
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 6.072056770324707,
+      "learning_rate": 4.4449976016657185e-05,
+      "loss": 1.843,
+      "step": 11690
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.8844799995422363,
+      "learning_rate": 4.4439983449793585e-05,
+      "loss": 1.6013,
+      "step": 11700
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.102245807647705,
+      "learning_rate": 4.442998302077277e-05,
+      "loss": 1.5863,
+      "step": 11710
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.803674697875977,
+      "learning_rate": 4.4419974733639244e-05,
+      "loss": 1.6002,
+      "step": 11720
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.105443954467773,
+      "learning_rate": 4.440995859244067e-05,
+      "loss": 1.5296,
+      "step": 11730
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.575257301330566,
+      "learning_rate": 4.439993460122791e-05,
+      "loss": 1.6627,
+      "step": 11740
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.023557662963867,
+      "learning_rate": 4.438990276405499e-05,
+      "loss": 1.4708,
+      "step": 11750
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.555371284484863,
+      "learning_rate": 4.43798630849791e-05,
+      "loss": 1.4226,
+      "step": 11760
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.876458644866943,
+      "learning_rate": 4.436981556806061e-05,
+      "loss": 1.9134,
+      "step": 11770
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.92462158203125,
+      "learning_rate": 4.435976021736307e-05,
+      "loss": 1.5589,
+      "step": 11780
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.126759052276611,
+      "learning_rate": 4.4349697036953186e-05,
+      "loss": 1.5,
+      "step": 11790
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.3090500831604,
+      "learning_rate": 4.433962603090083e-05,
+      "loss": 1.7171,
+      "step": 11800
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.329404830932617,
+      "learning_rate": 4.432954720327904e-05,
+      "loss": 1.5186,
+      "step": 11810
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.43467378616333,
+      "learning_rate": 4.431946055816403e-05,
+      "loss": 1.6543,
+      "step": 11820
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.861675262451172,
+      "learning_rate": 4.430936609963515e-05,
+      "loss": 1.7306,
+      "step": 11830
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.122840404510498,
+      "learning_rate": 4.429926383177494e-05,
+      "loss": 1.4537,
+      "step": 11840
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 7.410120964050293,
+      "learning_rate": 4.4289153758669075e-05,
+      "loss": 1.7531,
+      "step": 11850
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.340906620025635,
+      "learning_rate": 4.4279035884406395e-05,
+      "loss": 1.7773,
+      "step": 11860
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.850555419921875,
+      "learning_rate": 4.4268910213078896e-05,
+      "loss": 1.6794,
+      "step": 11870
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.56842041015625,
+      "learning_rate": 4.4258776748781725e-05,
+      "loss": 1.5125,
+      "step": 11880
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.323385715484619,
+      "learning_rate": 4.424863549561319e-05,
+      "loss": 1.6437,
+      "step": 11890
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.173916339874268,
+      "learning_rate": 4.423848645767473e-05,
+      "loss": 1.5748,
+      "step": 11900
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.211900234222412,
+      "learning_rate": 4.4228329639070954e-05,
+      "loss": 1.5492,
+      "step": 11910
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.843668460845947,
+      "learning_rate": 4.42181650439096e-05,
+      "loss": 1.4393,
+      "step": 11920
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 7.61888313293457,
+      "learning_rate": 4.4207992676301563e-05,
+      "loss": 1.6835,
+      "step": 11930
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.9813103675842285,
+      "learning_rate": 4.419781254036088e-05,
+      "loss": 1.3988,
+      "step": 11940
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.993699073791504,
+      "learning_rate": 4.4187624640204716e-05,
+      "loss": 1.6954,
+      "step": 11950
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.204382419586182,
+      "learning_rate": 4.4177428979953394e-05,
+      "loss": 1.5512,
+      "step": 11960
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 7.01641321182251,
+      "learning_rate": 4.416722556373037e-05,
+      "loss": 1.5539,
+      "step": 11970
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.7084059715271,
+      "learning_rate": 4.415701439566223e-05,
+      "loss": 1.7911,
+      "step": 11980
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 7.116555213928223,
+      "learning_rate": 4.4146795479878705e-05,
+      "loss": 1.6712,
+      "step": 11990
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.768601417541504,
+      "learning_rate": 4.413656882051266e-05,
+      "loss": 1.5553,
+      "step": 12000
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 1.352138876914978,
+      "eval_runtime": 423.4085,
+      "eval_samples_per_second": 50.119,
+      "eval_steps_per_second": 0.392,
+      "step": 12000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.654323577880859,
+      "learning_rate": 4.412633442170007e-05,
+      "loss": 1.4904,
+      "step": 12010
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.876643180847168,
+      "learning_rate": 4.411609228758007e-05,
+      "loss": 1.78,
+      "step": 12020
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.229610443115234,
+      "learning_rate": 4.41058424222949e-05,
+      "loss": 1.4487,
+      "step": 12030
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.311570167541504,
+      "learning_rate": 4.409558482998994e-05,
+      "loss": 1.6014,
+      "step": 12040
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.150774955749512,
+      "learning_rate": 4.40853195148137e-05,
+      "loss": 1.5214,
+      "step": 12050
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.989500522613525,
+      "learning_rate": 4.407504648091779e-05,
+      "loss": 1.6158,
+      "step": 12060
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.883471488952637,
+      "learning_rate": 4.406476573245697e-05,
+      "loss": 1.5151,
+      "step": 12070
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.21392822265625,
+      "learning_rate": 4.4054477273589115e-05,
+      "loss": 1.6264,
+      "step": 12080
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 7.5838236808776855,
+      "learning_rate": 4.404418110847518e-05,
+      "loss": 1.7264,
+      "step": 12090
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.685169696807861,
+      "learning_rate": 4.403387724127929e-05,
+      "loss": 1.6152,
+      "step": 12100
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.65505313873291,
+      "learning_rate": 4.4023565676168655e-05,
+      "loss": 1.6846,
+      "step": 12110
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.558564186096191,
+      "learning_rate": 4.4013246417313604e-05,
+      "loss": 1.7264,
+      "step": 12120
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.605981349945068,
+      "learning_rate": 4.400291946888758e-05,
+      "loss": 1.5809,
+      "step": 12130
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.0989670753479,
+      "learning_rate": 4.399258483506714e-05,
+      "loss": 1.2892,
+      "step": 12140
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.56234884262085,
+      "learning_rate": 4.398224252003193e-05,
+      "loss": 1.4583,
+      "step": 12150
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.174875736236572,
+      "learning_rate": 4.397189252796473e-05,
+      "loss": 1.7491,
+      "step": 12160
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.127572536468506,
+      "learning_rate": 4.39615348630514e-05,
+      "loss": 1.5392,
+      "step": 12170
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.131232261657715,
+      "learning_rate": 4.395116952948093e-05,
+      "loss": 1.4782,
+      "step": 12180
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.0297980308532715,
+      "learning_rate": 4.394079653144538e-05,
+      "loss": 1.6276,
+      "step": 12190
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.747319221496582,
+      "learning_rate": 4.393041587313993e-05,
+      "loss": 1.3473,
+      "step": 12200
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.469211101531982,
+      "learning_rate": 4.3920027558762855e-05,
+      "loss": 1.7429,
+      "step": 12210
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.476499080657959,
+      "learning_rate": 4.390963159251552e-05,
+      "loss": 1.7891,
+      "step": 12220
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.6036200523376465,
+      "learning_rate": 4.38992279786024e-05,
+      "loss": 1.5147,
+      "step": 12230
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 6.28443717956543,
+      "learning_rate": 4.388881672123105e-05,
+      "loss": 1.7051,
+      "step": 12240
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.224653244018555,
+      "learning_rate": 4.38783978246121e-05,
+      "loss": 1.445,
+      "step": 12250
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.014994144439697,
+      "learning_rate": 4.386797129295932e-05,
+      "loss": 1.6525,
+      "step": 12260
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.2429118156433105,
+      "learning_rate": 4.385753713048951e-05,
+      "loss": 1.5287,
+      "step": 12270
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.849613189697266,
+      "learning_rate": 4.38470953414226e-05,
+      "loss": 1.497,
+      "step": 12280
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.8472418785095215,
+      "learning_rate": 4.383664592998158e-05,
+      "loss": 1.6662,
+      "step": 12290
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.683506965637207,
+      "learning_rate": 4.382618890039252e-05,
+      "loss": 1.5714,
+      "step": 12300
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.4897780418396,
+      "learning_rate": 4.381572425688459e-05,
+      "loss": 1.4527,
+      "step": 12310
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.370377063751221,
+      "learning_rate": 4.380525200369003e-05,
+      "loss": 1.5004,
+      "step": 12320
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.105432510375977,
+      "learning_rate": 4.379477214504415e-05,
+      "loss": 1.554,
+      "step": 12330
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.320200443267822,
+      "learning_rate": 4.3784284685185364e-05,
+      "loss": 1.6465,
+      "step": 12340
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.400229454040527,
+      "learning_rate": 4.377378962835511e-05,
+      "loss": 1.4433,
+      "step": 12350
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.23899507522583,
+      "learning_rate": 4.376328697879796e-05,
+      "loss": 1.5693,
+      "step": 12360
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.675478935241699,
+      "learning_rate": 4.375277674076149e-05,
+      "loss": 1.3086,
+      "step": 12370
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.473839282989502,
+      "learning_rate": 4.374225891849641e-05,
+      "loss": 1.6332,
+      "step": 12380
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 7.072377681732178,
+      "learning_rate": 4.373173351625645e-05,
+      "loss": 1.6003,
+      "step": 12390
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.378538131713867,
+      "learning_rate": 4.372120053829842e-05,
+      "loss": 1.7182,
+      "step": 12400
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.857398986816406,
+      "learning_rate": 4.371065998888221e-05,
+      "loss": 1.4698,
+      "step": 12410
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.611136436462402,
+      "learning_rate": 4.370011187227076e-05,
+      "loss": 1.473,
+      "step": 12420
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.341707706451416,
+      "learning_rate": 4.3689556192730044e-05,
+      "loss": 1.8078,
+      "step": 12430
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.309572696685791,
+      "learning_rate": 4.3678992954529144e-05,
+      "loss": 1.5006,
+      "step": 12440
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 7.722212791442871,
+      "learning_rate": 4.366842216194017e-05,
+      "loss": 1.7005,
+      "step": 12450
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.579885482788086,
+      "learning_rate": 4.3657843819238284e-05,
+      "loss": 1.3846,
+      "step": 12460
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 6.808528423309326,
+      "learning_rate": 4.3647257930701724e-05,
+      "loss": 1.6517,
+      "step": 12470
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.950486183166504,
+      "learning_rate": 4.3636664500611756e-05,
+      "loss": 1.4428,
+      "step": 12480
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.783468723297119,
+      "learning_rate": 4.362606353325271e-05,
+      "loss": 1.698,
+      "step": 12490
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 6.811148643493652,
+      "learning_rate": 4.3615455032911946e-05,
+      "loss": 1.8498,
+      "step": 12500
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 1.3111711740493774,
+      "eval_runtime": 423.3291,
+      "eval_samples_per_second": 50.129,
+      "eval_steps_per_second": 0.392,
+      "step": 12500
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.396823883056641,
+      "learning_rate": 4.360483900387989e-05,
+      "loss": 1.4509,
+      "step": 12510
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.437137603759766,
+      "learning_rate": 4.359421545045002e-05,
+      "loss": 1.6226,
+      "step": 12520
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 6.149616718292236,
+      "learning_rate": 4.358358437691882e-05,
+      "loss": 1.4159,
+      "step": 12530
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.7397027015686035,
+      "learning_rate": 4.357294578758587e-05,
+      "loss": 1.2936,
+      "step": 12540
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.202317237854004,
+      "learning_rate": 4.356229968675374e-05,
+      "loss": 1.4471,
+      "step": 12550
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 6.386228561401367,
+      "learning_rate": 4.355164607872806e-05,
+      "loss": 1.6917,
+      "step": 12560
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.868130683898926,
+      "learning_rate": 4.3540984967817486e-05,
+      "loss": 1.558,
+      "step": 12570
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.320023059844971,
+      "learning_rate": 4.3530316358333734e-05,
+      "loss": 1.4657,
+      "step": 12580
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.57085657119751,
+      "learning_rate": 4.351964025459151e-05,
+      "loss": 1.4878,
+      "step": 12590
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 6.250162601470947,
+      "learning_rate": 4.35089566609086e-05,
+      "loss": 1.661,
+      "step": 12600
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.138901233673096,
+      "learning_rate": 4.349826558160578e-05,
+      "loss": 1.5959,
+      "step": 12610
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.0128068923950195,
+      "learning_rate": 4.348756702100687e-05,
+      "loss": 1.5687,
+      "step": 12620
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.661235332489014,
+      "learning_rate": 4.3476860983438714e-05,
+      "loss": 1.764,
+      "step": 12630
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.733764171600342,
+      "learning_rate": 4.3466147473231184e-05,
+      "loss": 1.2875,
+      "step": 12640
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.74668025970459,
+      "learning_rate": 4.3455426494717164e-05,
+      "loss": 1.5642,
+      "step": 12650
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 6.375096797943115,
+      "learning_rate": 4.344469805223256e-05,
+      "loss": 1.6291,
+      "step": 12660
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 6.215402603149414,
+      "learning_rate": 4.34339621501163e-05,
+      "loss": 1.6082,
+      "step": 12670
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 6.86801290512085,
+      "learning_rate": 4.342321879271035e-05,
+      "loss": 1.4994,
+      "step": 12680
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 6.0276641845703125,
+      "learning_rate": 4.341246798435965e-05,
+      "loss": 1.6687,
+      "step": 12690
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.529504776000977,
+      "learning_rate": 4.340170972941218e-05,
+      "loss": 1.6767,
+      "step": 12700
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 9.499716758728027,
+      "learning_rate": 4.339094403221893e-05,
+      "loss": 1.6431,
+      "step": 12710
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.8651652336120605,
+      "learning_rate": 4.338017089713389e-05,
+      "loss": 1.5028,
+      "step": 12720
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 6.01951265335083,
+      "learning_rate": 4.3369390328514056e-05,
+      "loss": 1.4069,
+      "step": 12730
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.871748447418213,
+      "learning_rate": 4.335860233071946e-05,
+      "loss": 1.392,
+      "step": 12740
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.011544704437256,
+      "learning_rate": 4.33478069081131e-05,
+      "loss": 1.6994,
+      "step": 12750
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 6.346103191375732,
+      "learning_rate": 4.3337004065061e-05,
+      "loss": 1.6418,
+      "step": 12760
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.917518615722656,
+      "learning_rate": 4.332619380593218e-05,
+      "loss": 1.3363,
+      "step": 12770
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 6.4693450927734375,
+      "learning_rate": 4.331537613509864e-05,
+      "loss": 1.6764,
+      "step": 12780
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.622341632843018,
+      "learning_rate": 4.3304551056935425e-05,
+      "loss": 1.4579,
+      "step": 12790
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 6.959853172302246,
+      "learning_rate": 4.3293718575820524e-05,
+      "loss": 1.5953,
+      "step": 12800
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.779293060302734,
+      "learning_rate": 4.3282878696134954e-05,
+      "loss": 1.7715,
+      "step": 12810
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.7977519035339355,
+      "learning_rate": 4.32720314222627e-05,
+      "loss": 1.6007,
+      "step": 12820
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.147169589996338,
+      "learning_rate": 4.3261176758590774e-05,
+      "loss": 1.3815,
+      "step": 12830
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.628530979156494,
+      "learning_rate": 4.325031470950912e-05,
+      "loss": 1.5828,
+      "step": 12840
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.895701885223389,
+      "learning_rate": 4.3239445279410726e-05,
+      "loss": 1.4952,
+      "step": 12850
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 7.570559978485107,
+      "learning_rate": 4.3228568472691534e-05,
+      "loss": 1.7566,
+      "step": 12860
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 6.089484691619873,
+      "learning_rate": 4.321768429375047e-05,
+      "loss": 1.4786,
+      "step": 12870
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.187694549560547,
+      "learning_rate": 4.320679274698946e-05,
+      "loss": 1.4087,
+      "step": 12880
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.053955554962158,
+      "learning_rate": 4.319589383681338e-05,
+      "loss": 1.3166,
+      "step": 12890
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.116620063781738,
+      "learning_rate": 4.3184987567630116e-05,
+      "loss": 1.468,
+      "step": 12900
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.72169828414917,
+      "learning_rate": 4.3174073943850504e-05,
+      "loss": 1.4468,
+      "step": 12910
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.469881534576416,
+      "learning_rate": 4.3163152969888384e-05,
+      "loss": 1.6363,
+      "step": 12920
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.1773552894592285,
+      "learning_rate": 4.315222465016053e-05,
+      "loss": 1.5785,
+      "step": 12930
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.658343315124512,
+      "learning_rate": 4.314128898908672e-05,
+      "loss": 1.5201,
+      "step": 12940
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.179915904998779,
+      "learning_rate": 4.313034599108968e-05,
+      "loss": 1.4282,
+      "step": 12950
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 7.206972599029541,
+      "learning_rate": 4.3119395660595126e-05,
+      "loss": 1.5281,
+      "step": 12960
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 6.760249614715576,
+      "learning_rate": 4.3108438002031706e-05,
+      "loss": 1.4842,
+      "step": 12970
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.5984978675842285,
+      "learning_rate": 4.309747301983107e-05,
+      "loss": 1.4941,
+      "step": 12980
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.022587776184082,
+      "learning_rate": 4.30865007184278e-05,
+      "loss": 1.5531,
+      "step": 12990
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.920048236846924,
+      "learning_rate": 4.307552110225945e-05,
+      "loss": 1.5089,
+      "step": 13000
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 1.2926665544509888,
+      "eval_runtime": 423.3962,
+      "eval_samples_per_second": 50.121,
+      "eval_steps_per_second": 0.392,
+      "step": 13000
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.030763149261475,
+      "learning_rate": 4.3064534175766533e-05,
+      "loss": 1.6161,
+      "step": 13010
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 6.073363780975342,
+      "learning_rate": 4.305353994339252e-05,
+      "loss": 1.4059,
+      "step": 13020
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.027052879333496,
+      "learning_rate": 4.304253840958382e-05,
+      "loss": 1.6348,
+      "step": 13030
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.7227044105529785,
+      "learning_rate": 4.303152957878983e-05,
+      "loss": 1.6145,
+      "step": 13040
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.840976238250732,
+      "learning_rate": 4.302051345546286e-05,
+      "loss": 1.3041,
+      "step": 13050
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.210689544677734,
+      "learning_rate": 4.300949004405819e-05,
+      "loss": 1.4798,
+      "step": 13060
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.268500804901123,
+      "learning_rate": 4.299845934903405e-05,
+      "loss": 1.8248,
+      "step": 13070
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 6.304892539978027,
+      "learning_rate": 4.2987421374851604e-05,
+      "loss": 1.6951,
+      "step": 13080
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.317906856536865,
+      "learning_rate": 4.297637612597497e-05,
+      "loss": 1.4274,
+      "step": 13090
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.343708038330078,
+      "learning_rate": 4.296532360687119e-05,
+      "loss": 1.7517,
+      "step": 13100
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.536047458648682,
+      "learning_rate": 4.2954263822010274e-05,
+      "loss": 1.442,
+      "step": 13110
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.51957893371582,
+      "learning_rate": 4.294319677586516e-05,
+      "loss": 1.4277,
+      "step": 13120
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.247564315795898,
+      "learning_rate": 4.29321224729117e-05,
+      "loss": 1.4482,
+      "step": 13130
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.230405330657959,
+      "learning_rate": 4.2921040917628706e-05,
+      "loss": 1.446,
+      "step": 13140
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.593940258026123,
+      "learning_rate": 4.2909952114497925e-05,
+      "loss": 1.5358,
+      "step": 13150
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 6.715129375457764,
+      "learning_rate": 4.289885606800402e-05,
+      "loss": 1.5706,
+      "step": 13160
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.610743045806885,
+      "learning_rate": 4.2887752782634606e-05,
+      "loss": 1.3979,
+      "step": 13170
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.410665988922119,
+      "learning_rate": 4.2876642262880196e-05,
+      "loss": 1.4915,
+      "step": 13180
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 6.392048358917236,
+      "learning_rate": 4.2865524513234234e-05,
+      "loss": 1.5759,
+      "step": 13190
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 7.24970006942749,
+      "learning_rate": 4.285439953819311e-05,
+      "loss": 1.5208,
+      "step": 13200
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 6.112131595611572,
+      "learning_rate": 4.284326734225613e-05,
+      "loss": 1.4937,
+      "step": 13210
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.898817539215088,
+      "learning_rate": 4.28321279299255e-05,
+      "loss": 1.4515,
+      "step": 13220
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 6.257785320281982,
+      "learning_rate": 4.2820981305706364e-05,
+      "loss": 1.6071,
+      "step": 13230
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.535906791687012,
+      "learning_rate": 4.2809827474106775e-05,
+      "loss": 1.4904,
+      "step": 13240
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.786548614501953,
+      "learning_rate": 4.279866643963771e-05,
+      "loss": 1.634,
+      "step": 13250
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 6.041201114654541,
+      "learning_rate": 4.278749820681304e-05,
+      "loss": 1.6218,
+      "step": 13260
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.758951663970947,
+      "learning_rate": 4.2776322780149565e-05,
+      "loss": 1.4682,
+      "step": 13270
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.944740295410156,
+      "learning_rate": 4.2765140164166984e-05,
+      "loss": 1.4629,
+      "step": 13280
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.891012191772461,
+      "learning_rate": 4.2753950363387916e-05,
+      "loss": 1.2639,
+      "step": 13290
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.920727729797363,
+      "learning_rate": 4.274275338233788e-05,
+      "loss": 1.5577,
+      "step": 13300
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.73730993270874,
+      "learning_rate": 4.273154922554528e-05,
+      "loss": 1.3221,
+      "step": 13310
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.401158332824707,
+      "learning_rate": 4.272033789754146e-05,
+      "loss": 1.3241,
+      "step": 13320
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.525073051452637,
+      "learning_rate": 4.2709119402860626e-05,
+      "loss": 1.5812,
+      "step": 13330
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.745188236236572,
+      "learning_rate": 4.2697893746039905e-05,
+      "loss": 1.4632,
+      "step": 13340
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 6.591483116149902,
+      "learning_rate": 4.268666093161933e-05,
+      "loss": 1.4419,
+      "step": 13350
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.667323589324951,
+      "learning_rate": 4.267542096414179e-05,
+      "loss": 1.6322,
+      "step": 13360
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.23670768737793,
+      "learning_rate": 4.26641738481531e-05,
+      "loss": 1.313,
+      "step": 13370
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.662097930908203,
+      "learning_rate": 4.265291958820196e-05,
+      "loss": 1.4252,
+      "step": 13380
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 6.018895149230957,
+      "learning_rate": 4.264165818883996e-05,
+      "loss": 1.6186,
+      "step": 13390
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.775135040283203,
+      "learning_rate": 4.263038965462156e-05,
+      "loss": 1.3846,
+      "step": 13400
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.29066276550293,
+      "learning_rate": 4.261911399010413e-05,
+      "loss": 1.3985,
+      "step": 13410
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 6.596442699432373,
+      "learning_rate": 4.260783119984792e-05,
+      "loss": 1.4038,
+      "step": 13420
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 6.022501468658447,
+      "learning_rate": 4.2596541288416037e-05,
+      "loss": 1.6066,
+      "step": 13430
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.6616106033325195,
+      "learning_rate": 4.258524426037449e-05,
+      "loss": 1.3102,
+      "step": 13440
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.347101211547852,
+      "learning_rate": 4.2573940120292175e-05,
+      "loss": 1.1036,
+      "step": 13450
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.300200939178467,
+      "learning_rate": 4.2562628872740846e-05,
+      "loss": 1.3468,
+      "step": 13460
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.665987014770508,
+      "learning_rate": 4.255131052229512e-05,
+      "loss": 1.4704,
+      "step": 13470
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.978814125061035,
+      "learning_rate": 4.253998507353253e-05,
+      "loss": 1.595,
+      "step": 13480
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 6.704648017883301,
+      "learning_rate": 4.252865253103343e-05,
+      "loss": 1.5977,
+      "step": 13490
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.576136112213135,
+      "learning_rate": 4.251731289938109e-05,
+      "loss": 1.4312,
+      "step": 13500
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 1.2380741834640503,
+      "eval_runtime": 423.2033,
+      "eval_samples_per_second": 50.144,
+      "eval_steps_per_second": 0.392,
+      "step": 13500
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.1742095947265625,
+      "learning_rate": 4.2505966183161594e-05,
+      "loss": 1.3508,
+      "step": 13510
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.386550426483154,
+      "learning_rate": 4.249461238696395e-05,
+      "loss": 1.4714,
+      "step": 13520
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 6.069825649261475,
+      "learning_rate": 4.248325151537998e-05,
+      "loss": 1.6404,
+      "step": 13530
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.645625114440918,
+      "learning_rate": 4.247188357300439e-05,
+      "loss": 1.6421,
+      "step": 13540
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.380563259124756,
+      "learning_rate": 4.2460508564434756e-05,
+      "loss": 1.4504,
+      "step": 13550
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.783507823944092,
+      "learning_rate": 4.2449126494271485e-05,
+      "loss": 1.32,
+      "step": 13560
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.294303894042969,
+      "learning_rate": 4.243773736711786e-05,
+      "loss": 1.1827,
+      "step": 13570
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.025012969970703,
+      "learning_rate": 4.242634118758001e-05,
+      "loss": 1.2954,
+      "step": 13580
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.992012023925781,
+      "learning_rate": 4.2414937960266914e-05,
+      "loss": 1.4897,
+      "step": 13590
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 6.171781539916992,
+      "learning_rate": 4.240352768979041e-05,
+      "loss": 1.5298,
+      "step": 13600
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 6.134835720062256,
+      "learning_rate": 4.239211038076519e-05,
+      "loss": 1.2998,
+      "step": 13610
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 7.044719219207764,
+      "learning_rate": 4.238068603780876e-05,
+      "loss": 1.6675,
+      "step": 13620
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.010741233825684,
+      "learning_rate": 4.236925466554151e-05,
+      "loss": 1.4997,
+      "step": 13630
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 7.9801716804504395,
+      "learning_rate": 4.235781626858666e-05,
+      "loss": 1.524,
+      "step": 13640
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.047695159912109,
+      "learning_rate": 4.234637085157025e-05,
+      "loss": 1.5436,
+      "step": 13650
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.539044380187988,
+      "learning_rate": 4.233491841912118e-05,
+      "loss": 1.403,
+      "step": 13660
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 7.978495121002197,
+      "learning_rate": 4.23234589758712e-05,
+      "loss": 1.6048,
+      "step": 13670
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.434673309326172,
+      "learning_rate": 4.231199252645487e-05,
+      "loss": 1.4996,
+      "step": 13680
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.410276889801025,
+      "learning_rate": 4.2300519075509583e-05,
+      "loss": 1.699,
+      "step": 13690
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 6.65217399597168,
+      "learning_rate": 4.228903862767558e-05,
+      "loss": 1.5541,
+      "step": 13700
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 6.513009548187256,
+      "learning_rate": 4.227755118759593e-05,
+      "loss": 1.3758,
+      "step": 13710
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.00607967376709,
+      "learning_rate": 4.2266056759916515e-05,
+      "loss": 1.5366,
+      "step": 13720
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.527690887451172,
+      "learning_rate": 4.225455534928607e-05,
+      "loss": 1.535,
+      "step": 13730
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.513370990753174,
+      "learning_rate": 4.224304696035611e-05,
+      "loss": 1.3079,
+      "step": 13740
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 6.4584197998046875,
+      "learning_rate": 4.223153159778103e-05,
+      "loss": 1.3724,
+      "step": 13750
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.3216352462768555,
+      "learning_rate": 4.222000926621799e-05,
+      "loss": 1.3896,
+      "step": 13760
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.951451778411865,
+      "learning_rate": 4.2208479970327e-05,
+      "loss": 1.2385,
+      "step": 13770
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 7.510885238647461,
+      "learning_rate": 4.219694371477088e-05,
+      "loss": 1.6211,
+      "step": 13780
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 6.100176811218262,
+      "learning_rate": 4.2185400504215265e-05,
+      "loss": 1.5985,
+      "step": 13790
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.841367721557617,
+      "learning_rate": 4.217385034332861e-05,
+      "loss": 1.3501,
+      "step": 13800
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.256452560424805,
+      "learning_rate": 4.216229323678217e-05,
+      "loss": 1.4354,
+      "step": 13810
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 6.605254650115967,
+      "learning_rate": 4.2150729189250005e-05,
+      "loss": 1.3359,
+      "step": 13820
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.197912693023682,
+      "learning_rate": 4.213915820540898e-05,
+      "loss": 1.3429,
+      "step": 13830
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 7.422953128814697,
+      "learning_rate": 4.212758028993881e-05,
+      "loss": 1.2199,
+      "step": 13840
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.43693733215332,
+      "learning_rate": 4.211599544752194e-05,
+      "loss": 1.465,
+      "step": 13850
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.025950908660889,
+      "learning_rate": 4.210440368284367e-05,
+      "loss": 1.245,
+      "step": 13860
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 7.20076847076416,
+      "learning_rate": 4.209280500059209e-05,
+      "loss": 1.3572,
+      "step": 13870
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.6596574783325195,
+      "learning_rate": 4.2081199405458074e-05,
+      "loss": 1.3353,
+      "step": 13880
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 6.5892157554626465,
+      "learning_rate": 4.2069586902135296e-05,
+      "loss": 1.5138,
+      "step": 13890
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 6.405339241027832,
+      "learning_rate": 4.205796749532024e-05,
+      "loss": 1.3152,
+      "step": 13900
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.329562187194824,
+      "learning_rate": 4.2046341189712154e-05,
+      "loss": 1.4369,
+      "step": 13910
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 6.811237812042236,
+      "learning_rate": 4.203470799001311e-05,
+      "loss": 1.598,
+      "step": 13920
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.798642158508301,
+      "learning_rate": 4.202306790092792e-05,
+      "loss": 1.419,
+      "step": 13930
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 6.391123294830322,
+      "learning_rate": 4.201142092716426e-05,
+      "loss": 1.5633,
+      "step": 13940
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.4787538051605225,
+      "learning_rate": 4.199976707343249e-05,
+      "loss": 1.3451,
+      "step": 13950
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.148783206939697,
+      "learning_rate": 4.198810634444584e-05,
+      "loss": 1.3242,
+      "step": 13960
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 6.200438022613525,
+      "learning_rate": 4.197643874492027e-05,
+      "loss": 1.5343,
+      "step": 13970
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.820129871368408,
+      "learning_rate": 4.1964764279574544e-05,
+      "loss": 1.4614,
+      "step": 13980
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 6.187035083770752,
+      "learning_rate": 4.195308295313018e-05,
+      "loss": 1.5567,
+      "step": 13990
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.901625633239746,
+      "learning_rate": 4.194139477031149e-05,
+      "loss": 1.4524,
+      "step": 14000
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 1.2039812803268433,
+      "eval_runtime": 422.6707,
+      "eval_samples_per_second": 50.207,
+      "eval_steps_per_second": 0.393,
+      "step": 14000
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.694302558898926,
+      "learning_rate": 4.192969973584556e-05,
+      "loss": 1.4586,
+      "step": 14010
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.519773960113525,
+      "learning_rate": 4.191799785446222e-05,
+      "loss": 1.3855,
+      "step": 14020
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.454830169677734,
+      "learning_rate": 4.1906289130894114e-05,
+      "loss": 1.3032,
+      "step": 14030
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 6.581485271453857,
+      "learning_rate": 4.1894573569876614e-05,
+      "loss": 1.5035,
+      "step": 14040
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 6.112485885620117,
+      "learning_rate": 4.188285117614787e-05,
+      "loss": 1.3223,
+      "step": 14050
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.18966817855835,
+      "learning_rate": 4.18711219544488e-05,
+      "loss": 1.4248,
+      "step": 14060
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.011171340942383,
+      "learning_rate": 4.185938590952309e-05,
+      "loss": 1.3983,
+      "step": 14070
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.915748119354248,
+      "learning_rate": 4.184764304611715e-05,
+      "loss": 1.3545,
+      "step": 14080
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.680284023284912,
+      "learning_rate": 4.183589336898019e-05,
+      "loss": 1.4662,
+      "step": 14090
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.400657653808594,
+      "learning_rate": 4.1824136882864166e-05,
+      "loss": 1.4658,
+      "step": 14100
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 6.576321601867676,
+      "learning_rate": 4.181237359252377e-05,
+      "loss": 1.6105,
+      "step": 14110
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.652660846710205,
+      "learning_rate": 4.180060350271645e-05,
+      "loss": 1.1901,
+      "step": 14120
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.588864803314209,
+      "learning_rate": 4.178882661820242e-05,
+      "loss": 1.3388,
+      "step": 14130
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.945875883102417,
+      "learning_rate": 4.1777042943744647e-05,
+      "loss": 1.4253,
+      "step": 14140
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.304834842681885,
+      "learning_rate": 4.1765252484108806e-05,
+      "loss": 1.369,
+      "step": 14150
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.2396392822265625,
+      "learning_rate": 4.175345524406335e-05,
+      "loss": 1.5466,
+      "step": 14160
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 6.9206624031066895,
+      "learning_rate": 4.1741651228379456e-05,
+      "loss": 1.3143,
+      "step": 14170
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.08700704574585,
+      "learning_rate": 4.1729840441831065e-05,
+      "loss": 1.3677,
+      "step": 14180
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.640330791473389,
+      "learning_rate": 4.171802288919482e-05,
+      "loss": 1.3894,
+      "step": 14190
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 6.228893280029297,
+      "learning_rate": 4.1706198575250146e-05,
+      "loss": 1.3178,
+      "step": 14200
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 6.123324394226074,
+      "learning_rate": 4.1694367504779156e-05,
+      "loss": 1.3004,
+      "step": 14210
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.580305576324463,
+      "learning_rate": 4.168252968256672e-05,
+      "loss": 1.1406,
+      "step": 14220
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 6.379098892211914,
+      "learning_rate": 4.167068511340044e-05,
+      "loss": 1.3433,
+      "step": 14230
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.925345420837402,
+      "learning_rate": 4.165883380207064e-05,
+      "loss": 1.4756,
+      "step": 14240
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.81134033203125,
+      "learning_rate": 4.164697575337039e-05,
+      "loss": 1.4024,
+      "step": 14250
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.875208377838135,
+      "learning_rate": 4.1635110972095435e-05,
+      "loss": 1.3796,
+      "step": 14260
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.351586818695068,
+      "learning_rate": 4.162323946304431e-05,
+      "loss": 1.3654,
+      "step": 14270
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 7.102518558502197,
+      "learning_rate": 4.161136123101821e-05,
+      "loss": 1.6611,
+      "step": 14280
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.741572856903076,
+      "learning_rate": 4.159947628082109e-05,
+      "loss": 1.662,
+      "step": 14290
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.783633232116699,
+      "learning_rate": 4.1587584617259605e-05,
+      "loss": 1.3581,
+      "step": 14300
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 7.395619869232178,
+      "learning_rate": 4.157568624514313e-05,
+      "loss": 1.6291,
+      "step": 14310
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.274905204772949,
+      "learning_rate": 4.156378116928375e-05,
+      "loss": 1.4451,
+      "step": 14320
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.388856410980225,
+      "learning_rate": 4.155186939449625e-05,
+      "loss": 1.3535,
+      "step": 14330
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.824211120605469,
+      "learning_rate": 4.1539950925598155e-05,
+      "loss": 1.279,
+      "step": 14340
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.400022029876709,
+      "learning_rate": 4.152802576740966e-05,
+      "loss": 1.587,
+      "step": 14350
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.7362847328186035,
+      "learning_rate": 4.151609392475371e-05,
+      "loss": 1.6354,
+      "step": 14360
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 7.173040866851807,
+      "learning_rate": 4.15041554024559e-05,
+      "loss": 1.6131,
+      "step": 14370
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.964481353759766,
+      "learning_rate": 4.1492210205344574e-05,
+      "loss": 1.6113,
+      "step": 14380
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.03452205657959,
+      "learning_rate": 4.1480258338250744e-05,
+      "loss": 1.5132,
+      "step": 14390
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.337208271026611,
+      "learning_rate": 4.1468299806008134e-05,
+      "loss": 1.2467,
+      "step": 14400
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.453386306762695,
+      "learning_rate": 4.145633461345316e-05,
+      "loss": 1.4429,
+      "step": 14410
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.697633266448975,
+      "learning_rate": 4.1444362765424936e-05,
+      "loss": 1.3794,
+      "step": 14420
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.01288366317749,
+      "learning_rate": 4.1432384266765264e-05,
+      "loss": 1.2765,
+      "step": 14430
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.066435813903809,
+      "learning_rate": 4.142039912231864e-05,
+      "loss": 1.5255,
+      "step": 14440
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.142197608947754,
+      "learning_rate": 4.140840733693223e-05,
+      "loss": 1.4664,
+      "step": 14450
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.393848896026611,
+      "learning_rate": 4.139640891545591e-05,
+      "loss": 1.4885,
+      "step": 14460
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 15.337352752685547,
+      "learning_rate": 4.138440386274222e-05,
+      "loss": 1.4605,
+      "step": 14470
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.26066780090332,
+      "learning_rate": 4.13723921836464e-05,
+      "loss": 1.5111,
+      "step": 14480
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.657282829284668,
+      "learning_rate": 4.136037388302636e-05,
+      "loss": 1.3115,
+      "step": 14490
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 6.4021525382995605,
+      "learning_rate": 4.134834896574269e-05,
+      "loss": 1.423,
+      "step": 14500
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 1.2047580480575562,
+      "eval_runtime": 422.7989,
+      "eval_samples_per_second": 50.192,
+      "eval_steps_per_second": 0.393,
+      "step": 14500
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.063785552978516,
+      "learning_rate": 4.1336317436658636e-05,
+      "loss": 1.3393,
+      "step": 14510
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.269222259521484,
+      "learning_rate": 4.1324279300640165e-05,
+      "loss": 1.4269,
+      "step": 14520
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 7.072628021240234,
+      "learning_rate": 4.1312234562555875e-05,
+      "loss": 1.4443,
+      "step": 14530
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.179318428039551,
+      "learning_rate": 4.130018322727705e-05,
+      "loss": 1.3808,
+      "step": 14540
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.315336227416992,
+      "learning_rate": 4.128812529967763e-05,
+      "loss": 1.4635,
+      "step": 14550
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 6.5437541007995605,
+      "learning_rate": 4.127606078463424e-05,
+      "loss": 1.4484,
+      "step": 14560
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 8.814804077148438,
+      "learning_rate": 4.126398968702616e-05,
+      "loss": 1.3596,
+      "step": 14570
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 6.811517238616943,
+      "learning_rate": 4.1251912011735326e-05,
+      "loss": 1.4437,
+      "step": 14580
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.859598636627197,
+      "learning_rate": 4.123982776364635e-05,
+      "loss": 1.4392,
+      "step": 14590
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.631993293762207,
+      "learning_rate": 4.122773694764648e-05,
+      "loss": 1.4582,
+      "step": 14600
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.026217460632324,
+      "learning_rate": 4.1215639568625644e-05,
+      "loss": 1.1777,
+      "step": 14610
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.466883182525635,
+      "learning_rate": 4.12035356314764e-05,
+      "loss": 1.5073,
+      "step": 14620
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.44281005859375,
+      "learning_rate": 4.119142514109397e-05,
+      "loss": 1.3269,
+      "step": 14630
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 7.500271797180176,
+      "learning_rate": 4.117930810237625e-05,
+      "loss": 1.4027,
+      "step": 14640
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 6.872348785400391,
+      "learning_rate": 4.1167184520223734e-05,
+      "loss": 1.4361,
+      "step": 14650
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.404297828674316,
+      "learning_rate": 4.115505439953961e-05,
+      "loss": 1.3718,
+      "step": 14660
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.596623420715332,
+      "learning_rate": 4.1142917745229686e-05,
+      "loss": 1.3586,
+      "step": 14670
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.425731182098389,
+      "learning_rate": 4.113077456220241e-05,
+      "loss": 1.2302,
+      "step": 14680
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.91450834274292,
+      "learning_rate": 4.111862485536888e-05,
+      "loss": 1.2041,
+      "step": 14690
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 6.293668746948242,
+      "learning_rate": 4.110646862964284e-05,
+      "loss": 1.2934,
+      "step": 14700
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.863705158233643,
+      "learning_rate": 4.1094305889940646e-05,
+      "loss": 1.4319,
+      "step": 14710
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.59689474105835,
+      "learning_rate": 4.108213664118131e-05,
+      "loss": 1.2319,
+      "step": 14720
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.928666114807129,
+      "learning_rate": 4.1069960888286476e-05,
+      "loss": 1.2845,
+      "step": 14730
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 6.364851474761963,
+      "learning_rate": 4.10577786361804e-05,
+      "loss": 1.6162,
+      "step": 14740
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.940974235534668,
+      "learning_rate": 4.104558988978997e-05,
+      "loss": 1.3045,
+      "step": 14750
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.9831390380859375,
+      "learning_rate": 4.103339465404474e-05,
+      "loss": 1.3442,
+      "step": 14760
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.107690811157227,
+      "learning_rate": 4.102119293387683e-05,
+      "loss": 1.3325,
+      "step": 14770
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.088287830352783,
+      "learning_rate": 4.100898473422101e-05,
+      "loss": 1.3377,
+      "step": 14780
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.367318153381348,
+      "learning_rate": 4.099677006001469e-05,
+      "loss": 1.461,
+      "step": 14790
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.532168865203857,
+      "learning_rate": 4.098454891619787e-05,
+      "loss": 1.2492,
+      "step": 14800
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.846711158752441,
+      "learning_rate": 4.097232130771317e-05,
+      "loss": 1.281,
+      "step": 14810
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.030317783355713,
+      "learning_rate": 4.096008723950582e-05,
+      "loss": 1.4126,
+      "step": 14820
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 7.5767364501953125,
+      "learning_rate": 4.09478467165237e-05,
+      "loss": 1.445,
+      "step": 14830
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 7.347666263580322,
+      "learning_rate": 4.093559974371725e-05,
+      "loss": 1.283,
+      "step": 14840
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.339590549468994,
+      "learning_rate": 4.0923346326039543e-05,
+      "loss": 1.4574,
+      "step": 14850
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 6.707940101623535,
+      "learning_rate": 4.091108646844627e-05,
+      "loss": 1.3477,
+      "step": 14860
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.814685821533203,
+      "learning_rate": 4.0898820175895706e-05,
+      "loss": 1.2665,
+      "step": 14870
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 6.189144611358643,
+      "learning_rate": 4.0886547453348734e-05,
+      "loss": 1.4252,
+      "step": 14880
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.622840404510498,
+      "learning_rate": 4.0874268305768847e-05,
+      "loss": 1.4959,
+      "step": 14890
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.850718975067139,
+      "learning_rate": 4.0861982738122115e-05,
+      "loss": 1.2829,
+      "step": 14900
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.684026718139648,
+      "learning_rate": 4.0849690755377235e-05,
+      "loss": 1.4226,
+      "step": 14910
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 6.287352561950684,
+      "learning_rate": 4.0837392362505464e-05,
+      "loss": 1.4048,
+      "step": 14920
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.713840484619141,
+      "learning_rate": 4.0825087564480694e-05,
+      "loss": 1.1692,
+      "step": 14930
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 6.009079933166504,
+      "learning_rate": 4.081277636627936e-05,
+      "loss": 1.4785,
+      "step": 14940
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.710774898529053,
+      "learning_rate": 4.0800458772880514e-05,
+      "loss": 1.1754,
+      "step": 14950
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.261815071105957,
+      "learning_rate": 4.0788134789265784e-05,
+      "loss": 1.4076,
+      "step": 14960
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.314476490020752,
+      "learning_rate": 4.0775804420419404e-05,
+      "loss": 1.2346,
+      "step": 14970
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.6869306564331055,
+      "learning_rate": 4.076346767132816e-05,
+      "loss": 1.522,
+      "step": 14980
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.193082332611084,
+      "learning_rate": 4.0751124546981435e-05,
+      "loss": 1.4676,
+      "step": 14990
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.26650857925415,
+      "learning_rate": 4.073877505237118e-05,
+      "loss": 1.3972,
+      "step": 15000
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 1.1875303983688354,
+      "eval_runtime": 422.7473,
+      "eval_samples_per_second": 50.198,
+      "eval_steps_per_second": 0.393,
+      "step": 15000
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.977046489715576,
+      "learning_rate": 4.0726419192491946e-05,
+      "loss": 1.3338,
+      "step": 15010
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 6.557323455810547,
+      "learning_rate": 4.071405697234082e-05,
+      "loss": 1.2829,
+      "step": 15020
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 6.982044219970703,
+      "learning_rate": 4.070168839691751e-05,
+      "loss": 1.4146,
+      "step": 15030
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.319934368133545,
+      "learning_rate": 4.068931347122424e-05,
+      "loss": 1.4224,
+      "step": 15040
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.294476509094238,
+      "learning_rate": 4.0676932200265854e-05,
+      "loss": 1.3346,
+      "step": 15050
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.89024019241333,
+      "learning_rate": 4.066454458904973e-05,
+      "loss": 1.2246,
+      "step": 15060
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.022579193115234,
+      "learning_rate": 4.0652150642585804e-05,
+      "loss": 1.2931,
+      "step": 15070
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.805153846740723,
+      "learning_rate": 4.0639750365886595e-05,
+      "loss": 1.2412,
+      "step": 15080
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 6.5985236167907715,
+      "learning_rate": 4.0627343763967185e-05,
+      "loss": 1.3521,
+      "step": 15090
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 6.953591823577881,
+      "learning_rate": 4.06149308418452e-05,
+      "loss": 1.4378,
+      "step": 15100
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.721883773803711,
+      "learning_rate": 4.0602511604540824e-05,
+      "loss": 1.4803,
+      "step": 15110
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.27775764465332,
+      "learning_rate": 4.05900860570768e-05,
+      "loss": 1.3535,
+      "step": 15120
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 6.432738304138184,
+      "learning_rate": 4.057765420447842e-05,
+      "loss": 1.5568,
+      "step": 15130
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.384369850158691,
+      "learning_rate": 4.0565216051773526e-05,
+      "loss": 1.1974,
+      "step": 15140
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 6.226934909820557,
+      "learning_rate": 4.05527716039925e-05,
+      "loss": 1.7431,
+      "step": 15150
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 6.568231582641602,
+      "learning_rate": 4.05403208661683e-05,
+      "loss": 1.5949,
+      "step": 15160
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.112946510314941,
+      "learning_rate": 4.052786384333639e-05,
+      "loss": 1.5105,
+      "step": 15170
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.808885097503662,
+      "learning_rate": 4.0515400540534795e-05,
+      "loss": 1.2202,
+      "step": 15180
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.511651039123535,
+      "learning_rate": 4.050293096280408e-05,
+      "loss": 1.241,
+      "step": 15190
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.525788307189941,
+      "learning_rate": 4.0490455115187345e-05,
+      "loss": 1.2465,
+      "step": 15200
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 6.322134971618652,
+      "learning_rate": 4.047797300273023e-05,
+      "loss": 1.4229,
+      "step": 15210
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.499948978424072,
+      "learning_rate": 4.046548463048089e-05,
+      "loss": 1.4353,
+      "step": 15220
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.226128101348877,
+      "learning_rate": 4.0452990003490047e-05,
+      "loss": 1.4493,
+      "step": 15230
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 6.206206321716309,
+      "learning_rate": 4.0440489126810905e-05,
+      "loss": 1.4239,
+      "step": 15240
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.620721817016602,
+      "learning_rate": 4.042798200549924e-05,
+      "loss": 1.3828,
+      "step": 15250
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.580987930297852,
+      "learning_rate": 4.041546864461334e-05,
+      "loss": 1.3089,
+      "step": 15260
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.741886615753174,
+      "learning_rate": 4.0402949049214e-05,
+      "loss": 1.337,
+      "step": 15270
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.680039882659912,
+      "learning_rate": 4.039042322436456e-05,
+      "loss": 1.3618,
+      "step": 15280
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.636205196380615,
+      "learning_rate": 4.0377891175130866e-05,
+      "loss": 1.3015,
+      "step": 15290
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.7052717208862305,
+      "learning_rate": 4.036535290658127e-05,
+      "loss": 1.4406,
+      "step": 15300
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 10.381632804870605,
+      "learning_rate": 4.0352808423786656e-05,
+      "loss": 1.4643,
+      "step": 15310
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.433605670928955,
+      "learning_rate": 4.034025773182044e-05,
+      "loss": 1.3619,
+      "step": 15320
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.702324390411377,
+      "learning_rate": 4.032770083575851e-05,
+      "loss": 1.3252,
+      "step": 15330
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.662522315979004,
+      "learning_rate": 4.031513774067928e-05,
+      "loss": 1.2505,
+      "step": 15340
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.96264123916626,
+      "learning_rate": 4.030256845166367e-05,
+      "loss": 1.1774,
+      "step": 15350
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.252078056335449,
+      "learning_rate": 4.028999297379511e-05,
+      "loss": 1.3467,
+      "step": 15360
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.914453029632568,
+      "learning_rate": 4.027741131215953e-05,
+      "loss": 1.3049,
+      "step": 15370
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.962696552276611,
+      "learning_rate": 4.026482347184536e-05,
+      "loss": 1.5633,
+      "step": 15380
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.900122165679932,
+      "learning_rate": 4.025222945794354e-05,
+      "loss": 1.5329,
+      "step": 15390
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.105041980743408,
+      "learning_rate": 4.023962927554747e-05,
+      "loss": 1.3803,
+      "step": 15400
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.27079963684082,
+      "learning_rate": 4.022702292975309e-05,
+      "loss": 1.3135,
+      "step": 15410
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.644437789916992,
+      "learning_rate": 4.021441042565881e-05,
+      "loss": 1.5243,
+      "step": 15420
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.418429851531982,
+      "learning_rate": 4.0201791768365535e-05,
+      "loss": 1.6302,
+      "step": 15430
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.456400394439697,
+      "learning_rate": 4.018916696297665e-05,
+      "loss": 1.1917,
+      "step": 15440
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.891478538513184,
+      "learning_rate": 4.0176536014598046e-05,
+      "loss": 1.4113,
+      "step": 15450
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.790380954742432,
+      "learning_rate": 4.0163898928338086e-05,
+      "loss": 1.48,
+      "step": 15460
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.691829204559326,
+      "learning_rate": 4.01512557093076e-05,
+      "loss": 1.3284,
+      "step": 15470
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 7.182213306427002,
+      "learning_rate": 4.0138606362619926e-05,
+      "loss": 1.375,
+      "step": 15480
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.13661003112793,
+      "learning_rate": 4.0125950893390876e-05,
+      "loss": 1.1884,
+      "step": 15490
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 6.533120155334473,
+      "learning_rate": 4.011328930673871e-05,
+      "loss": 1.3406,
+      "step": 15500
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 1.1359351873397827,
+      "eval_runtime": 422.7419,
+      "eval_samples_per_second": 50.198,
+      "eval_steps_per_second": 0.393,
+      "step": 15500
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.910820960998535,
+      "learning_rate": 4.010062160778419e-05,
+      "loss": 1.4145,
+      "step": 15510
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.160300254821777,
+      "learning_rate": 4.008794780165056e-05,
+      "loss": 1.2538,
+      "step": 15520
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.033262252807617,
+      "learning_rate": 4.0075267893463494e-05,
+      "loss": 1.3626,
+      "step": 15530
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.409912109375,
+      "learning_rate": 4.006258188835117e-05,
+      "loss": 1.4094,
+      "step": 15540
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 7.784324645996094,
+      "learning_rate": 4.0049889791444215e-05,
+      "loss": 1.2388,
+      "step": 15550
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.941953659057617,
+      "learning_rate": 4.003719160787572e-05,
+      "loss": 1.2564,
+      "step": 15560
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.0910725593566895,
+      "learning_rate": 4.0024487342781244e-05,
+      "loss": 1.5045,
+      "step": 15570
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 6.084527492523193,
+      "learning_rate": 4.00117770012988e-05,
+      "loss": 1.4067,
+      "step": 15580
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 7.873744487762451,
+      "learning_rate": 3.999906058856886e-05,
+      "loss": 1.4861,
+      "step": 15590
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.216774940490723,
+      "learning_rate": 3.998633810973435e-05,
+      "loss": 1.1987,
+      "step": 15600
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.941842555999756,
+      "learning_rate": 3.9973609569940665e-05,
+      "loss": 1.1738,
+      "step": 15610
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.540301322937012,
+      "learning_rate": 3.996087497433562e-05,
+      "loss": 1.3491,
+      "step": 15620
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 6.421250820159912,
+      "learning_rate": 3.99481343280695e-05,
+      "loss": 1.3776,
+      "step": 15630
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 6.641364097595215,
+      "learning_rate": 3.993538763629504e-05,
+      "loss": 1.5226,
+      "step": 15640
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.815140247344971,
+      "learning_rate": 3.992263490416741e-05,
+      "loss": 1.5592,
+      "step": 15650
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.087193012237549,
+      "learning_rate": 3.9909876136844234e-05,
+      "loss": 1.194,
+      "step": 15660
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 6.01862096786499,
+      "learning_rate": 3.989711133948556e-05,
+      "loss": 1.4539,
+      "step": 15670
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 8.232246398925781,
+      "learning_rate": 3.988561787044916e-05,
+      "loss": 1.3606,
+      "step": 15680
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 7.1438889503479,
+      "learning_rate": 3.987284163024774e-05,
+      "loss": 1.51,
+      "step": 15690
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.273068904876709,
+      "learning_rate": 3.986005937498878e-05,
+      "loss": 1.1812,
+      "step": 15700
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.936359405517578,
+      "learning_rate": 3.984727110984183e-05,
+      "loss": 1.4972,
+      "step": 15710
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.9473676681518555,
+      "learning_rate": 3.9834476839978885e-05,
+      "loss": 1.3558,
+      "step": 15720
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.207148551940918,
+      "learning_rate": 3.982167657057435e-05,
+      "loss": 1.1452,
+      "step": 15730
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.698793411254883,
+      "learning_rate": 3.980887030680507e-05,
+      "loss": 1.3265,
+      "step": 15740
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 6.656704902648926,
+      "learning_rate": 3.9796058053850325e-05,
+      "loss": 1.327,
+      "step": 15750
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.735909461975098,
+      "learning_rate": 3.97832398168918e-05,
+      "loss": 1.0668,
+      "step": 15760
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.310609340667725,
+      "learning_rate": 3.977041560111359e-05,
+      "loss": 1.5557,
+      "step": 15770
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.834068298339844,
+      "learning_rate": 3.9757585411702234e-05,
+      "loss": 1.2306,
+      "step": 15780
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 6.531064510345459,
+      "learning_rate": 3.9744749253846674e-05,
+      "loss": 1.4083,
+      "step": 15790
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.1445231437683105,
+      "learning_rate": 3.9731907132738266e-05,
+      "loss": 1.3991,
+      "step": 15800
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.940733432769775,
+      "learning_rate": 3.971905905357077e-05,
+      "loss": 1.378,
+      "step": 15810
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 6.144431114196777,
+      "learning_rate": 3.970620502154038e-05,
+      "loss": 1.2744,
+      "step": 15820
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.148155212402344,
+      "learning_rate": 3.9693345041845675e-05,
+      "loss": 1.2378,
+      "step": 15830
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.666192054748535,
+      "learning_rate": 3.968047911968764e-05,
+      "loss": 1.2144,
+      "step": 15840
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.626368522644043,
+      "learning_rate": 3.9667607260269676e-05,
+      "loss": 1.1964,
+      "step": 15850
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 6.228915691375732,
+      "learning_rate": 3.965472946879758e-05,
+      "loss": 1.4976,
+      "step": 15860
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 6.5518903732299805,
+      "learning_rate": 3.964184575047953e-05,
+      "loss": 1.4334,
+      "step": 15870
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.513093948364258,
+      "learning_rate": 3.962895611052615e-05,
+      "loss": 1.2937,
+      "step": 15880
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.038384437561035,
+      "learning_rate": 3.961606055415039e-05,
+      "loss": 1.4432,
+      "step": 15890
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.855319499969482,
+      "learning_rate": 3.9603159086567663e-05,
+      "loss": 1.4869,
+      "step": 15900
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.632646560668945,
+      "learning_rate": 3.959025171299572e-05,
+      "loss": 1.2318,
+      "step": 15910
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.503358364105225,
+      "learning_rate": 3.957733843865472e-05,
+      "loss": 1.3042,
+      "step": 15920
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.59611177444458,
+      "learning_rate": 3.9564419268767216e-05,
+      "loss": 1.6136,
+      "step": 15930
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 7.608952045440674,
+      "learning_rate": 3.955149420855813e-05,
+      "loss": 1.3806,
+      "step": 15940
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 6.060898303985596,
+      "learning_rate": 3.953856326325478e-05,
+      "loss": 1.2115,
+      "step": 15950
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.162904739379883,
+      "learning_rate": 3.952562643808686e-05,
+      "loss": 1.2384,
+      "step": 15960
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.473962783813477,
+      "learning_rate": 3.9512683738286425e-05,
+      "loss": 1.2406,
+      "step": 15970
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 6.094630241394043,
+      "learning_rate": 3.9499735169087935e-05,
+      "loss": 1.2053,
+      "step": 15980
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.800478935241699,
+      "learning_rate": 3.948678073572821e-05,
+      "loss": 1.3536,
+      "step": 15990
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.067087650299072,
+      "learning_rate": 3.947382044344644e-05,
+      "loss": 1.4316,
+      "step": 16000
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 1.0881279706954956,
+      "eval_runtime": 422.7064,
+      "eval_samples_per_second": 50.203,
+      "eval_steps_per_second": 0.393,
+      "step": 16000
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.532190322875977,
+      "learning_rate": 3.946085429748417e-05,
+      "loss": 1.2301,
+      "step": 16010
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.910928726196289,
+      "learning_rate": 3.9447882303085346e-05,
+      "loss": 1.4426,
+      "step": 16020
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.150907039642334,
+      "learning_rate": 3.943490446549626e-05,
+      "loss": 1.0841,
+      "step": 16030
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.879810810089111,
+      "learning_rate": 3.942192078996557e-05,
+      "loss": 1.3347,
+      "step": 16040
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 7.33631706237793,
+      "learning_rate": 3.940893128174428e-05,
+      "loss": 1.5175,
+      "step": 16050
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.534356117248535,
+      "learning_rate": 3.939593594608578e-05,
+      "loss": 1.1567,
+      "step": 16060
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.3950042724609375,
+      "learning_rate": 3.9382934788245796e-05,
+      "loss": 1.5167,
+      "step": 16070
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 6.453834533691406,
+      "learning_rate": 3.9369927813482424e-05,
+      "loss": 1.4956,
+      "step": 16080
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 6.632455825805664,
+      "learning_rate": 3.93569150270561e-05,
+      "loss": 1.1387,
+      "step": 16090
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.63433313369751,
+      "learning_rate": 3.934389643422961e-05,
+      "loss": 1.3557,
+      "step": 16100
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.399450302124023,
+      "learning_rate": 3.9330872040268105e-05,
+      "loss": 1.2347,
+      "step": 16110
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.170933723449707,
+      "learning_rate": 3.931784185043906e-05,
+      "loss": 1.3735,
+      "step": 16120
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.9142746925354004,
+      "learning_rate": 3.930480587001231e-05,
+      "loss": 1.2401,
+      "step": 16130
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 7.65895938873291,
+      "learning_rate": 3.929176410426003e-05,
+      "loss": 1.4134,
+      "step": 16140
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 6.046780586242676,
+      "learning_rate": 3.927871655845672e-05,
+      "loss": 1.4892,
+      "step": 16150
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.716516494750977,
+      "learning_rate": 3.926566323787924e-05,
+      "loss": 1.4031,
+      "step": 16160
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.8535733222961426,
+      "learning_rate": 3.925260414780677e-05,
+      "loss": 1.2103,
+      "step": 16170
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.6399307250976562,
+      "learning_rate": 3.923953929352083e-05,
+      "loss": 1.2162,
+      "step": 16180
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.716061592102051,
+      "learning_rate": 3.922646868030526e-05,
+      "loss": 1.2601,
+      "step": 16190
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.244365215301514,
+      "learning_rate": 3.9213392313446254e-05,
+      "loss": 1.2144,
+      "step": 16200
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.070780277252197,
+      "learning_rate": 3.92003101982323e-05,
+      "loss": 1.4203,
+      "step": 16210
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.236509799957275,
+      "learning_rate": 3.9187222339954244e-05,
+      "loss": 1.2607,
+      "step": 16220
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 7.095030784606934,
+      "learning_rate": 3.917412874390523e-05,
+      "loss": 1.4368,
+      "step": 16230
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.535121917724609,
+      "learning_rate": 3.916102941538072e-05,
+      "loss": 1.2474,
+      "step": 16240
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.435385227203369,
+      "learning_rate": 3.914792435967853e-05,
+      "loss": 1.1781,
+      "step": 16250
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.988912105560303,
+      "learning_rate": 3.9134813582098753e-05,
+      "loss": 1.4351,
+      "step": 16260
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 6.389806747436523,
+      "learning_rate": 3.912169708794381e-05,
+      "loss": 1.5292,
+      "step": 16270
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.733895778656006,
+      "learning_rate": 3.9108574882518446e-05,
+      "loss": 1.1951,
+      "step": 16280
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 6.386997699737549,
+      "learning_rate": 3.909544697112969e-05,
+      "loss": 1.281,
+      "step": 16290
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.145005226135254,
+      "learning_rate": 3.908231335908691e-05,
+      "loss": 1.1915,
+      "step": 16300
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.459034442901611,
+      "learning_rate": 3.906917405170174e-05,
+      "loss": 1.2664,
+      "step": 16310
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 6.048981189727783,
+      "learning_rate": 3.9056029054288164e-05,
+      "loss": 1.4481,
+      "step": 16320
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.027734279632568,
+      "learning_rate": 3.904287837216243e-05,
+      "loss": 1.2802,
+      "step": 16330
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.500889778137207,
+      "learning_rate": 3.90297220106431e-05,
+      "loss": 1.4253,
+      "step": 16340
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 7.287274360656738,
+      "learning_rate": 3.901655997505103e-05,
+      "loss": 1.4855,
+      "step": 16350
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 6.854940891265869,
+      "learning_rate": 3.9003392270709383e-05,
+      "loss": 1.3279,
+      "step": 16360
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 7.035317420959473,
+      "learning_rate": 3.8990218902943594e-05,
+      "loss": 1.2614,
+      "step": 16370
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.95227575302124,
+      "learning_rate": 3.8977039877081403e-05,
+      "loss": 1.3839,
+      "step": 16380
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 6.689998626708984,
+      "learning_rate": 3.8963855198452825e-05,
+      "loss": 1.4866,
+      "step": 16390
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 6.5214409828186035,
+      "learning_rate": 3.895066487239019e-05,
+      "loss": 1.379,
+      "step": 16400
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.5395684242248535,
+      "learning_rate": 3.8937468904228065e-05,
+      "loss": 1.1929,
+      "step": 16410
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 7.409101963043213,
+      "learning_rate": 3.892426729930334e-05,
+      "loss": 1.1966,
+      "step": 16420
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.251775741577148,
+      "learning_rate": 3.891106006295517e-05,
+      "loss": 1.2691,
+      "step": 16430
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 6.069132328033447,
+      "learning_rate": 3.8897847200525e-05,
+      "loss": 1.5639,
+      "step": 16440
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 8.514307975769043,
+      "learning_rate": 3.888462871735652e-05,
+      "loss": 1.2468,
+      "step": 16450
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 7.389209270477295,
+      "learning_rate": 3.887140461879572e-05,
+      "loss": 1.1494,
+      "step": 16460
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.613709449768066,
+      "learning_rate": 3.8858174910190836e-05,
+      "loss": 1.253,
+      "step": 16470
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 6.2350029945373535,
+      "learning_rate": 3.884493959689241e-05,
+      "loss": 1.316,
+      "step": 16480
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.432788848876953,
+      "learning_rate": 3.8831698684253236e-05,
+      "loss": 1.2194,
+      "step": 16490
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 6.1027116775512695,
+      "learning_rate": 3.881845217762834e-05,
+      "loss": 1.2576,
+      "step": 16500
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 1.0837162733078003,
+      "eval_runtime": 424.5934,
+      "eval_samples_per_second": 49.98,
+      "eval_steps_per_second": 0.391,
+      "step": 16500
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 6.772694110870361,
+      "learning_rate": 3.880520008237506e-05,
+      "loss": 1.1353,
+      "step": 16510
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.501481056213379,
+      "learning_rate": 3.879194240385297e-05,
+      "loss": 1.2903,
+      "step": 16520
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.109957695007324,
+      "learning_rate": 3.877867914742388e-05,
+      "loss": 1.3333,
+      "step": 16530
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.654473781585693,
+      "learning_rate": 3.87654103184519e-05,
+      "loss": 1.3164,
+      "step": 16540
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.647524833679199,
+      "learning_rate": 3.875213592230338e-05,
+      "loss": 1.3941,
+      "step": 16550
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.180333137512207,
+      "learning_rate": 3.8738855964346895e-05,
+      "loss": 1.3035,
+      "step": 16560
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.741043567657471,
+      "learning_rate": 3.87255704499533e-05,
+      "loss": 1.3921,
+      "step": 16570
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 6.4629082679748535,
+      "learning_rate": 3.871227938449567e-05,
+      "loss": 1.339,
+      "step": 16580
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.685945510864258,
+      "learning_rate": 3.869898277334938e-05,
+      "loss": 1.4017,
+      "step": 16590
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.676754474639893,
+      "learning_rate": 3.868568062189196e-05,
+      "loss": 1.3345,
+      "step": 16600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.361738681793213,
+      "learning_rate": 3.8672372935503264e-05,
+      "loss": 1.2851,
+      "step": 16610
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.013650417327881,
+      "learning_rate": 3.865905971956534e-05,
+      "loss": 1.0779,
+      "step": 16620
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 7.297702312469482,
+      "learning_rate": 3.864574097946249e-05,
+      "loss": 1.3065,
+      "step": 16630
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.867257595062256,
+      "learning_rate": 3.8632416720581225e-05,
+      "loss": 1.2559,
+      "step": 16640
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 6.169046878814697,
+      "learning_rate": 3.861908694831031e-05,
+      "loss": 1.4102,
+      "step": 16650
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.501685619354248,
+      "learning_rate": 3.8605751668040745e-05,
+      "loss": 1.1252,
+      "step": 16660
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 6.068757057189941,
+      "learning_rate": 3.8592410885165755e-05,
+      "loss": 1.1634,
+      "step": 16670
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.935568332672119,
+      "learning_rate": 3.8579064605080764e-05,
+      "loss": 1.3675,
+      "step": 16680
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.720357894897461,
+      "learning_rate": 3.8565712833183446e-05,
+      "loss": 1.3728,
+      "step": 16690
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 6.775291442871094,
+      "learning_rate": 3.855235557487369e-05,
+      "loss": 1.1792,
+      "step": 16700
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.8963799476623535,
+      "learning_rate": 3.853899283555361e-05,
+      "loss": 1.097,
+      "step": 16710
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 7.333351135253906,
+      "learning_rate": 3.8525624620627507e-05,
+      "loss": 1.2568,
+      "step": 16720
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 6.668841361999512,
+      "learning_rate": 3.851225093550195e-05,
+      "loss": 1.4923,
+      "step": 16730
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.743914604187012,
+      "learning_rate": 3.8498871785585666e-05,
+      "loss": 1.151,
+      "step": 16740
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.761590957641602,
+      "learning_rate": 3.848548717628963e-05,
+      "loss": 1.3164,
+      "step": 16750
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.771350383758545,
+      "learning_rate": 3.847209711302699e-05,
+      "loss": 1.2282,
+      "step": 16760
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.366697788238525,
+      "learning_rate": 3.845870160121314e-05,
+      "loss": 1.2248,
+      "step": 16770
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.376469612121582,
+      "learning_rate": 3.844530064626565e-05,
+      "loss": 1.3282,
+      "step": 16780
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.593537330627441,
+      "learning_rate": 3.84318942536043e-05,
+      "loss": 1.2155,
+      "step": 16790
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 7.0101776123046875,
+      "learning_rate": 3.841848242865107e-05,
+      "loss": 1.2623,
+      "step": 16800
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.499823093414307,
+      "learning_rate": 3.840506517683012e-05,
+      "loss": 1.2206,
+      "step": 16810
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 6.118846893310547,
+      "learning_rate": 3.839164250356785e-05,
+      "loss": 1.2309,
+      "step": 16820
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.666083812713623,
+      "learning_rate": 3.83782144142928e-05,
+      "loss": 1.2501,
+      "step": 16830
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 6.0595173835754395,
+      "learning_rate": 3.836478091443573e-05,
+      "loss": 1.4198,
+      "step": 16840
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 6.150108337402344,
+      "learning_rate": 3.835134200942958e-05,
+      "loss": 1.36,
+      "step": 16850
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.785310745239258,
+      "learning_rate": 3.833789770470948e-05,
+      "loss": 1.3114,
+      "step": 16860
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.706027030944824,
+      "learning_rate": 3.832444800571273e-05,
+      "loss": 1.1982,
+      "step": 16870
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.6808207035064697,
+      "learning_rate": 3.831099291787884e-05,
+      "loss": 1.1148,
+      "step": 16880
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.5226335525512695,
+      "learning_rate": 3.829753244664946e-05,
+      "loss": 1.3141,
+      "step": 16890
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.448121070861816,
+      "learning_rate": 3.8284066597468474e-05,
+      "loss": 1.2691,
+      "step": 16900
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.62360143661499,
+      "learning_rate": 3.827059537578187e-05,
+      "loss": 1.2562,
+      "step": 16910
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 6.009716510772705,
+      "learning_rate": 3.825711878703786e-05,
+      "loss": 1.2901,
+      "step": 16920
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.977668762207031,
+      "learning_rate": 3.824363683668681e-05,
+      "loss": 1.2988,
+      "step": 16930
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.765778064727783,
+      "learning_rate": 3.823014953018127e-05,
+      "loss": 1.3024,
+      "step": 16940
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.70220947265625,
+      "learning_rate": 3.821665687297592e-05,
+      "loss": 1.0985,
+      "step": 16950
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 7.918168544769287,
+      "learning_rate": 3.820315887052764e-05,
+      "loss": 1.5052,
+      "step": 16960
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.946806907653809,
+      "learning_rate": 3.8189655528295454e-05,
+      "loss": 1.3336,
+      "step": 16970
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.795207500457764,
+      "learning_rate": 3.817614685174055e-05,
+      "loss": 1.2338,
+      "step": 16980
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 6.338341236114502,
+      "learning_rate": 3.816263284632627e-05,
+      "loss": 1.4383,
+      "step": 16990
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.565619468688965,
+      "learning_rate": 3.814911351751813e-05,
+      "loss": 1.387,
+      "step": 17000
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 1.044659972190857,
+      "eval_runtime": 422.6378,
+      "eval_samples_per_second": 50.211,
+      "eval_steps_per_second": 0.393,
+      "step": 17000
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 6.713719844818115,
+      "learning_rate": 3.8135588870783766e-05,
+      "loss": 1.3544,
+      "step": 17010
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 6.5251383781433105,
+      "learning_rate": 3.812205891159298e-05,
+      "loss": 1.2589,
+      "step": 17020
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.792092323303223,
+      "learning_rate": 3.810852364541774e-05,
+      "loss": 1.3022,
+      "step": 17030
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.9584269523620605,
+      "learning_rate": 3.809498307773214e-05,
+      "loss": 1.1615,
+      "step": 17040
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.160969257354736,
+      "learning_rate": 3.8081437214012414e-05,
+      "loss": 1.2068,
+      "step": 17050
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.709597587585449,
+      "learning_rate": 3.806788605973695e-05,
+      "loss": 1.214,
+      "step": 17060
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 7.621508598327637,
+      "learning_rate": 3.8054329620386275e-05,
+      "loss": 1.264,
+      "step": 17070
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.529162883758545,
+      "learning_rate": 3.804076790144305e-05,
+      "loss": 1.3597,
+      "step": 17080
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 7.313896179199219,
+      "learning_rate": 3.802720090839208e-05,
+      "loss": 1.4012,
+      "step": 17090
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.968306541442871,
+      "learning_rate": 3.801362864672029e-05,
+      "loss": 1.3357,
+      "step": 17100
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 6.597377300262451,
+      "learning_rate": 3.800005112191673e-05,
+      "loss": 1.2657,
+      "step": 17110
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.945515155792236,
+      "learning_rate": 3.79864683394726e-05,
+      "loss": 1.3432,
+      "step": 17120
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.914671897888184,
+      "learning_rate": 3.797288030488122e-05,
+      "loss": 1.2738,
+      "step": 17130
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.4619460105896,
+      "learning_rate": 3.7959287023638025e-05,
+      "loss": 1.2859,
+      "step": 17140
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 6.749204635620117,
+      "learning_rate": 3.7945688501240575e-05,
+      "loss": 1.5406,
+      "step": 17150
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.18287467956543,
+      "learning_rate": 3.793208474318854e-05,
+      "loss": 1.2077,
+      "step": 17160
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.32201623916626,
+      "learning_rate": 3.7918475754983754e-05,
+      "loss": 1.0533,
+      "step": 17170
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.163656234741211,
+      "learning_rate": 3.79048615421301e-05,
+      "loss": 1.3605,
+      "step": 17180
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.54106330871582,
+      "learning_rate": 3.789124211013362e-05,
+      "loss": 1.384,
+      "step": 17190
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.656838893890381,
+      "learning_rate": 3.787761746450244e-05,
+      "loss": 1.1631,
+      "step": 17200
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 6.608431339263916,
+      "learning_rate": 3.786398761074682e-05,
+      "loss": 1.2269,
+      "step": 17210
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.582605838775635,
+      "learning_rate": 3.785035255437911e-05,
+      "loss": 1.1863,
+      "step": 17220
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.8159966468811035,
+      "learning_rate": 3.783671230091376e-05,
+      "loss": 1.4262,
+      "step": 17230
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.179230690002441,
+      "learning_rate": 3.782306685586734e-05,
+      "loss": 1.252,
+      "step": 17240
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 6.152403831481934,
+      "learning_rate": 3.78094162247585e-05,
+      "loss": 1.2345,
+      "step": 17250
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.948228359222412,
+      "learning_rate": 3.7795760413107996e-05,
+      "loss": 1.3103,
+      "step": 17260
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 6.83307409286499,
+      "learning_rate": 3.778209942643867e-05,
+      "loss": 1.2482,
+      "step": 17270
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.7010018825531006,
+      "learning_rate": 3.7768433270275494e-05,
+      "loss": 1.3851,
+      "step": 17280
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 6.152671813964844,
+      "learning_rate": 3.775476195014548e-05,
+      "loss": 0.939,
+      "step": 17290
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.54927921295166,
+      "learning_rate": 3.774108547157776e-05,
+      "loss": 1.0888,
+      "step": 17300
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.867654800415039,
+      "learning_rate": 3.772740384010355e-05,
+      "loss": 1.2175,
+      "step": 17310
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 6.372371196746826,
+      "learning_rate": 3.7713717061256126e-05,
+      "loss": 1.1794,
+      "step": 17320
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.274842262268066,
+      "learning_rate": 3.770002514057088e-05,
+      "loss": 1.1586,
+      "step": 17330
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.6965556144714355,
+      "learning_rate": 3.768632808358526e-05,
+      "loss": 1.0384,
+      "step": 17340
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.417512893676758,
+      "learning_rate": 3.76726258958388e-05,
+      "loss": 1.0245,
+      "step": 17350
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.877015113830566,
+      "learning_rate": 3.76589185828731e-05,
+      "loss": 1.0279,
+      "step": 17360
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.958017826080322,
+      "learning_rate": 3.764520615023186e-05,
+      "loss": 1.3543,
+      "step": 17370
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.18721342086792,
+      "learning_rate": 3.76314886034608e-05,
+      "loss": 1.2023,
+      "step": 17380
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.118999481201172,
+      "learning_rate": 3.7617765948107764e-05,
+      "loss": 1.1819,
+      "step": 17390
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 6.0342559814453125,
+      "learning_rate": 3.7604038189722626e-05,
+      "loss": 1.188,
+      "step": 17400
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 6.279386520385742,
+      "learning_rate": 3.759030533385734e-05,
+      "loss": 1.2823,
+      "step": 17410
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 6.285080432891846,
+      "learning_rate": 3.75765673860659e-05,
+      "loss": 1.2753,
+      "step": 17420
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 8.364779472351074,
+      "learning_rate": 3.75628243519044e-05,
+      "loss": 1.3537,
+      "step": 17430
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 6.025515556335449,
+      "learning_rate": 3.754907623693096e-05,
+      "loss": 1.1277,
+      "step": 17440
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.373194217681885,
+      "learning_rate": 3.753532304670576e-05,
+      "loss": 1.1462,
+      "step": 17450
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.5704121589660645,
+      "learning_rate": 3.752156478679103e-05,
+      "loss": 1.1748,
+      "step": 17460
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.857534408569336,
+      "learning_rate": 3.7507801462751056e-05,
+      "loss": 1.179,
+      "step": 17470
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.12694787979126,
+      "learning_rate": 3.749403308015218e-05,
+      "loss": 1.2148,
+      "step": 17480
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.542255401611328,
+      "learning_rate": 3.748025964456277e-05,
+      "loss": 1.2986,
+      "step": 17490
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.159937381744385,
+      "learning_rate": 3.746648116155325e-05,
+      "loss": 1.0923,
+      "step": 17500
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 0.9958468079566956,
+      "eval_runtime": 423.0481,
+      "eval_samples_per_second": 50.162,
+      "eval_steps_per_second": 0.392,
+      "step": 17500
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.758975982666016,
+      "learning_rate": 3.7452697636696086e-05,
+      "loss": 1.4349,
+      "step": 17510
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.408443450927734,
+      "learning_rate": 3.743890907556578e-05,
+      "loss": 1.1076,
+      "step": 17520
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 6.099447727203369,
+      "learning_rate": 3.7425115483738866e-05,
+      "loss": 1.2092,
+      "step": 17530
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.1920294761657715,
+      "learning_rate": 3.741131686679393e-05,
+      "loss": 1.1028,
+      "step": 17540
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 6.446952819824219,
+      "learning_rate": 3.739751323031156e-05,
+      "loss": 1.142,
+      "step": 17550
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 6.614532470703125,
+      "learning_rate": 3.7383704579874416e-05,
+      "loss": 1.2543,
+      "step": 17560
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.392438888549805,
+      "learning_rate": 3.736989092106714e-05,
+      "loss": 1.2918,
+      "step": 17570
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.369213581085205,
+      "learning_rate": 3.735607225947644e-05,
+      "loss": 1.1161,
+      "step": 17580
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.794151306152344,
+      "learning_rate": 3.7342248600691014e-05,
+      "loss": 1.306,
+      "step": 17590
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.5254974365234375,
+      "learning_rate": 3.732841995030161e-05,
+      "loss": 1.0868,
+      "step": 17600
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.277831077575684,
+      "learning_rate": 3.7314586313900976e-05,
+      "loss": 1.2188,
+      "step": 17610
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.63234806060791,
+      "learning_rate": 3.730074769708388e-05,
+      "loss": 1.0694,
+      "step": 17620
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.33595609664917,
+      "learning_rate": 3.72869041054471e-05,
+      "loss": 1.2462,
+      "step": 17630
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 12.84477710723877,
+      "learning_rate": 3.7273055544589436e-05,
+      "loss": 1.2412,
+      "step": 17640
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.196050643920898,
+      "learning_rate": 3.7259202020111696e-05,
+      "loss": 1.3901,
+      "step": 17650
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.992873668670654,
+      "learning_rate": 3.724534353761669e-05,
+      "loss": 1.2365,
+      "step": 17660
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 7.373627185821533,
+      "learning_rate": 3.723148010270924e-05,
+      "loss": 1.1194,
+      "step": 17670
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.824416637420654,
+      "learning_rate": 3.721761172099616e-05,
+      "loss": 1.2703,
+      "step": 17680
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 6.399817943572998,
+      "learning_rate": 3.7203738398086276e-05,
+      "loss": 1.1051,
+      "step": 17690
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.0744309425354,
+      "learning_rate": 3.7189860139590405e-05,
+      "loss": 1.1408,
+      "step": 17700
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 7.6162214279174805,
+      "learning_rate": 3.717597695112136e-05,
+      "loss": 1.311,
+      "step": 17710
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.730576992034912,
+      "learning_rate": 3.716208883829396e-05,
+      "loss": 1.234,
+      "step": 17720
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 9.725975036621094,
+      "learning_rate": 3.7148195806725005e-05,
+      "loss": 1.133,
+      "step": 17730
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.541728496551514,
+      "learning_rate": 3.7134297862033284e-05,
+      "loss": 1.2321,
+      "step": 17740
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.881216049194336,
+      "learning_rate": 3.712039500983957e-05,
+      "loss": 1.247,
+      "step": 17750
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 8.307670593261719,
+      "learning_rate": 3.710648725576664e-05,
+      "loss": 1.2383,
+      "step": 17760
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.254453659057617,
+      "learning_rate": 3.7092574605439215e-05,
+      "loss": 1.3467,
+      "step": 17770
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 6.207208633422852,
+      "learning_rate": 3.707865706448404e-05,
+      "loss": 1.3152,
+      "step": 17780
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.729865312576294,
+      "learning_rate": 3.706473463852981e-05,
+      "loss": 0.9822,
+      "step": 17790
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.125967502593994,
+      "learning_rate": 3.70508073332072e-05,
+      "loss": 1.2209,
+      "step": 17800
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.6768181324005127,
+      "learning_rate": 3.703687515414887e-05,
+      "loss": 1.1339,
+      "step": 17810
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 21.38343048095703,
+      "learning_rate": 3.702293810698945e-05,
+      "loss": 1.2236,
+      "step": 17820
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 22.4886531829834,
+      "learning_rate": 3.7010390606978125e-05,
+      "loss": 2.2058,
+      "step": 17830
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 9.057819366455078,
+      "learning_rate": 3.699923397048174e-05,
+      "loss": 2.2409,
+      "step": 17840
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.9954376220703125,
+      "learning_rate": 3.698528380760968e-05,
+      "loss": 1.2043,
+      "step": 17850
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.8414082527160645,
+      "learning_rate": 3.697132879750174e-05,
+      "loss": 1.1875,
+      "step": 17860
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 6.413239479064941,
+      "learning_rate": 3.695736894580176e-05,
+      "loss": 1.2509,
+      "step": 17870
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.1529998779296875,
+      "learning_rate": 3.694340425815555e-05,
+      "loss": 1.1347,
+      "step": 17880
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.692809581756592,
+      "learning_rate": 3.69294347402109e-05,
+      "loss": 1.2004,
+      "step": 17890
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.42519474029541,
+      "learning_rate": 3.691546039761752e-05,
+      "loss": 1.2574,
+      "step": 17900
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.131531238555908,
+      "learning_rate": 3.6901481236027086e-05,
+      "loss": 1.3062,
+      "step": 17910
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 6.283532619476318,
+      "learning_rate": 3.688749726109323e-05,
+      "loss": 1.0368,
+      "step": 17920
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.127588272094727,
+      "learning_rate": 3.687350847847152e-05,
+      "loss": 1.2512,
+      "step": 17930
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.94773530960083,
+      "learning_rate": 3.685951489381947e-05,
+      "loss": 1.2949,
+      "step": 17940
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 6.109265327453613,
+      "learning_rate": 3.684551651279654e-05,
+      "loss": 1.3509,
+      "step": 17950
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 6.710040092468262,
+      "learning_rate": 3.683151334106413e-05,
+      "loss": 1.2345,
+      "step": 17960
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.863633871078491,
+      "learning_rate": 3.681750538428556e-05,
+      "loss": 1.2099,
+      "step": 17970
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 7.031714916229248,
+      "learning_rate": 3.680349264812612e-05,
+      "loss": 1.3313,
+      "step": 17980
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 6.305208206176758,
+      "learning_rate": 3.678947513825299e-05,
+      "loss": 1.0984,
+      "step": 17990
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.624855995178223,
+      "learning_rate": 3.677545286033533e-05,
+      "loss": 1.0058,
+      "step": 18000
+    },
+    {
+      "epoch": 0.71,
+      "eval_loss": 0.999274730682373,
+      "eval_runtime": 423.014,
+      "eval_samples_per_second": 50.166,
+      "eval_steps_per_second": 0.392,
+      "step": 18000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 6.157956123352051,
+      "learning_rate": 3.676142582004419e-05,
+      "loss": 0.9764,
+      "step": 18010
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.534373760223389,
+      "learning_rate": 3.6747394023052554e-05,
+      "loss": 1.1359,
+      "step": 18020
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.630082607269287,
+      "learning_rate": 3.673335747503534e-05,
+      "loss": 1.2114,
+      "step": 18030
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.868686199188232,
+      "learning_rate": 3.671931618166938e-05,
+      "loss": 1.2667,
+      "step": 18040
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.993005275726318,
+      "learning_rate": 3.6705270148633425e-05,
+      "loss": 1.1216,
+      "step": 18050
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.423141002655029,
+      "learning_rate": 3.669121938160815e-05,
+      "loss": 1.1374,
+      "step": 18060
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.673520565032959,
+      "learning_rate": 3.6677163886276127e-05,
+      "loss": 1.2182,
+      "step": 18070
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.341485500335693,
+      "learning_rate": 3.666310366832187e-05,
+      "loss": 1.0818,
+      "step": 18080
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.989436626434326,
+      "learning_rate": 3.664903873343176e-05,
+      "loss": 1.1738,
+      "step": 18090
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 7.993595600128174,
+      "learning_rate": 3.6634969087294144e-05,
+      "loss": 1.2692,
+      "step": 18100
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.427123069763184,
+      "learning_rate": 3.662089473559921e-05,
+      "loss": 1.2029,
+      "step": 18110
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 7.414592742919922,
+      "learning_rate": 3.66068156840391e-05,
+      "loss": 1.1924,
+      "step": 18120
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.597836494445801,
+      "learning_rate": 3.659273193830782e-05,
+      "loss": 1.3162,
+      "step": 18130
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.294476509094238,
+      "learning_rate": 3.6578643504101304e-05,
+      "loss": 1.2607,
+      "step": 18140
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 7.323800563812256,
+      "learning_rate": 3.656455038711738e-05,
+      "loss": 1.2322,
+      "step": 18150
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.3606157302856445,
+      "learning_rate": 3.6550452593055746e-05,
+      "loss": 1.1703,
+      "step": 18160
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.750335693359375,
+      "learning_rate": 3.6536350127618003e-05,
+      "loss": 1.1672,
+      "step": 18170
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.703127384185791,
+      "learning_rate": 3.652224299650765e-05,
+      "loss": 1.4331,
+      "step": 18180
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.751081943511963,
+      "learning_rate": 3.650813120543006e-05,
+      "loss": 1.0799,
+      "step": 18190
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.81510066986084,
+      "learning_rate": 3.649401476009251e-05,
+      "loss": 1.1139,
+      "step": 18200
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.828704357147217,
+      "learning_rate": 3.6479893666204136e-05,
+      "loss": 1.2178,
+      "step": 18210
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.051114559173584,
+      "learning_rate": 3.646576792947596e-05,
+      "loss": 1.3687,
+      "step": 18220
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.098495960235596,
+      "learning_rate": 3.64516375556209e-05,
+      "loss": 1.1869,
+      "step": 18230
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.546419620513916,
+      "learning_rate": 3.6437502550353725e-05,
+      "loss": 1.1196,
+      "step": 18240
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 18.026338577270508,
+      "learning_rate": 3.642336291939109e-05,
+      "loss": 1.1291,
+      "step": 18250
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.646088123321533,
+      "learning_rate": 3.640921866845152e-05,
+      "loss": 1.3037,
+      "step": 18260
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.589871406555176,
+      "learning_rate": 3.639506980325541e-05,
+      "loss": 1.1657,
+      "step": 18270
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.844313144683838,
+      "learning_rate": 3.638091632952501e-05,
+      "loss": 1.062,
+      "step": 18280
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.114562034606934,
+      "learning_rate": 3.6366758252984446e-05,
+      "loss": 1.0401,
+      "step": 18290
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.473639488220215,
+      "learning_rate": 3.63525955793597e-05,
+      "loss": 1.2128,
+      "step": 18300
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.314550399780273,
+      "learning_rate": 3.6338428314378624e-05,
+      "loss": 1.1442,
+      "step": 18310
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.832433223724365,
+      "learning_rate": 3.632425646377091e-05,
+      "loss": 1.2718,
+      "step": 18320
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.2858710289001465,
+      "learning_rate": 3.63100800332681e-05,
+      "loss": 1.1316,
+      "step": 18330
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 16.22000503540039,
+      "learning_rate": 3.629589902860363e-05,
+      "loss": 1.1344,
+      "step": 18340
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 6.151425361633301,
+      "learning_rate": 3.628171345551273e-05,
+      "loss": 1.2864,
+      "step": 18350
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 7.1131367683410645,
+      "learning_rate": 3.626752331973251e-05,
+      "loss": 1.2014,
+      "step": 18360
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.380446434020996,
+      "learning_rate": 3.6253328627001927e-05,
+      "loss": 1.1151,
+      "step": 18370
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 7.08466100692749,
+      "learning_rate": 3.623912938306176e-05,
+      "loss": 1.4563,
+      "step": 18380
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.9606010913848877,
+      "learning_rate": 3.6224925593654664e-05,
+      "loss": 1.4146,
+      "step": 18390
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.536890506744385,
+      "learning_rate": 3.6210717264525085e-05,
+      "loss": 1.1595,
+      "step": 18400
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.11704683303833,
+      "learning_rate": 3.6196504401419345e-05,
+      "loss": 1.1238,
+      "step": 18410
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.4873576164245605,
+      "learning_rate": 3.6182287010085586e-05,
+      "loss": 1.0774,
+      "step": 18420
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.146674156188965,
+      "learning_rate": 3.616806509627377e-05,
+      "loss": 1.1298,
+      "step": 18430
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 6.896640300750732,
+      "learning_rate": 3.61538386657357e-05,
+      "loss": 1.1547,
+      "step": 18440
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.913506507873535,
+      "learning_rate": 3.6139607724225016e-05,
+      "loss": 1.1119,
+      "step": 18450
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 7.31114387512207,
+      "learning_rate": 3.612537227749716e-05,
+      "loss": 1.2397,
+      "step": 18460
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.9692702293396,
+      "learning_rate": 3.611113233130942e-05,
+      "loss": 0.9785,
+      "step": 18470
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 6.780041217803955,
+      "learning_rate": 3.609688789142086e-05,
+      "loss": 1.2358,
+      "step": 18480
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.687527179718018,
+      "learning_rate": 3.608263896359242e-05,
+      "loss": 1.032,
+      "step": 18490
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.15678596496582,
+      "learning_rate": 3.606838555358682e-05,
+      "loss": 1.063,
+      "step": 18500
+    },
+    {
+      "epoch": 0.73,
+      "eval_loss": 0.9674268960952759,
+      "eval_runtime": 423.0316,
+      "eval_samples_per_second": 50.164,
+      "eval_steps_per_second": 0.392,
+      "step": 18500
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.980559825897217,
+      "learning_rate": 3.6054127667168596e-05,
+      "loss": 1.2599,
+      "step": 18510
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 6.483269691467285,
+      "learning_rate": 3.603986531010409e-05,
+      "loss": 1.22,
+      "step": 18520
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.8804402351379395,
+      "learning_rate": 3.602559848816148e-05,
+      "loss": 1.2753,
+      "step": 18530
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 6.696392059326172,
+      "learning_rate": 3.6011327207110715e-05,
+      "loss": 1.2061,
+      "step": 18540
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 6.158512592315674,
+      "learning_rate": 3.5997051472723566e-05,
+      "loss": 1.1812,
+      "step": 18550
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.274986743927002,
+      "learning_rate": 3.5982771290773595e-05,
+      "loss": 1.1922,
+      "step": 18560
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 7.274186611175537,
+      "learning_rate": 3.596848666703617e-05,
+      "loss": 1.1734,
+      "step": 18570
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 6.14224100112915,
+      "learning_rate": 3.595419760728847e-05,
+      "loss": 1.2075,
+      "step": 18580
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.483425140380859,
+      "learning_rate": 3.593990411730944e-05,
+      "loss": 1.0428,
+      "step": 18590
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.514137268066406,
+      "learning_rate": 3.592560620287981e-05,
+      "loss": 1.1947,
+      "step": 18600
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.059767723083496,
+      "learning_rate": 3.591130386978216e-05,
+      "loss": 1.2259,
+      "step": 18610
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 7.321201324462891,
+      "learning_rate": 3.589699712380079e-05,
+      "loss": 1.2213,
+      "step": 18620
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 6.80574369430542,
+      "learning_rate": 3.588268597072181e-05,
+      "loss": 1.2357,
+      "step": 18630
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 6.115262031555176,
+      "learning_rate": 3.586837041633312e-05,
+      "loss": 1.2932,
+      "step": 18640
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 6.366672992706299,
+      "learning_rate": 3.585405046642438e-05,
+      "loss": 1.1897,
+      "step": 18650
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 7.872925281524658,
+      "learning_rate": 3.5839726126787056e-05,
+      "loss": 1.1749,
+      "step": 18660
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.9373064041137695,
+      "learning_rate": 3.5825397403214364e-05,
+      "loss": 1.3522,
+      "step": 18670
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.59600305557251,
+      "learning_rate": 3.5811064301501305e-05,
+      "loss": 1.1961,
+      "step": 18680
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.240240573883057,
+      "learning_rate": 3.5796726827444646e-05,
+      "loss": 1.16,
+      "step": 18690
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 7.09135627746582,
+      "learning_rate": 3.578238498684293e-05,
+      "loss": 1.2919,
+      "step": 18700
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.37578010559082,
+      "learning_rate": 3.576803878549646e-05,
+      "loss": 1.3254,
+      "step": 18710
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.90376091003418,
+      "learning_rate": 3.57536882292073e-05,
+      "loss": 1.14,
+      "step": 18720
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 6.709833145141602,
+      "learning_rate": 3.573933332377928e-05,
+      "loss": 1.289,
+      "step": 18730
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.5887227058410645,
+      "learning_rate": 3.5724974075017984e-05,
+      "loss": 1.1695,
+      "step": 18740
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 7.202672481536865,
+      "learning_rate": 3.5710610488730764e-05,
+      "loss": 1.2715,
+      "step": 18750
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.508922100067139,
+      "learning_rate": 3.569624257072671e-05,
+      "loss": 1.1112,
+      "step": 18760
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.2764153480529785,
+      "learning_rate": 3.568187032681667e-05,
+      "loss": 1.0467,
+      "step": 18770
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.631936550140381,
+      "learning_rate": 3.566749376281325e-05,
+      "loss": 1.1183,
+      "step": 18780
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 7.92403507232666,
+      "learning_rate": 3.56531128845308e-05,
+      "loss": 1.2647,
+      "step": 18790
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 6.557210922241211,
+      "learning_rate": 3.563872769778541e-05,
+      "loss": 1.396,
+      "step": 18800
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.717602252960205,
+      "learning_rate": 3.562433820839491e-05,
+      "loss": 1.5048,
+      "step": 18810
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 8.679744720458984,
+      "learning_rate": 3.560994442217887e-05,
+      "loss": 1.2151,
+      "step": 18820
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.491687297821045,
+      "learning_rate": 3.5595546344958614e-05,
+      "loss": 1.0843,
+      "step": 18830
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.3524065017700195,
+      "learning_rate": 3.558114398255718e-05,
+      "loss": 1.1617,
+      "step": 18840
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.459160804748535,
+      "learning_rate": 3.556673734079936e-05,
+      "loss": 1.2338,
+      "step": 18850
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.755857944488525,
+      "learning_rate": 3.555232642551164e-05,
+      "loss": 1.0757,
+      "step": 18860
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 14.724515914916992,
+      "learning_rate": 3.553791124252229e-05,
+      "loss": 1.1567,
+      "step": 18870
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.496631145477295,
+      "learning_rate": 3.552349179766126e-05,
+      "loss": 1.2369,
+      "step": 18880
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.554177284240723,
+      "learning_rate": 3.550906809676024e-05,
+      "loss": 1.1991,
+      "step": 18890
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.716100692749023,
+      "learning_rate": 3.549464014565265e-05,
+      "loss": 1.26,
+      "step": 18900
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.523488521575928,
+      "learning_rate": 3.54802079501736e-05,
+      "loss": 1.0827,
+      "step": 18910
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.5706892013549805,
+      "learning_rate": 3.546577151615995e-05,
+      "loss": 0.9701,
+      "step": 18920
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.8000569343566895,
+      "learning_rate": 3.5451330849450257e-05,
+      "loss": 0.9931,
+      "step": 18930
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.145776748657227,
+      "learning_rate": 3.5436885955884804e-05,
+      "loss": 0.9592,
+      "step": 18940
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 7.698915481567383,
+      "learning_rate": 3.542243684130555e-05,
+      "loss": 1.3278,
+      "step": 18950
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 6.6572418212890625,
+      "learning_rate": 3.54079835115562e-05,
+      "loss": 1.4969,
+      "step": 18960
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.524096488952637,
+      "learning_rate": 3.5393525972482146e-05,
+      "loss": 1.2471,
+      "step": 18970
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 6.782894611358643,
+      "learning_rate": 3.537906422993048e-05,
+      "loss": 1.3066,
+      "step": 18980
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.772091865539551,
+      "learning_rate": 3.536459828975e-05,
+      "loss": 1.1938,
+      "step": 18990
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 7.265366077423096,
+      "learning_rate": 3.5350128157791195e-05,
+      "loss": 1.2417,
+      "step": 19000
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.9906503558158875,
+      "eval_runtime": 423.0609,
+      "eval_samples_per_second": 50.161,
+      "eval_steps_per_second": 0.392,
+      "step": 19000
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.862709999084473,
+      "learning_rate": 3.533565383990626e-05,
+      "loss": 1.1364,
+      "step": 19010
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 7.6012983322143555,
+      "learning_rate": 3.532117534194907e-05,
+      "loss": 1.2322,
+      "step": 19020
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.332521915435791,
+      "learning_rate": 3.530669266977521e-05,
+      "loss": 1.0388,
+      "step": 19030
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 6.732748985290527,
+      "learning_rate": 3.529220582924193e-05,
+      "loss": 1.2087,
+      "step": 19040
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.4480390548706055,
+      "learning_rate": 3.527771482620817e-05,
+      "loss": 1.3163,
+      "step": 19050
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.523915767669678,
+      "learning_rate": 3.5263219666534566e-05,
+      "loss": 1.1824,
+      "step": 19060
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 6.780529022216797,
+      "learning_rate": 3.524872035608344e-05,
+      "loss": 1.0323,
+      "step": 19070
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.73441743850708,
+      "learning_rate": 3.523421690071876e-05,
+      "loss": 1.2144,
+      "step": 19080
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 6.4305338859558105,
+      "learning_rate": 3.521970930630621e-05,
+      "loss": 1.1832,
+      "step": 19090
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.339730739593506,
+      "learning_rate": 3.520519757871313e-05,
+      "loss": 1.1666,
+      "step": 19100
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.925113201141357,
+      "learning_rate": 3.5190681723808515e-05,
+      "loss": 1.0603,
+      "step": 19110
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 6.413649559020996,
+      "learning_rate": 3.5176161747463056e-05,
+      "loss": 1.1603,
+      "step": 19120
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.945598125457764,
+      "learning_rate": 3.51616376555491e-05,
+      "loss": 1.1038,
+      "step": 19130
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.29184627532959,
+      "learning_rate": 3.514710945394066e-05,
+      "loss": 1.0526,
+      "step": 19140
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.155411243438721,
+      "learning_rate": 3.513257714851341e-05,
+      "loss": 0.9027,
+      "step": 19150
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 7.066675662994385,
+      "learning_rate": 3.511804074514468e-05,
+      "loss": 1.0971,
+      "step": 19160
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 6.2751665115356445,
+      "learning_rate": 3.510350024971347e-05,
+      "loss": 1.1794,
+      "step": 19170
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.440314292907715,
+      "learning_rate": 3.5088955668100414e-05,
+      "loss": 1.0865,
+      "step": 19180
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 6.728399753570557,
+      "learning_rate": 3.5074407006187816e-05,
+      "loss": 1.2261,
+      "step": 19190
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.077253341674805,
+      "learning_rate": 3.505985426985962e-05,
+      "loss": 1.0304,
+      "step": 19200
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 7.649025917053223,
+      "learning_rate": 3.504529746500142e-05,
+      "loss": 1.0394,
+      "step": 19210
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 6.106729984283447,
+      "learning_rate": 3.503073659750048e-05,
+      "loss": 1.3419,
+      "step": 19220
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.529127597808838,
+      "learning_rate": 3.5016171673245665e-05,
+      "loss": 1.2145,
+      "step": 19230
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.942320346832275,
+      "learning_rate": 3.500160269812751e-05,
+      "loss": 1.2657,
+      "step": 19240
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.263607978820801,
+      "learning_rate": 3.498702967803817e-05,
+      "loss": 1.2116,
+      "step": 19250
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.790194034576416,
+      "learning_rate": 3.4972452618871456e-05,
+      "loss": 1.2305,
+      "step": 19260
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 6.1989264488220215,
+      "learning_rate": 3.495787152652279e-05,
+      "loss": 1.1694,
+      "step": 19270
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.459064960479736,
+      "learning_rate": 3.494328640688925e-05,
+      "loss": 1.3178,
+      "step": 19280
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.873173713684082,
+      "learning_rate": 3.4928697265869515e-05,
+      "loss": 1.146,
+      "step": 19290
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 6.925245761871338,
+      "learning_rate": 3.491410410936392e-05,
+      "loss": 1.3765,
+      "step": 19300
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 6.337075233459473,
+      "learning_rate": 3.489950694327439e-05,
+      "loss": 1.1616,
+      "step": 19310
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.319609642028809,
+      "learning_rate": 3.488490577350451e-05,
+      "loss": 1.2877,
+      "step": 19320
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 8.024974822998047,
+      "learning_rate": 3.487030060595945e-05,
+      "loss": 1.1263,
+      "step": 19330
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.91800594329834,
+      "learning_rate": 3.485569144654603e-05,
+      "loss": 1.2216,
+      "step": 19340
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.766550064086914,
+      "learning_rate": 3.4841078301172635e-05,
+      "loss": 1.0855,
+      "step": 19350
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 6.872048377990723,
+      "learning_rate": 3.482646117574932e-05,
+      "loss": 1.1821,
+      "step": 19360
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 7.4065399169921875,
+      "learning_rate": 3.4811840076187714e-05,
+      "loss": 1.276,
+      "step": 19370
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.74197244644165,
+      "learning_rate": 3.4797215008401056e-05,
+      "loss": 1.0844,
+      "step": 19380
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 6.256606578826904,
+      "learning_rate": 3.478258597830421e-05,
+      "loss": 1.322,
+      "step": 19390
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 6.4911956787109375,
+      "learning_rate": 3.476795299181362e-05,
+      "loss": 1.3724,
+      "step": 19400
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 7.156044006347656,
+      "learning_rate": 3.475331605484733e-05,
+      "loss": 1.2932,
+      "step": 19410
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 6.791740417480469,
+      "learning_rate": 3.473867517332501e-05,
+      "loss": 1.1424,
+      "step": 19420
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.755488872528076,
+      "learning_rate": 3.472403035316789e-05,
+      "loss": 1.1266,
+      "step": 19430
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 6.027830123901367,
+      "learning_rate": 3.470938160029881e-05,
+      "loss": 1.3218,
+      "step": 19440
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 7.475457668304443,
+      "learning_rate": 3.46947289206422e-05,
+      "loss": 1.0366,
+      "step": 19450
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 6.005370140075684,
+      "learning_rate": 3.468007232012408e-05,
+      "loss": 1.094,
+      "step": 19460
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 8.539311408996582,
+      "learning_rate": 3.466541180467204e-05,
+      "loss": 1.1254,
+      "step": 19470
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.714039325714111,
+      "learning_rate": 3.4650747380215294e-05,
+      "loss": 1.1311,
+      "step": 19480
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 6.291929721832275,
+      "learning_rate": 3.4636079052684576e-05,
+      "loss": 1.2993,
+      "step": 19490
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.029024124145508,
+      "learning_rate": 3.4621406828012257e-05,
+      "loss": 1.0889,
+      "step": 19500
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 0.9187302589416504,
+      "eval_runtime": 422.8354,
+      "eval_samples_per_second": 50.187,
+      "eval_steps_per_second": 0.393,
+      "step": 19500
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 6.371516704559326,
+      "learning_rate": 3.460673071213224e-05,
+      "loss": 1.116,
+      "step": 19510
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.552900791168213,
+      "learning_rate": 3.4592050710980035e-05,
+      "loss": 1.0279,
+      "step": 19520
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 6.127098560333252,
+      "learning_rate": 3.4577366830492703e-05,
+      "loss": 1.0784,
+      "step": 19530
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 6.475603103637695,
+      "learning_rate": 3.456267907660888e-05,
+      "loss": 1.0933,
+      "step": 19540
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.963050365447998,
+      "learning_rate": 3.454798745526876e-05,
+      "loss": 1.1908,
+      "step": 19550
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.997339248657227,
+      "learning_rate": 3.453329197241413e-05,
+      "loss": 1.0984,
+      "step": 19560
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9980947971343994,
+      "learning_rate": 3.451859263398829e-05,
+      "loss": 1.133,
+      "step": 19570
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.910024642944336,
+      "learning_rate": 3.4503889445936145e-05,
+      "loss": 0.8625,
+      "step": 19580
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.042624473571777,
+      "learning_rate": 3.448918241420414e-05,
+      "loss": 0.9984,
+      "step": 19590
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.292120933532715,
+      "learning_rate": 3.447447154474026e-05,
+      "loss": 1.1071,
+      "step": 19600
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.484497547149658,
+      "learning_rate": 3.4459756843494075e-05,
+      "loss": 0.8867,
+      "step": 19610
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.6151628494262695,
+      "learning_rate": 3.444503831641667e-05,
+      "loss": 1.0767,
+      "step": 19620
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.444524765014648,
+      "learning_rate": 3.44303159694607e-05,
+      "loss": 1.1059,
+      "step": 19630
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.467438697814941,
+      "learning_rate": 3.441558980858035e-05,
+      "loss": 1.0952,
+      "step": 19640
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.955422878265381,
+      "learning_rate": 3.4400859839731366e-05,
+      "loss": 1.0664,
+      "step": 19650
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 6.156907558441162,
+      "learning_rate": 3.438612606887101e-05,
+      "loss": 1.2699,
+      "step": 19660
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.79284143447876,
+      "learning_rate": 3.4371388501958116e-05,
+      "loss": 1.0393,
+      "step": 19670
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.800482273101807,
+      "learning_rate": 3.435664714495301e-05,
+      "loss": 1.1332,
+      "step": 19680
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.0411601066589355,
+      "learning_rate": 3.434190200381758e-05,
+      "loss": 1.1577,
+      "step": 19690
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.855594635009766,
+      "learning_rate": 3.432715308451525e-05,
+      "loss": 0.9627,
+      "step": 19700
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.704545021057129,
+      "learning_rate": 3.431240039301095e-05,
+      "loss": 0.9416,
+      "step": 19710
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.06236457824707,
+      "learning_rate": 3.4297643935271133e-05,
+      "loss": 1.0789,
+      "step": 19720
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 7.57991886138916,
+      "learning_rate": 3.4282883717263803e-05,
+      "loss": 1.2021,
+      "step": 19730
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.65207052230835,
+      "learning_rate": 3.426811974495846e-05,
+      "loss": 1.0169,
+      "step": 19740
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.683870792388916,
+      "learning_rate": 3.425335202432614e-05,
+      "loss": 1.101,
+      "step": 19750
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 6.760273456573486,
+      "learning_rate": 3.423858056133937e-05,
+      "loss": 0.9897,
+      "step": 19760
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.750148773193359,
+      "learning_rate": 3.422380536197223e-05,
+      "loss": 1.1489,
+      "step": 19770
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 6.0316925048828125,
+      "learning_rate": 3.420902643220027e-05,
+      "loss": 0.9949,
+      "step": 19780
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.253840923309326,
+      "learning_rate": 3.419424377800057e-05,
+      "loss": 1.117,
+      "step": 19790
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.083964824676514,
+      "learning_rate": 3.417945740535172e-05,
+      "loss": 1.074,
+      "step": 19800
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.166134834289551,
+      "learning_rate": 3.41646673202338e-05,
+      "loss": 0.9774,
+      "step": 19810
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.710581302642822,
+      "learning_rate": 3.414987352862839e-05,
+      "loss": 1.1299,
+      "step": 19820
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.255856037139893,
+      "learning_rate": 3.413507603651861e-05,
+      "loss": 1.1568,
+      "step": 19830
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.32375431060791,
+      "learning_rate": 3.412027484988901e-05,
+      "loss": 1.1712,
+      "step": 19840
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.535093784332275,
+      "learning_rate": 3.4105469974725674e-05,
+      "loss": 1.0975,
+      "step": 19850
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.150576114654541,
+      "learning_rate": 3.4090661417016175e-05,
+      "loss": 1.0086,
+      "step": 19860
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 7.549522399902344,
+      "learning_rate": 3.407584918274958e-05,
+      "loss": 1.2344,
+      "step": 19870
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.6424384117126465,
+      "learning_rate": 3.4061033277916424e-05,
+      "loss": 1.1765,
+      "step": 19880
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.286909103393555,
+      "learning_rate": 3.404621370850875e-05,
+      "loss": 1.1065,
+      "step": 19890
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.180936336517334,
+      "learning_rate": 3.4031390480520056e-05,
+      "loss": 1.1878,
+      "step": 19900
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.357401132583618,
+      "learning_rate": 3.401656359994534e-05,
+      "loss": 0.9703,
+      "step": 19910
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.909946441650391,
+      "learning_rate": 3.400173307278107e-05,
+      "loss": 0.9868,
+      "step": 19920
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.067035675048828,
+      "learning_rate": 3.398689890502519e-05,
+      "loss": 1.3108,
+      "step": 19930
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.286576747894287,
+      "learning_rate": 3.397206110267713e-05,
+      "loss": 1.2323,
+      "step": 19940
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.54276704788208,
+      "learning_rate": 3.3957219671737764e-05,
+      "loss": 1.0686,
+      "step": 19950
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.040194034576416,
+      "learning_rate": 3.394237461820944e-05,
+      "loss": 0.9858,
+      "step": 19960
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.379824638366699,
+      "learning_rate": 3.3927525948095984e-05,
+      "loss": 1.075,
+      "step": 19970
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.724083423614502,
+      "learning_rate": 3.391267366740268e-05,
+      "loss": 1.1752,
+      "step": 19980
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.3904337882995605,
+      "learning_rate": 3.389781778213628e-05,
+      "loss": 1.1794,
+      "step": 19990
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.583755016326904,
+      "learning_rate": 3.388295829830496e-05,
+      "loss": 1.1985,
+      "step": 20000
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 0.9072844982147217,
+      "eval_runtime": 423.2237,
+      "eval_samples_per_second": 50.141,
+      "eval_steps_per_second": 0.392,
+      "step": 20000
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.750321865081787,
+      "learning_rate": 3.3868095221918385e-05,
+      "loss": 1.0204,
+      "step": 20010
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.180161952972412,
+      "learning_rate": 3.385322855898767e-05,
+      "loss": 1.3302,
+      "step": 20020
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.926629066467285,
+      "learning_rate": 3.383835831552537e-05,
+      "loss": 1.1546,
+      "step": 20030
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.6702046394348145,
+      "learning_rate": 3.3823484497545495e-05,
+      "loss": 1.2066,
+      "step": 20040
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.437082290649414,
+      "learning_rate": 3.380860711106348e-05,
+      "loss": 1.0266,
+      "step": 20050
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.182374954223633,
+      "learning_rate": 3.3793726162096254e-05,
+      "loss": 1.0211,
+      "step": 20060
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 6.240570068359375,
+      "learning_rate": 3.377884165666212e-05,
+      "loss": 1.1501,
+      "step": 20070
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.38503360748291,
+      "learning_rate": 3.376395360078087e-05,
+      "loss": 1.0631,
+      "step": 20080
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 6.204507827758789,
+      "learning_rate": 3.374906200047371e-05,
+      "loss": 1.2123,
+      "step": 20090
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.898905277252197,
+      "learning_rate": 3.3734166861763284e-05,
+      "loss": 1.3479,
+      "step": 20100
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.1880645751953125,
+      "learning_rate": 3.371926819067367e-05,
+      "loss": 1.2629,
+      "step": 20110
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.070500373840332,
+      "learning_rate": 3.3704365993230366e-05,
+      "loss": 1.0535,
+      "step": 20120
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 7.216237545013428,
+      "learning_rate": 3.368946027546032e-05,
+      "loss": 1.3136,
+      "step": 20130
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.297421932220459,
+      "learning_rate": 3.3674551043391855e-05,
+      "loss": 1.025,
+      "step": 20140
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 10.107786178588867,
+      "learning_rate": 3.365963830305477e-05,
+      "loss": 1.1476,
+      "step": 20150
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.5583109855651855,
+      "learning_rate": 3.3644722060480246e-05,
+      "loss": 1.3907,
+      "step": 20160
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.232701301574707,
+      "learning_rate": 3.362980232170091e-05,
+      "loss": 1.0455,
+      "step": 20170
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 6.205572128295898,
+      "learning_rate": 3.361487909275077e-05,
+      "loss": 1.0491,
+      "step": 20180
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 6.285012245178223,
+      "learning_rate": 3.3599952379665265e-05,
+      "loss": 1.336,
+      "step": 20190
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.088164329528809,
+      "learning_rate": 3.358502218848125e-05,
+      "loss": 1.2031,
+      "step": 20200
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.3296942710876465,
+      "learning_rate": 3.357008852523695e-05,
+      "loss": 1.093,
+      "step": 20210
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.5712761878967285,
+      "learning_rate": 3.355515139597206e-05,
+      "loss": 0.9851,
+      "step": 20220
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.023895740509033,
+      "learning_rate": 3.3540210806727605e-05,
+      "loss": 1.0393,
+      "step": 20230
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.967472553253174,
+      "learning_rate": 3.352526676354606e-05,
+      "loss": 1.1072,
+      "step": 20240
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.862733840942383,
+      "learning_rate": 3.351031927247127e-05,
+      "loss": 1.1496,
+      "step": 20250
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.912438869476318,
+      "learning_rate": 3.349536833954849e-05,
+      "loss": 1.0439,
+      "step": 20260
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 6.724765777587891,
+      "learning_rate": 3.348041397082434e-05,
+      "loss": 1.0515,
+      "step": 20270
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 6.460733413696289,
+      "learning_rate": 3.346545617234688e-05,
+      "loss": 0.9933,
+      "step": 20280
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 7.006014347076416,
+      "learning_rate": 3.345049495016551e-05,
+      "loss": 1.1974,
+      "step": 20290
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.221912384033203,
+      "learning_rate": 3.343553031033103e-05,
+      "loss": 1.1806,
+      "step": 20300
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.629359722137451,
+      "learning_rate": 3.342056225889563e-05,
+      "loss": 1.2486,
+      "step": 20310
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.0083770751953125,
+      "learning_rate": 3.3405590801912874e-05,
+      "loss": 1.1473,
+      "step": 20320
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.067394256591797,
+      "learning_rate": 3.33906159454377e-05,
+      "loss": 1.3352,
+      "step": 20330
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.185932159423828,
+      "learning_rate": 3.337563769552642e-05,
+      "loss": 1.3216,
+      "step": 20340
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.5378618240356445,
+      "learning_rate": 3.336065605823672e-05,
+      "loss": 1.2144,
+      "step": 20350
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.7176525592803955,
+      "learning_rate": 3.3345671039627666e-05,
+      "loss": 1.0522,
+      "step": 20360
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.243635177612305,
+      "learning_rate": 3.333068264575969e-05,
+      "loss": 1.2082,
+      "step": 20370
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.2928547859191895,
+      "learning_rate": 3.331569088269457e-05,
+      "loss": 1.2175,
+      "step": 20380
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.688693523406982,
+      "learning_rate": 3.330069575649546e-05,
+      "loss": 1.0414,
+      "step": 20390
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.3996357917785645,
+      "learning_rate": 3.328569727322688e-05,
+      "loss": 1.0109,
+      "step": 20400
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.147793292999268,
+      "learning_rate": 3.32706954389547e-05,
+      "loss": 1.3541,
+      "step": 20410
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.977090835571289,
+      "learning_rate": 3.3255690259746136e-05,
+      "loss": 1.2724,
+      "step": 20420
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.9551801681518555,
+      "learning_rate": 3.324068174166977e-05,
+      "loss": 1.1175,
+      "step": 20430
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.526800155639648,
+      "learning_rate": 3.322566989079554e-05,
+      "loss": 1.1945,
+      "step": 20440
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.269755840301514,
+      "learning_rate": 3.3210654713194726e-05,
+      "loss": 1.0807,
+      "step": 20450
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.062760829925537,
+      "learning_rate": 3.319563621493994e-05,
+      "loss": 1.39,
+      "step": 20460
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.831636905670166,
+      "learning_rate": 3.318061440210516e-05,
+      "loss": 1.2049,
+      "step": 20470
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.105273246765137,
+      "learning_rate": 3.3165589280765675e-05,
+      "loss": 1.1115,
+      "step": 20480
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.841034889221191,
+      "learning_rate": 3.315056085699814e-05,
+      "loss": 1.0631,
+      "step": 20490
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.890759468078613,
+      "learning_rate": 3.313552913688054e-05,
+      "loss": 1.1866,
+      "step": 20500
+    },
+    {
+      "epoch": 0.81,
+      "eval_loss": 0.9173963069915771,
+      "eval_runtime": 423.2558,
+      "eval_samples_per_second": 50.138,
+      "eval_steps_per_second": 0.392,
+      "step": 20500
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.454107284545898,
+      "learning_rate": 3.312049412649218e-05,
+      "loss": 1.0349,
+      "step": 20510
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.466691970825195,
+      "learning_rate": 3.310545583191371e-05,
+      "loss": 1.0189,
+      "step": 20520
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.449139595031738,
+      "learning_rate": 3.3090414259227084e-05,
+      "loss": 1.0081,
+      "step": 20530
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.641279697418213,
+      "learning_rate": 3.307536941451563e-05,
+      "loss": 1.1935,
+      "step": 20540
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.352484226226807,
+      "learning_rate": 3.306032130386394e-05,
+      "loss": 0.8546,
+      "step": 20550
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 6.614115238189697,
+      "learning_rate": 3.304526993335797e-05,
+      "loss": 1.0819,
+      "step": 20560
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4167211055755615,
+      "learning_rate": 3.303021530908499e-05,
+      "loss": 1.1848,
+      "step": 20570
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 6.5440239906311035,
+      "learning_rate": 3.3015157437133556e-05,
+      "loss": 1.2462,
+      "step": 20580
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.0716633796691895,
+      "learning_rate": 3.300009632359357e-05,
+      "loss": 1.3046,
+      "step": 20590
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.870248794555664,
+      "learning_rate": 3.298503197455623e-05,
+      "loss": 0.9689,
+      "step": 20600
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.813103199005127,
+      "learning_rate": 3.296996439611404e-05,
+      "loss": 1.0596,
+      "step": 20610
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.267066955566406,
+      "learning_rate": 3.295489359436083e-05,
+      "loss": 1.1432,
+      "step": 20620
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.67086935043335,
+      "learning_rate": 3.293981957539171e-05,
+      "loss": 0.9654,
+      "step": 20630
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 7.342175483703613,
+      "learning_rate": 3.292474234530309e-05,
+      "loss": 1.1707,
+      "step": 20640
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.687649726867676,
+      "learning_rate": 3.290966191019271e-05,
+      "loss": 1.1953,
+      "step": 20650
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.920400619506836,
+      "learning_rate": 3.2894578276159576e-05,
+      "loss": 1.2348,
+      "step": 20660
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 6.004182815551758,
+      "learning_rate": 3.287949144930399e-05,
+      "loss": 1.3107,
+      "step": 20670
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.966498851776123,
+      "learning_rate": 3.286440143572756e-05,
+      "loss": 1.0045,
+      "step": 20680
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.098879814147949,
+      "learning_rate": 3.284930824153316e-05,
+      "loss": 1.102,
+      "step": 20690
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.4774041175842285,
+      "learning_rate": 3.283421187282498e-05,
+      "loss": 1.2144,
+      "step": 20700
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.517295837402344,
+      "learning_rate": 3.281911233570848e-05,
+      "loss": 1.0952,
+      "step": 20710
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 6.00652551651001,
+      "learning_rate": 3.28040096362904e-05,
+      "loss": 1.0454,
+      "step": 20720
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.340013027191162,
+      "learning_rate": 3.2788903780678746e-05,
+      "loss": 1.1245,
+      "step": 20730
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.672094345092773,
+      "learning_rate": 3.277379477498284e-05,
+      "loss": 1.0511,
+      "step": 20740
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 6.4503068923950195,
+      "learning_rate": 3.275868262531322e-05,
+      "loss": 1.3228,
+      "step": 20750
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.11310338973999,
+      "learning_rate": 3.274356733778175e-05,
+      "loss": 1.2805,
+      "step": 20760
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.5658721923828125,
+      "learning_rate": 3.2728448918501536e-05,
+      "loss": 1.16,
+      "step": 20770
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 7.478603363037109,
+      "learning_rate": 3.271332737358695e-05,
+      "loss": 1.281,
+      "step": 20780
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.312899589538574,
+      "learning_rate": 3.269820270915364e-05,
+      "loss": 1.0734,
+      "step": 20790
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 7.332522869110107,
+      "learning_rate": 3.26830749313185e-05,
+      "loss": 1.2745,
+      "step": 20800
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.943617820739746,
+      "learning_rate": 3.2667944046199704e-05,
+      "loss": 0.9687,
+      "step": 20810
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 6.301081657409668,
+      "learning_rate": 3.265281005991666e-05,
+      "loss": 1.1392,
+      "step": 20820
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 6.474895000457764,
+      "learning_rate": 3.2637672978590036e-05,
+      "loss": 1.2522,
+      "step": 20830
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 7.805530548095703,
+      "learning_rate": 3.262253280834178e-05,
+      "loss": 1.1372,
+      "step": 20840
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.137313365936279,
+      "learning_rate": 3.260738955529504e-05,
+      "loss": 0.9912,
+      "step": 20850
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 7.606447219848633,
+      "learning_rate": 3.2592243225574246e-05,
+      "loss": 1.3441,
+      "step": 20860
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.675898551940918,
+      "learning_rate": 3.257709382530507e-05,
+      "loss": 0.8928,
+      "step": 20870
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.256418704986572,
+      "learning_rate": 3.2561941360614414e-05,
+      "loss": 1.0389,
+      "step": 20880
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.057074546813965,
+      "learning_rate": 3.254678583763042e-05,
+      "loss": 1.1096,
+      "step": 20890
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.805013418197632,
+      "learning_rate": 3.253162726248248e-05,
+      "loss": 1.0386,
+      "step": 20900
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 7.048025131225586,
+      "learning_rate": 3.25164656413012e-05,
+      "loss": 1.1397,
+      "step": 20910
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.4225094318389893,
+      "learning_rate": 3.250130098021844e-05,
+      "loss": 1.1548,
+      "step": 20920
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 6.029432773590088,
+      "learning_rate": 3.248613328536727e-05,
+      "loss": 1.0185,
+      "step": 20930
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.50508975982666,
+      "learning_rate": 3.2470962562882e-05,
+      "loss": 1.1095,
+      "step": 20940
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.5588302612304688,
+      "learning_rate": 3.245578881889816e-05,
+      "loss": 0.9348,
+      "step": 20950
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.503541469573975,
+      "learning_rate": 3.244061205955251e-05,
+      "loss": 1.1805,
+      "step": 20960
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.023471832275391,
+      "learning_rate": 3.242543229098301e-05,
+      "loss": 1.1385,
+      "step": 20970
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.005465984344482,
+      "learning_rate": 3.241024951932885e-05,
+      "loss": 1.1415,
+      "step": 20980
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 6.451344966888428,
+      "learning_rate": 3.2395063750730434e-05,
+      "loss": 1.0934,
+      "step": 20990
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.293153762817383,
+      "learning_rate": 3.2379874991329374e-05,
+      "loss": 1.0381,
+      "step": 21000
+    },
+    {
+      "epoch": 0.83,
+      "eval_loss": 0.909953773021698,
+      "eval_runtime": 423.1425,
+      "eval_samples_per_second": 50.151,
+      "eval_steps_per_second": 0.392,
+      "step": 21000
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 10.161764144897461,
+      "learning_rate": 3.2364683247268506e-05,
+      "loss": 1.2178,
+      "step": 21010
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.921660900115967,
+      "learning_rate": 3.2349488524691855e-05,
+      "loss": 1.0175,
+      "step": 21020
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 6.362030029296875,
+      "learning_rate": 3.2334290829744665e-05,
+      "loss": 1.224,
+      "step": 21030
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 7.064016342163086,
+      "learning_rate": 3.231909016857335e-05,
+      "loss": 1.2064,
+      "step": 21040
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 6.590982437133789,
+      "learning_rate": 3.230388654732559e-05,
+      "loss": 1.2402,
+      "step": 21050
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.597902774810791,
+      "learning_rate": 3.228867997215018e-05,
+      "loss": 0.9706,
+      "step": 21060
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 6.809738636016846,
+      "learning_rate": 3.227347044919718e-05,
+      "loss": 1.225,
+      "step": 21070
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 6.783113956451416,
+      "learning_rate": 3.225825798461779e-05,
+      "loss": 1.2031,
+      "step": 21080
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1968164443969727,
+      "learning_rate": 3.224304258456444e-05,
+      "loss": 1.0427,
+      "step": 21090
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.888051986694336,
+      "learning_rate": 3.222782425519071e-05,
+      "loss": 1.1442,
+      "step": 21100
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.595452785491943,
+      "learning_rate": 3.22126030026514e-05,
+      "loss": 1.0922,
+      "step": 21110
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.382979393005371,
+      "learning_rate": 3.219737883310248e-05,
+      "loss": 1.0216,
+      "step": 21120
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.942633152008057,
+      "learning_rate": 3.2182151752701074e-05,
+      "loss": 1.14,
+      "step": 21130
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.8574371337890625,
+      "learning_rate": 3.216692176760552e-05,
+      "loss": 1.2912,
+      "step": 21140
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 6.248178958892822,
+      "learning_rate": 3.215168888397532e-05,
+      "loss": 1.2275,
+      "step": 21150
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 7.462525844573975,
+      "learning_rate": 3.213645310797113e-05,
+      "loss": 1.0915,
+      "step": 21160
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.811999797821045,
+      "learning_rate": 3.21212144457548e-05,
+      "loss": 1.0597,
+      "step": 21170
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.5101189613342285,
+      "learning_rate": 3.210597290348934e-05,
+      "loss": 1.1122,
+      "step": 21180
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 7.957448959350586,
+      "learning_rate": 3.209072848733892e-05,
+      "loss": 0.9938,
+      "step": 21190
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 6.3079705238342285,
+      "learning_rate": 3.207548120346886e-05,
+      "loss": 1.2257,
+      "step": 21200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.911131858825684,
+      "learning_rate": 3.206023105804567e-05,
+      "loss": 0.9406,
+      "step": 21210
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.180386543273926,
+      "learning_rate": 3.204497805723699e-05,
+      "loss": 1.1214,
+      "step": 21220
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 7.178151607513428,
+      "learning_rate": 3.202972220721165e-05,
+      "loss": 1.3259,
+      "step": 21230
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.242873191833496,
+      "learning_rate": 3.201446351413958e-05,
+      "loss": 0.9677,
+      "step": 21240
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 6.060952663421631,
+      "learning_rate": 3.199920198419191e-05,
+      "loss": 1.0428,
+      "step": 21250
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.427149772644043,
+      "learning_rate": 3.1983937623540877e-05,
+      "loss": 1.0296,
+      "step": 21260
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.713008403778076,
+      "learning_rate": 3.1968670438359895e-05,
+      "loss": 1.0906,
+      "step": 21270
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.909712791442871,
+      "learning_rate": 3.1953400434823494e-05,
+      "loss": 1.0355,
+      "step": 21280
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.506474018096924,
+      "learning_rate": 3.193812761910737e-05,
+      "loss": 1.175,
+      "step": 21290
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.022580623626709,
+      "learning_rate": 3.1922851997388334e-05,
+      "loss": 0.7978,
+      "step": 21300
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.540780067443848,
+      "learning_rate": 3.190757357584435e-05,
+      "loss": 0.9821,
+      "step": 21310
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.833585739135742,
+      "learning_rate": 3.18922923606545e-05,
+      "loss": 1.002,
+      "step": 21320
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.8618292808532715,
+      "learning_rate": 3.187700835799899e-05,
+      "loss": 0.9968,
+      "step": 21330
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.432217121124268,
+      "learning_rate": 3.186172157405919e-05,
+      "loss": 1.0482,
+      "step": 21340
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 7.451992988586426,
+      "learning_rate": 3.184643201501754e-05,
+      "loss": 1.0294,
+      "step": 21350
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 6.576083183288574,
+      "learning_rate": 3.183113968705766e-05,
+      "loss": 1.1118,
+      "step": 21360
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 6.387075424194336,
+      "learning_rate": 3.181584459636423e-05,
+      "loss": 1.0532,
+      "step": 21370
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 6.323338031768799,
+      "learning_rate": 3.180054674912311e-05,
+      "loss": 1.1036,
+      "step": 21380
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.137307643890381,
+      "learning_rate": 3.178524615152122e-05,
+      "loss": 1.0352,
+      "step": 21390
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.188426494598389,
+      "learning_rate": 3.176994280974663e-05,
+      "loss": 0.9973,
+      "step": 21400
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.588771343231201,
+      "learning_rate": 3.17546367299885e-05,
+      "loss": 1.1341,
+      "step": 21410
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.626752853393555,
+      "learning_rate": 3.17393279184371e-05,
+      "loss": 1.0873,
+      "step": 21420
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 7.133699893951416,
+      "learning_rate": 3.172401638128382e-05,
+      "loss": 0.9561,
+      "step": 21430
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.1029486656188965,
+      "learning_rate": 3.1708702124721125e-05,
+      "loss": 0.8692,
+      "step": 21440
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.230268478393555,
+      "learning_rate": 3.1693385154942603e-05,
+      "loss": 1.1171,
+      "step": 21450
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 6.337332725524902,
+      "learning_rate": 3.1678065478142924e-05,
+      "loss": 1.234,
+      "step": 21460
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.035854339599609,
+      "learning_rate": 3.1662743100517876e-05,
+      "loss": 1.1916,
+      "step": 21470
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 8.146329879760742,
+      "learning_rate": 3.164741802826431e-05,
+      "loss": 1.1265,
+      "step": 21480
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.540589809417725,
+      "learning_rate": 3.1632090267580186e-05,
+      "loss": 0.9533,
+      "step": 21490
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 6.112457275390625,
+      "learning_rate": 3.161675982466454e-05,
+      "loss": 0.9814,
+      "step": 21500
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.889907956123352,
+      "eval_runtime": 423.3519,
+      "eval_samples_per_second": 50.126,
+      "eval_steps_per_second": 0.392,
+      "step": 21500
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.6886186599731445,
+      "learning_rate": 3.1601426705717504e-05,
+      "loss": 0.9442,
+      "step": 21510
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 6.8706955909729,
+      "learning_rate": 3.158609091694028e-05,
+      "loss": 1.0343,
+      "step": 21520
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 6.704661846160889,
+      "learning_rate": 3.157075246453517e-05,
+      "loss": 1.0022,
+      "step": 21530
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.385902404785156,
+      "learning_rate": 3.155541135470552e-05,
+      "loss": 1.041,
+      "step": 21540
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.963711261749268,
+      "learning_rate": 3.154006759365578e-05,
+      "loss": 0.9071,
+      "step": 21550
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 6.317327976226807,
+      "learning_rate": 3.1524721187591476e-05,
+      "loss": 1.2691,
+      "step": 21560
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 6.511353969573975,
+      "learning_rate": 3.150937214271917e-05,
+      "loss": 0.8812,
+      "step": 21570
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.590428352355957,
+      "learning_rate": 3.1494020465246516e-05,
+      "loss": 1.2435,
+      "step": 21580
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.968519687652588,
+      "learning_rate": 3.1478666161382234e-05,
+      "loss": 1.206,
+      "step": 21590
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.978015899658203,
+      "learning_rate": 3.1463309237336096e-05,
+      "loss": 1.1667,
+      "step": 21600
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.9925689697265625,
+      "learning_rate": 3.144794969931895e-05,
+      "loss": 1.1302,
+      "step": 21610
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 6.255640983581543,
+      "learning_rate": 3.1432587553542664e-05,
+      "loss": 1.3034,
+      "step": 21620
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.81433629989624,
+      "learning_rate": 3.141722280622021e-05,
+      "loss": 1.0502,
+      "step": 21630
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.786376953125,
+      "learning_rate": 3.140185546356557e-05,
+      "loss": 1.1996,
+      "step": 21640
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.25999641418457,
+      "learning_rate": 3.13864855317938e-05,
+      "loss": 0.9497,
+      "step": 21650
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.933833599090576,
+      "learning_rate": 3.137111301712101e-05,
+      "loss": 0.9244,
+      "step": 21660
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.952049732208252,
+      "learning_rate": 3.135573792576433e-05,
+      "loss": 1.0377,
+      "step": 21670
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 7.929764270782471,
+      "learning_rate": 3.1340360263941944e-05,
+      "loss": 1.0928,
+      "step": 21680
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 7.117997646331787,
+      "learning_rate": 3.1324980037873074e-05,
+      "loss": 1.0214,
+      "step": 21690
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 6.464271545410156,
+      "learning_rate": 3.130959725377798e-05,
+      "loss": 1.1581,
+      "step": 21700
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 7.101837158203125,
+      "learning_rate": 3.1294211917877956e-05,
+      "loss": 1.1656,
+      "step": 21710
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.358852386474609,
+      "learning_rate": 3.127882403639534e-05,
+      "loss": 1.0547,
+      "step": 21720
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.918743133544922,
+      "learning_rate": 3.126343361555348e-05,
+      "loss": 1.2658,
+      "step": 21730
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 8.720705032348633,
+      "learning_rate": 3.1248040661576756e-05,
+      "loss": 1.2704,
+      "step": 21740
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 6.142070293426514,
+      "learning_rate": 3.123264518069058e-05,
+      "loss": 1.0327,
+      "step": 21750
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 6.692007541656494,
+      "learning_rate": 3.121724717912138e-05,
+      "loss": 1.0429,
+      "step": 21760
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 6.9781928062438965,
+      "learning_rate": 3.12018466630966e-05,
+      "loss": 1.1784,
+      "step": 21770
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 7.5509796142578125,
+      "learning_rate": 3.118644363884471e-05,
+      "loss": 1.3064,
+      "step": 21780
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.989879608154297,
+      "learning_rate": 3.11710381125952e-05,
+      "loss": 1.033,
+      "step": 21790
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 8.063882827758789,
+      "learning_rate": 3.1155630090578545e-05,
+      "loss": 1.0755,
+      "step": 21800
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 7.077784538269043,
+      "learning_rate": 3.114021957902627e-05,
+      "loss": 1.0925,
+      "step": 21810
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 7.2486395835876465,
+      "learning_rate": 3.112480658417086e-05,
+      "loss": 1.0496,
+      "step": 21820
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 6.045034408569336,
+      "learning_rate": 3.1109391112245844e-05,
+      "loss": 1.1877,
+      "step": 21830
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 6.299131870269775,
+      "learning_rate": 3.1093973169485726e-05,
+      "loss": 1.0999,
+      "step": 21840
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 6.0803542137146,
+      "learning_rate": 3.107855276212603e-05,
+      "loss": 1.2298,
+      "step": 21850
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.020075798034668,
+      "learning_rate": 3.106312989640326e-05,
+      "loss": 1.0,
+      "step": 21860
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.906154155731201,
+      "learning_rate": 3.1047704578554926e-05,
+      "loss": 1.0317,
+      "step": 21870
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.399689197540283,
+      "learning_rate": 3.1032276814819524e-05,
+      "loss": 1.0374,
+      "step": 21880
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.056970119476318,
+      "learning_rate": 3.101684661143653e-05,
+      "loss": 0.8561,
+      "step": 21890
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 7.343419075012207,
+      "learning_rate": 3.100141397464642e-05,
+      "loss": 1.0456,
+      "step": 21900
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.343579292297363,
+      "learning_rate": 3.098597891069066e-05,
+      "loss": 0.9818,
+      "step": 21910
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.233598232269287,
+      "learning_rate": 3.0970541425811685e-05,
+      "loss": 0.9902,
+      "step": 21920
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 6.544045925140381,
+      "learning_rate": 3.095510152625291e-05,
+      "loss": 1.1162,
+      "step": 21930
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 6.728771686553955,
+      "learning_rate": 3.093965921825873e-05,
+      "loss": 1.2518,
+      "step": 21940
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.558053016662598,
+      "learning_rate": 3.092421450807451e-05,
+      "loss": 1.158,
+      "step": 21950
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.78532600402832,
+      "learning_rate": 3.090876740194659e-05,
+      "loss": 1.0531,
+      "step": 21960
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.058490753173828,
+      "learning_rate": 3.089331790612229e-05,
+      "loss": 1.0834,
+      "step": 21970
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.480714797973633,
+      "learning_rate": 3.087786602684987e-05,
+      "loss": 1.1361,
+      "step": 21980
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.6793415546417236,
+      "learning_rate": 3.086241177037858e-05,
+      "loss": 1.0025,
+      "step": 21990
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 6.248135566711426,
+      "learning_rate": 3.084695514295861e-05,
+      "loss": 0.8621,
+      "step": 22000
+    },
+    {
+      "epoch": 0.87,
+      "eval_loss": 0.8591948747634888,
+      "eval_runtime": 423.0965,
+      "eval_samples_per_second": 50.156,
+      "eval_steps_per_second": 0.392,
+      "step": 22000
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.4856109619140625,
+      "learning_rate": 3.083149615084114e-05,
+      "loss": 1.0407,
+      "step": 22010
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.636837005615234,
+      "learning_rate": 3.081603480027826e-05,
+      "loss": 1.0011,
+      "step": 22020
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 7.340846538543701,
+      "learning_rate": 3.080057109752305e-05,
+      "loss": 1.0809,
+      "step": 22030
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.9157421588897705,
+      "learning_rate": 3.0785105048829537e-05,
+      "loss": 0.9145,
+      "step": 22040
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.720076084136963,
+      "learning_rate": 3.0769636660452686e-05,
+      "loss": 0.9346,
+      "step": 22050
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.848022699356079,
+      "learning_rate": 3.0754165938648406e-05,
+      "loss": 1.0307,
+      "step": 22060
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 7.542574882507324,
+      "learning_rate": 3.073869288967356e-05,
+      "loss": 1.089,
+      "step": 22070
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.601290702819824,
+      "learning_rate": 3.072321751978596e-05,
+      "loss": 0.8905,
+      "step": 22080
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 7.06060791015625,
+      "learning_rate": 3.070773983524433e-05,
+      "loss": 1.215,
+      "step": 22090
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.848817348480225,
+      "learning_rate": 3.0692259842308345e-05,
+      "loss": 0.8913,
+      "step": 22100
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.71605110168457,
+      "learning_rate": 3.0676777547238634e-05,
+      "loss": 1.1818,
+      "step": 22110
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.8453891277313232,
+      "learning_rate": 3.066129295629672e-05,
+      "loss": 0.8754,
+      "step": 22120
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 6.165750980377197,
+      "learning_rate": 3.064580607574508e-05,
+      "loss": 1.0401,
+      "step": 22130
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 6.419191837310791,
+      "learning_rate": 3.063031691184711e-05,
+      "loss": 1.195,
+      "step": 22140
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.988631248474121,
+      "learning_rate": 3.061482547086712e-05,
+      "loss": 0.9497,
+      "step": 22150
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 6.101799011230469,
+      "learning_rate": 3.0599331759070346e-05,
+      "loss": 0.9741,
+      "step": 22160
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 6.259200572967529,
+      "learning_rate": 3.058383578272296e-05,
+      "loss": 1.1817,
+      "step": 22170
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.6100852489471436,
+      "learning_rate": 3.0568337548092026e-05,
+      "loss": 0.9269,
+      "step": 22180
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 8.319791793823242,
+      "learning_rate": 3.055283706144553e-05,
+      "loss": 1.0099,
+      "step": 22190
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.644494533538818,
+      "learning_rate": 3.053733432905237e-05,
+      "loss": 0.9535,
+      "step": 22200
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.23773717880249,
+      "learning_rate": 3.052182935718235e-05,
+      "loss": 1.0283,
+      "step": 22210
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.452188014984131,
+      "learning_rate": 3.0506322152106192e-05,
+      "loss": 1.0248,
+      "step": 22220
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.8521552085876465,
+      "learning_rate": 3.04908127200955e-05,
+      "loss": 1.1009,
+      "step": 22230
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 9.736371994018555,
+      "learning_rate": 3.0475301067422795e-05,
+      "loss": 1.442,
+      "step": 22240
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 6.1204376220703125,
+      "learning_rate": 3.0459787200361483e-05,
+      "loss": 1.1315,
+      "step": 22250
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.582522392272949,
+      "learning_rate": 3.0444271125185876e-05,
+      "loss": 1.1421,
+      "step": 22260
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 6.34846830368042,
+      "learning_rate": 3.0428752848171188e-05,
+      "loss": 1.1799,
+      "step": 22270
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 6.126006603240967,
+      "learning_rate": 3.0413232375593497e-05,
+      "loss": 1.139,
+      "step": 22280
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.371376037597656,
+      "learning_rate": 3.0397709713729782e-05,
+      "loss": 0.9661,
+      "step": 22290
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.749190092086792,
+      "learning_rate": 3.038218486885792e-05,
+      "loss": 1.0492,
+      "step": 22300
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.900237560272217,
+      "learning_rate": 3.0366657847256653e-05,
+      "loss": 1.1914,
+      "step": 22310
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.922823429107666,
+      "learning_rate": 3.0351128655205614e-05,
+      "loss": 1.0599,
+      "step": 22320
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.98379373550415,
+      "learning_rate": 3.0335597298985303e-05,
+      "loss": 1.2475,
+      "step": 22330
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 6.054649353027344,
+      "learning_rate": 3.0320063784877107e-05,
+      "loss": 1.1106,
+      "step": 22340
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 7.189651966094971,
+      "learning_rate": 3.030452811916328e-05,
+      "loss": 0.9569,
+      "step": 22350
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.836315155029297,
+      "learning_rate": 3.0288990308126947e-05,
+      "loss": 0.8587,
+      "step": 22360
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.376640796661377,
+      "learning_rate": 3.0273450358052103e-05,
+      "loss": 1.1984,
+      "step": 22370
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 6.802809715270996,
+      "learning_rate": 3.0257908275223602e-05,
+      "loss": 1.0393,
+      "step": 22380
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.971006870269775,
+      "learning_rate": 3.0242364065927165e-05,
+      "loss": 1.0836,
+      "step": 22390
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 7.505317687988281,
+      "learning_rate": 3.022681773644937e-05,
+      "loss": 1.1302,
+      "step": 22400
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.797395706176758,
+      "learning_rate": 3.021126929307766e-05,
+      "loss": 0.9037,
+      "step": 22410
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 6.912562370300293,
+      "learning_rate": 3.0195718742100333e-05,
+      "loss": 1.0147,
+      "step": 22420
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 6.876853942871094,
+      "learning_rate": 3.0180166089806516e-05,
+      "loss": 0.9852,
+      "step": 22430
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.871789455413818,
+      "learning_rate": 3.016461134248622e-05,
+      "loss": 1.1026,
+      "step": 22440
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 7.126190662384033,
+      "learning_rate": 3.014905450643028e-05,
+      "loss": 1.2575,
+      "step": 22450
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 6.332014560699463,
+      "learning_rate": 3.013349558793039e-05,
+      "loss": 0.9037,
+      "step": 22460
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.1320977210998535,
+      "learning_rate": 3.0117934593279062e-05,
+      "loss": 0.9718,
+      "step": 22470
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.859474182128906,
+      "learning_rate": 3.010237152876969e-05,
+      "loss": 1.0757,
+      "step": 22480
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 9.253780364990234,
+      "learning_rate": 3.008680640069646e-05,
+      "loss": 1.0582,
+      "step": 22490
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.635983943939209,
+      "learning_rate": 3.0071239215354418e-05,
+      "loss": 0.9599,
+      "step": 22500
+    },
+    {
+      "epoch": 0.89,
+      "eval_loss": 0.8504640460014343,
+      "eval_runtime": 423.5521,
+      "eval_samples_per_second": 50.102,
+      "eval_steps_per_second": 0.392,
+      "step": 22500
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.691872596740723,
+      "learning_rate": 3.005566997903944e-05,
+      "loss": 1.0123,
+      "step": 22510
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 6.324728012084961,
+      "learning_rate": 3.004009869804823e-05,
+      "loss": 1.0119,
+      "step": 22520
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 6.235013961791992,
+      "learning_rate": 3.002452537867831e-05,
+      "loss": 0.9155,
+      "step": 22530
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5683515071868896,
+      "learning_rate": 3.0008950027228033e-05,
+      "loss": 0.8382,
+      "step": 22540
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 7.953617095947266,
+      "learning_rate": 2.999337264999658e-05,
+      "loss": 1.0642,
+      "step": 22550
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.907096862792969,
+      "learning_rate": 2.9977793253283943e-05,
+      "loss": 1.1446,
+      "step": 22560
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.287431716918945,
+      "learning_rate": 2.9962211843390935e-05,
+      "loss": 1.1547,
+      "step": 22570
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.998079776763916,
+      "learning_rate": 2.9946628426619178e-05,
+      "loss": 0.9751,
+      "step": 22580
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.072854995727539,
+      "learning_rate": 2.993104300927111e-05,
+      "loss": 1.0368,
+      "step": 22590
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.493224143981934,
+      "learning_rate": 2.9915455597649977e-05,
+      "loss": 1.0431,
+      "step": 22600
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.0673604011535645,
+      "learning_rate": 2.9899866198059835e-05,
+      "loss": 1.0479,
+      "step": 22610
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.113644599914551,
+      "learning_rate": 2.988427481680554e-05,
+      "loss": 1.1038,
+      "step": 22620
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.650437831878662,
+      "learning_rate": 2.9868681460192745e-05,
+      "loss": 1.0209,
+      "step": 22630
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.620076656341553,
+      "learning_rate": 2.9853086134527913e-05,
+      "loss": 0.9421,
+      "step": 22640
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.25744104385376,
+      "learning_rate": 2.9837488846118295e-05,
+      "loss": 0.8188,
+      "step": 22650
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.3469038009643555,
+      "learning_rate": 2.9821889601271935e-05,
+      "loss": 1.1037,
+      "step": 22660
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.8933563232421875,
+      "learning_rate": 2.9806288406297676e-05,
+      "loss": 1.1282,
+      "step": 22670
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.419617652893066,
+      "learning_rate": 2.9790685267505146e-05,
+      "loss": 1.2271,
+      "step": 22680
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.185498237609863,
+      "learning_rate": 2.9775080191204756e-05,
+      "loss": 1.0469,
+      "step": 22690
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.537951707839966,
+      "learning_rate": 2.975947318370771e-05,
+      "loss": 0.9825,
+      "step": 22700
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 7.60770845413208,
+      "learning_rate": 2.9743864251325965e-05,
+      "loss": 1.0434,
+      "step": 22710
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.987367630004883,
+      "learning_rate": 2.9728253400372297e-05,
+      "loss": 1.0208,
+      "step": 22720
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.727715969085693,
+      "learning_rate": 2.9712640637160233e-05,
+      "loss": 1.1833,
+      "step": 22730
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.151361465454102,
+      "learning_rate": 2.9697025968004077e-05,
+      "loss": 1.0933,
+      "step": 22740
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.213372230529785,
+      "learning_rate": 2.968140939921891e-05,
+      "loss": 0.8194,
+      "step": 22750
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.646399021148682,
+      "learning_rate": 2.9665790937120575e-05,
+      "loss": 1.0136,
+      "step": 22760
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.7873430252075195,
+      "learning_rate": 2.9650170588025682e-05,
+      "loss": 1.053,
+      "step": 22770
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.841545581817627,
+      "learning_rate": 2.9634548358251603e-05,
+      "loss": 1.1731,
+      "step": 22780
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 8.664173126220703,
+      "learning_rate": 2.961892425411647e-05,
+      "loss": 0.9024,
+      "step": 22790
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.270088195800781,
+      "learning_rate": 2.9603298281939178e-05,
+      "loss": 1.1785,
+      "step": 22800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.5727858543396,
+      "learning_rate": 2.9587670448039385e-05,
+      "loss": 1.0825,
+      "step": 22810
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.703754425048828,
+      "learning_rate": 2.9572040758737474e-05,
+      "loss": 1.1539,
+      "step": 22820
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.344325304031372,
+      "learning_rate": 2.9556409220354613e-05,
+      "loss": 1.0239,
+      "step": 22830
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 6.108560085296631,
+      "learning_rate": 2.9540775839212687e-05,
+      "loss": 1.0691,
+      "step": 22840
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 6.379608154296875,
+      "learning_rate": 2.9525140621634345e-05,
+      "loss": 1.0678,
+      "step": 22850
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.147445201873779,
+      "learning_rate": 2.9509503573942977e-05,
+      "loss": 0.9371,
+      "step": 22860
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 6.11639404296875,
+      "learning_rate": 2.9493864702462703e-05,
+      "loss": 1.2253,
+      "step": 22870
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 7.699833393096924,
+      "learning_rate": 2.9478224013518396e-05,
+      "loss": 0.9342,
+      "step": 22880
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 6.273466110229492,
+      "learning_rate": 2.9462581513435654e-05,
+      "loss": 1.0323,
+      "step": 22890
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 6.1669182777404785,
+      "learning_rate": 2.9446937208540808e-05,
+      "loss": 1.112,
+      "step": 22900
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.359527111053467,
+      "learning_rate": 2.943129110516092e-05,
+      "loss": 1.2383,
+      "step": 22910
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 6.280796527862549,
+      "learning_rate": 2.941564320962378e-05,
+      "loss": 0.915,
+      "step": 22920
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 6.141207695007324,
+      "learning_rate": 2.9399993528257902e-05,
+      "loss": 1.1245,
+      "step": 22930
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 7.9414849281311035,
+      "learning_rate": 2.9384342067392524e-05,
+      "loss": 1.2677,
+      "step": 22940
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.869688034057617,
+      "learning_rate": 2.93686888333576e-05,
+      "loss": 0.9457,
+      "step": 22950
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 9.19936466217041,
+      "learning_rate": 2.9353033832483806e-05,
+      "loss": 1.0445,
+      "step": 22960
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 7.491847991943359,
+      "learning_rate": 2.933737707110253e-05,
+      "loss": 0.9163,
+      "step": 22970
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 6.724958419799805,
+      "learning_rate": 2.9321718555545867e-05,
+      "loss": 1.164,
+      "step": 22980
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 8.466373443603516,
+      "learning_rate": 2.9306058292146633e-05,
+      "loss": 1.1343,
+      "step": 22990
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 6.593084335327148,
+      "learning_rate": 2.9290396287238338e-05,
+      "loss": 1.0433,
+      "step": 23000
+    },
+    {
+      "epoch": 0.91,
+      "eval_loss": 0.8658052682876587,
+      "eval_runtime": 423.2708,
+      "eval_samples_per_second": 50.136,
+      "eval_steps_per_second": 0.392,
+      "step": 23000
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 6.445999622344971,
+      "learning_rate": 2.927473254715521e-05,
+      "loss": 1.2697,
+      "step": 23010
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.005995273590088,
+      "learning_rate": 2.9259067078232155e-05,
+      "loss": 1.0715,
+      "step": 23020
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.656961441040039,
+      "learning_rate": 2.9243399886804816e-05,
+      "loss": 1.0149,
+      "step": 23030
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.443883895874023,
+      "learning_rate": 2.9227730979209505e-05,
+      "loss": 1.0117,
+      "step": 23040
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.457942962646484,
+      "learning_rate": 2.9212060361783227e-05,
+      "loss": 1.2611,
+      "step": 23050
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 7.557255268096924,
+      "learning_rate": 2.9196388040863693e-05,
+      "loss": 1.1329,
+      "step": 23060
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.439605712890625,
+      "learning_rate": 2.9180714022789298e-05,
+      "loss": 1.158,
+      "step": 23070
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.7694597244262695,
+      "learning_rate": 2.9165038313899108e-05,
+      "loss": 1.122,
+      "step": 23080
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.453972816467285,
+      "learning_rate": 2.9149360920532896e-05,
+      "loss": 1.0408,
+      "step": 23090
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.740454196929932,
+      "learning_rate": 2.9133681849031115e-05,
+      "loss": 1.0582,
+      "step": 23100
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.465270519256592,
+      "learning_rate": 2.911800110573487e-05,
+      "loss": 1.0999,
+      "step": 23110
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.310168266296387,
+      "learning_rate": 2.910231869698597e-05,
+      "loss": 1.0576,
+      "step": 23120
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 13.463846206665039,
+      "learning_rate": 2.908663462912688e-05,
+      "loss": 1.1267,
+      "step": 23130
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.4687886238098145,
+      "learning_rate": 2.907094890850075e-05,
+      "loss": 1.1986,
+      "step": 23140
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.365592002868652,
+      "learning_rate": 2.9055261541451394e-05,
+      "loss": 1.0756,
+      "step": 23150
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.876799583435059,
+      "learning_rate": 2.903957253432329e-05,
+      "loss": 1.0403,
+      "step": 23160
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.1549601554870605,
+      "learning_rate": 2.9023881893461564e-05,
+      "loss": 1.0264,
+      "step": 23170
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 7.506237506866455,
+      "learning_rate": 2.9008189625212035e-05,
+      "loss": 0.9861,
+      "step": 23180
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.614490985870361,
+      "learning_rate": 2.8992495735921165e-05,
+      "loss": 1.0764,
+      "step": 23190
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.873629570007324,
+      "learning_rate": 2.897680023193606e-05,
+      "loss": 1.0816,
+      "step": 23200
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.994396686553955,
+      "learning_rate": 2.896110311960449e-05,
+      "loss": 1.1313,
+      "step": 23210
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.632668972015381,
+      "learning_rate": 2.8945404405274877e-05,
+      "loss": 1.1192,
+      "step": 23220
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 7.808846950531006,
+      "learning_rate": 2.8929704095296295e-05,
+      "loss": 1.0383,
+      "step": 23230
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.517245769500732,
+      "learning_rate": 2.891400219601844e-05,
+      "loss": 0.9253,
+      "step": 23240
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.232921123504639,
+      "learning_rate": 2.889829871379168e-05,
+      "loss": 0.9709,
+      "step": 23250
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.299297332763672,
+      "learning_rate": 2.8882593654967e-05,
+      "loss": 1.2047,
+      "step": 23260
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.690835475921631,
+      "learning_rate": 2.8866887025896044e-05,
+      "loss": 0.8339,
+      "step": 23270
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.03762674331665,
+      "learning_rate": 2.8851178832931074e-05,
+      "loss": 1.0892,
+      "step": 23280
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.220973014831543,
+      "learning_rate": 2.8835469082424993e-05,
+      "loss": 1.0665,
+      "step": 23290
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.758197784423828,
+      "learning_rate": 2.8819757780731325e-05,
+      "loss": 1.1177,
+      "step": 23300
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.888477802276611,
+      "learning_rate": 2.8804044934204243e-05,
+      "loss": 1.1363,
+      "step": 23310
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.058335304260254,
+      "learning_rate": 2.878833054919851e-05,
+      "loss": 0.8217,
+      "step": 23320
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.3837103843688965,
+      "learning_rate": 2.877261463206955e-05,
+      "loss": 0.9303,
+      "step": 23330
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.79655122756958,
+      "learning_rate": 2.875689718917337e-05,
+      "loss": 0.9166,
+      "step": 23340
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.885788440704346,
+      "learning_rate": 2.8741178226866626e-05,
+      "loss": 0.9216,
+      "step": 23350
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.639639854431152,
+      "learning_rate": 2.8725457751506574e-05,
+      "loss": 0.8781,
+      "step": 23360
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 6.610500812530518,
+      "learning_rate": 2.8709735769451075e-05,
+      "loss": 1.0288,
+      "step": 23370
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 6.682758808135986,
+      "learning_rate": 2.86940122870586e-05,
+      "loss": 0.9531,
+      "step": 23380
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 7.377542495727539,
+      "learning_rate": 2.8678287310688247e-05,
+      "loss": 1.1259,
+      "step": 23390
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 7.524086952209473,
+      "learning_rate": 2.8662560846699694e-05,
+      "loss": 1.3314,
+      "step": 23400
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 6.151552677154541,
+      "learning_rate": 2.864683290145323e-05,
+      "loss": 1.0736,
+      "step": 23410
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.9615020751953125,
+      "learning_rate": 2.8631103481309758e-05,
+      "loss": 0.9964,
+      "step": 23420
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 6.33573055267334,
+      "learning_rate": 2.8615372592630753e-05,
+      "loss": 1.256,
+      "step": 23430
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.995708703994751,
+      "learning_rate": 2.8599640241778293e-05,
+      "loss": 1.032,
+      "step": 23440
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 6.893513202667236,
+      "learning_rate": 2.8583906435115047e-05,
+      "loss": 0.9443,
+      "step": 23450
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 6.0982666015625,
+      "learning_rate": 2.856817117900429e-05,
+      "loss": 0.94,
+      "step": 23460
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.435002326965332,
+      "learning_rate": 2.8552434479809854e-05,
+      "loss": 0.9383,
+      "step": 23470
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.177101135253906,
+      "learning_rate": 2.8536696343896162e-05,
+      "loss": 1.0555,
+      "step": 23480
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.778027534484863,
+      "learning_rate": 2.8520956777628253e-05,
+      "loss": 0.8263,
+      "step": 23490
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 6.134400844573975,
+      "learning_rate": 2.8505215787371687e-05,
+      "loss": 0.9788,
+      "step": 23500
+    },
+    {
+      "epoch": 0.93,
+      "eval_loss": 0.8511847853660583,
+      "eval_runtime": 423.6489,
+      "eval_samples_per_second": 50.091,
+      "eval_steps_per_second": 0.392,
+      "step": 23500
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.521714210510254,
+      "learning_rate": 2.8489473379492643e-05,
+      "loss": 0.9882,
+      "step": 23510
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 6.953526973724365,
+      "learning_rate": 2.8473729560357852e-05,
+      "loss": 1.0879,
+      "step": 23520
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.372171401977539,
+      "learning_rate": 2.8457984336334625e-05,
+      "loss": 1.1522,
+      "step": 23530
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 6.562096118927002,
+      "learning_rate": 2.844223771379084e-05,
+      "loss": 1.2368,
+      "step": 23540
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.493814468383789,
+      "learning_rate": 2.842648969909495e-05,
+      "loss": 0.9871,
+      "step": 23550
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.67075252532959,
+      "learning_rate": 2.841074029861594e-05,
+      "loss": 0.9958,
+      "step": 23560
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.682605743408203,
+      "learning_rate": 2.8394989518723392e-05,
+      "loss": 0.7843,
+      "step": 23570
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 9.357209205627441,
+      "learning_rate": 2.8379237365787426e-05,
+      "loss": 1.15,
+      "step": 23580
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.497167110443115,
+      "learning_rate": 2.836348384617872e-05,
+      "loss": 1.1971,
+      "step": 23590
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 7.113755702972412,
+      "learning_rate": 2.8347728966268504e-05,
+      "loss": 1.2034,
+      "step": 23600
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 7.123961448669434,
+      "learning_rate": 2.833197273242857e-05,
+      "loss": 1.0299,
+      "step": 23610
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.0889763832092285,
+      "learning_rate": 2.831621515103124e-05,
+      "loss": 0.9152,
+      "step": 23620
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.366820335388184,
+      "learning_rate": 2.830045622844939e-05,
+      "loss": 0.983,
+      "step": 23630
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 6.343715667724609,
+      "learning_rate": 2.8284695971056436e-05,
+      "loss": 1.0513,
+      "step": 23640
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 7.200556755065918,
+      "learning_rate": 2.826893438522633e-05,
+      "loss": 1.1471,
+      "step": 23650
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 6.461779594421387,
+      "learning_rate": 2.8253171477333586e-05,
+      "loss": 1.3019,
+      "step": 23660
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.0565104484558105,
+      "learning_rate": 2.8237407253753213e-05,
+      "loss": 1.0261,
+      "step": 23670
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 6.044254302978516,
+      "learning_rate": 2.822164172086078e-05,
+      "loss": 0.9178,
+      "step": 23680
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.329671859741211,
+      "learning_rate": 2.820587488503238e-05,
+      "loss": 1.0119,
+      "step": 23690
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.487349510192871,
+      "learning_rate": 2.8190106752644623e-05,
+      "loss": 0.9167,
+      "step": 23700
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 7.3770432472229,
+      "learning_rate": 2.817433733007466e-05,
+      "loss": 1.1922,
+      "step": 23710
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.085980415344238,
+      "learning_rate": 2.8158566623700143e-05,
+      "loss": 0.9954,
+      "step": 23720
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 6.528548717498779,
+      "learning_rate": 2.814279463989927e-05,
+      "loss": 0.9675,
+      "step": 23730
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.6550452709198,
+      "learning_rate": 2.8127021385050733e-05,
+      "loss": 1.1213,
+      "step": 23740
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 9.975143432617188,
+      "learning_rate": 2.811124686553374e-05,
+      "loss": 1.0368,
+      "step": 23750
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.812924861907959,
+      "learning_rate": 2.8095471087728032e-05,
+      "loss": 0.8878,
+      "step": 23760
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.118997097015381,
+      "learning_rate": 2.8079694058013827e-05,
+      "loss": 1.0688,
+      "step": 23770
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.507280349731445,
+      "learning_rate": 2.8063915782771873e-05,
+      "loss": 1.0,
+      "step": 23780
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.998612403869629,
+      "learning_rate": 2.804813626838341e-05,
+      "loss": 0.9812,
+      "step": 23790
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.4506378173828125,
+      "learning_rate": 2.803235552123019e-05,
+      "loss": 0.9507,
+      "step": 23800
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.117337226867676,
+      "learning_rate": 2.8016573547694456e-05,
+      "loss": 1.0762,
+      "step": 23810
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.5635762214660645,
+      "learning_rate": 2.800079035415894e-05,
+      "loss": 1.0643,
+      "step": 23820
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.455548286437988,
+      "learning_rate": 2.798500594700688e-05,
+      "loss": 1.1218,
+      "step": 23830
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.004398345947266,
+      "learning_rate": 2.7969220332622003e-05,
+      "loss": 1.1033,
+      "step": 23840
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.115938186645508,
+      "learning_rate": 2.7953433517388523e-05,
+      "loss": 1.0689,
+      "step": 23850
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.625364303588867,
+      "learning_rate": 2.7937645507691123e-05,
+      "loss": 0.9577,
+      "step": 23860
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.219773292541504,
+      "learning_rate": 2.7921856309915e-05,
+      "loss": 1.0248,
+      "step": 23870
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.402790546417236,
+      "learning_rate": 2.7906065930445812e-05,
+      "loss": 0.7878,
+      "step": 23880
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.68462610244751,
+      "learning_rate": 2.7890274375669695e-05,
+      "loss": 1.0298,
+      "step": 23890
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.162065505981445,
+      "learning_rate": 2.7874481651973267e-05,
+      "loss": 1.015,
+      "step": 23900
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 7.081689834594727,
+      "learning_rate": 2.7858687765743615e-05,
+      "loss": 1.0656,
+      "step": 23910
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.693350791931152,
+      "learning_rate": 2.784289272336829e-05,
+      "loss": 0.9774,
+      "step": 23920
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.156042098999023,
+      "learning_rate": 2.7827096531235326e-05,
+      "loss": 0.817,
+      "step": 23930
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.233408451080322,
+      "learning_rate": 2.781129919573321e-05,
+      "loss": 1.348,
+      "step": 23940
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.981807231903076,
+      "learning_rate": 2.7795500723250895e-05,
+      "loss": 1.0813,
+      "step": 23950
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 7.200736045837402,
+      "learning_rate": 2.7779701120177797e-05,
+      "loss": 0.9782,
+      "step": 23960
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.438696384429932,
+      "learning_rate": 2.776390039290378e-05,
+      "loss": 1.0241,
+      "step": 23970
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 7.832723617553711,
+      "learning_rate": 2.7748098547819173e-05,
+      "loss": 0.9571,
+      "step": 23980
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.44420862197876,
+      "learning_rate": 2.773229559131475e-05,
+      "loss": 0.9868,
+      "step": 23990
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.3921737670898438,
+      "learning_rate": 2.7716491529781734e-05,
+      "loss": 0.7736,
+      "step": 24000
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 0.827427089214325,
+      "eval_runtime": 423.4834,
+      "eval_samples_per_second": 50.111,
+      "eval_steps_per_second": 0.392,
+      "step": 24000
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.789316177368164,
+      "learning_rate": 2.7700686369611807e-05,
+      "loss": 0.8924,
+      "step": 24010
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 7.132543563842773,
+      "learning_rate": 2.7684880117197076e-05,
+      "loss": 0.9204,
+      "step": 24020
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 7.354853630065918,
+      "learning_rate": 2.7669072778930105e-05,
+      "loss": 1.1944,
+      "step": 24030
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.282512903213501,
+      "learning_rate": 2.7653264361203895e-05,
+      "loss": 0.9165,
+      "step": 24040
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 8.07840633392334,
+      "learning_rate": 2.763745487041188e-05,
+      "loss": 0.9985,
+      "step": 24050
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.335934638977051,
+      "learning_rate": 2.7621644312947924e-05,
+      "loss": 1.0285,
+      "step": 24060
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.617362976074219,
+      "learning_rate": 2.760583269520633e-05,
+      "loss": 1.0075,
+      "step": 24070
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 6.323548793792725,
+      "learning_rate": 2.7590020023581824e-05,
+      "loss": 0.9737,
+      "step": 24080
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 6.636780738830566,
+      "learning_rate": 2.7574206304469565e-05,
+      "loss": 1.1059,
+      "step": 24090
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.078829765319824,
+      "learning_rate": 2.755839154426513e-05,
+      "loss": 1.089,
+      "step": 24100
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 7.382745742797852,
+      "learning_rate": 2.7542575749364513e-05,
+      "loss": 1.0648,
+      "step": 24110
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 7.366662979125977,
+      "learning_rate": 2.7526758926164148e-05,
+      "loss": 1.1096,
+      "step": 24120
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 6.612687110900879,
+      "learning_rate": 2.751094108106086e-05,
+      "loss": 0.9943,
+      "step": 24130
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.2188568115234375,
+      "learning_rate": 2.7495122220451892e-05,
+      "loss": 1.1958,
+      "step": 24140
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.635903835296631,
+      "learning_rate": 2.7479302350734902e-05,
+      "loss": 1.0704,
+      "step": 24150
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.741935729980469,
+      "learning_rate": 2.746348147830797e-05,
+      "loss": 1.0445,
+      "step": 24160
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.011616230010986,
+      "learning_rate": 2.7447659609569554e-05,
+      "loss": 0.7718,
+      "step": 24170
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 7.993719100952148,
+      "learning_rate": 2.7431836750918532e-05,
+      "loss": 1.0155,
+      "step": 24180
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.438990592956543,
+      "learning_rate": 2.741601290875419e-05,
+      "loss": 0.9026,
+      "step": 24190
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 7.700430393218994,
+      "learning_rate": 2.740018808947618e-05,
+      "loss": 1.2059,
+      "step": 24200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 6.956786632537842,
+      "learning_rate": 2.738436229948459e-05,
+      "loss": 1.1834,
+      "step": 24210
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.293816089630127,
+      "learning_rate": 2.7368535545179864e-05,
+      "loss": 0.9953,
+      "step": 24220
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.7796735763549805,
+      "learning_rate": 2.7352707832962865e-05,
+      "loss": 1.1763,
+      "step": 24230
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.533602237701416,
+      "learning_rate": 2.733687916923482e-05,
+      "loss": 0.9252,
+      "step": 24240
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.062287330627441,
+      "learning_rate": 2.732104956039736e-05,
+      "loss": 0.8569,
+      "step": 24250
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.258227825164795,
+      "learning_rate": 2.7305219012852494e-05,
+      "loss": 1.0029,
+      "step": 24260
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.090412139892578,
+      "learning_rate": 2.7289387533002598e-05,
+      "loss": 1.0225,
+      "step": 24270
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.199697494506836,
+      "learning_rate": 2.727355512725044e-05,
+      "loss": 1.009,
+      "step": 24280
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.499246835708618,
+      "learning_rate": 2.7257721801999152e-05,
+      "loss": 0.8921,
+      "step": 24290
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.5831828117370605,
+      "learning_rate": 2.7241887563652246e-05,
+      "loss": 1.0232,
+      "step": 24300
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 6.164170265197754,
+      "learning_rate": 2.7226052418613595e-05,
+      "loss": 0.983,
+      "step": 24310
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.465358257293701,
+      "learning_rate": 2.721021637328745e-05,
+      "loss": 1.1401,
+      "step": 24320
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 8.392667770385742,
+      "learning_rate": 2.719437943407841e-05,
+      "loss": 1.0317,
+      "step": 24330
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3071768283843994,
+      "learning_rate": 2.7178541607391446e-05,
+      "loss": 0.8796,
+      "step": 24340
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.605903625488281,
+      "learning_rate": 2.7162702899631893e-05,
+      "loss": 1.1653,
+      "step": 24350
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 10.53249740600586,
+      "learning_rate": 2.7146863317205427e-05,
+      "loss": 1.041,
+      "step": 24360
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 7.232144832611084,
+      "learning_rate": 2.7131022866518102e-05,
+      "loss": 0.9848,
+      "step": 24370
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 7.997378826141357,
+      "learning_rate": 2.7115181553976294e-05,
+      "loss": 1.1404,
+      "step": 24380
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.986005783081055,
+      "learning_rate": 2.7099339385986737e-05,
+      "loss": 1.1268,
+      "step": 24390
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.5063886642456055,
+      "learning_rate": 2.7083496368956535e-05,
+      "loss": 1.1118,
+      "step": 24400
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.6564483642578125,
+      "learning_rate": 2.7067652509293105e-05,
+      "loss": 0.9306,
+      "step": 24410
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.5761799812316895,
+      "learning_rate": 2.705180781340421e-05,
+      "loss": 1.1194,
+      "step": 24420
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 7.7314887046813965,
+      "learning_rate": 2.703596228769797e-05,
+      "loss": 1.2153,
+      "step": 24430
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.181602954864502,
+      "learning_rate": 2.7020115938582813e-05,
+      "loss": 0.7751,
+      "step": 24440
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.507518768310547,
+      "learning_rate": 2.7004268772467533e-05,
+      "loss": 1.0259,
+      "step": 24450
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.449952602386475,
+      "learning_rate": 2.6988420795761215e-05,
+      "loss": 0.8985,
+      "step": 24460
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.6933369636535645,
+      "learning_rate": 2.6972572014873306e-05,
+      "loss": 1.0284,
+      "step": 24470
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 7.7537841796875,
+      "learning_rate": 2.6956722436213554e-05,
+      "loss": 1.235,
+      "step": 24480
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.253195285797119,
+      "learning_rate": 2.6940872066192052e-05,
+      "loss": 0.987,
+      "step": 24490
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.385574817657471,
+      "learning_rate": 2.6925020911219195e-05,
+      "loss": 1.117,
+      "step": 24500
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 0.8338937163352966,
+      "eval_runtime": 423.6107,
+      "eval_samples_per_second": 50.096,
+      "eval_steps_per_second": 0.392,
+      "step": 24500
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 7.15852165222168,
+      "learning_rate": 2.6909168977705705e-05,
+      "loss": 1.0776,
+      "step": 24510
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.108470439910889,
+      "learning_rate": 2.6893316272062615e-05,
+      "loss": 0.8403,
+      "step": 24520
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.16189432144165,
+      "learning_rate": 2.687746280070127e-05,
+      "loss": 0.9358,
+      "step": 24530
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 7.410829544067383,
+      "learning_rate": 2.686160857003332e-05,
+      "loss": 1.1286,
+      "step": 24540
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.6660923957824707,
+      "learning_rate": 2.6845753586470747e-05,
+      "loss": 0.9658,
+      "step": 24550
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.532138824462891,
+      "learning_rate": 2.68298978564258e-05,
+      "loss": 1.0845,
+      "step": 24560
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.186542510986328,
+      "learning_rate": 2.6814041386311056e-05,
+      "loss": 0.9221,
+      "step": 24570
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.2704243659973145,
+      "learning_rate": 2.6798184182539388e-05,
+      "loss": 0.9736,
+      "step": 24580
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 7.539289474487305,
+      "learning_rate": 2.6782326251523947e-05,
+      "loss": 1.0909,
+      "step": 24590
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.351105213165283,
+      "learning_rate": 2.6766467599678207e-05,
+      "loss": 0.9931,
+      "step": 24600
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.677516460418701,
+      "learning_rate": 2.6750608233415913e-05,
+      "loss": 0.9594,
+      "step": 24610
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 7.198444366455078,
+      "learning_rate": 2.6734748159151102e-05,
+      "loss": 0.9686,
+      "step": 24620
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.8082363605499268,
+      "learning_rate": 2.6718887383298106e-05,
+      "loss": 1.097,
+      "step": 24630
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 7.070648670196533,
+      "learning_rate": 2.6703025912271533e-05,
+      "loss": 1.0404,
+      "step": 24640
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.771407604217529,
+      "learning_rate": 2.6687163752486272e-05,
+      "loss": 1.0464,
+      "step": 24650
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.672630310058594,
+      "learning_rate": 2.66713009103575e-05,
+      "loss": 0.9453,
+      "step": 24660
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.519309997558594,
+      "learning_rate": 2.6655437392300647e-05,
+      "loss": 0.8772,
+      "step": 24670
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 7.36231803894043,
+      "learning_rate": 2.663957320473145e-05,
+      "loss": 0.8882,
+      "step": 24680
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 7.726024627685547,
+      "learning_rate": 2.662370835406589e-05,
+      "loss": 0.9343,
+      "step": 24690
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 7.697350025177002,
+      "learning_rate": 2.660784284672022e-05,
+      "loss": 1.0461,
+      "step": 24700
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.224346160888672,
+      "learning_rate": 2.6591976689110976e-05,
+      "loss": 0.8042,
+      "step": 24710
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.329206943511963,
+      "learning_rate": 2.6576109887654942e-05,
+      "loss": 0.9287,
+      "step": 24720
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.652462005615234,
+      "learning_rate": 2.6560242448769164e-05,
+      "loss": 1.0605,
+      "step": 24730
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.980512619018555,
+      "learning_rate": 2.6544374378870934e-05,
+      "loss": 1.1796,
+      "step": 24740
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.899630069732666,
+      "learning_rate": 2.652850568437783e-05,
+      "loss": 1.1829,
+      "step": 24750
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.663119316101074,
+      "learning_rate": 2.6512636371707655e-05,
+      "loss": 0.859,
+      "step": 24760
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.833503246307373,
+      "learning_rate": 2.649676644727847e-05,
+      "loss": 0.8409,
+      "step": 24770
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.021016597747803,
+      "learning_rate": 2.648089591750859e-05,
+      "loss": 0.8664,
+      "step": 24780
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.999817371368408,
+      "learning_rate": 2.6465024788816573e-05,
+      "loss": 1.1146,
+      "step": 24790
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 7.344309329986572,
+      "learning_rate": 2.6449153067621207e-05,
+      "loss": 1.1525,
+      "step": 24800
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.338775157928467,
+      "learning_rate": 2.6433280760341532e-05,
+      "loss": 0.995,
+      "step": 24810
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.216484069824219,
+      "learning_rate": 2.6417407873396825e-05,
+      "loss": 0.936,
+      "step": 24820
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 6.019948959350586,
+      "learning_rate": 2.6401534413206596e-05,
+      "loss": 0.8391,
+      "step": 24830
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 6.263534069061279,
+      "learning_rate": 2.6385660386190582e-05,
+      "loss": 1.0688,
+      "step": 24840
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.723762512207031,
+      "learning_rate": 2.6369785798768744e-05,
+      "loss": 0.9624,
+      "step": 24850
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.153489589691162,
+      "learning_rate": 2.6353910657361287e-05,
+      "loss": 0.9957,
+      "step": 24860
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.928905487060547,
+      "learning_rate": 2.6338034968388632e-05,
+      "loss": 0.919,
+      "step": 24870
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.869629383087158,
+      "learning_rate": 2.632215873827142e-05,
+      "loss": 0.9661,
+      "step": 24880
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.619648456573486,
+      "learning_rate": 2.63062819734305e-05,
+      "loss": 0.9503,
+      "step": 24890
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.579214572906494,
+      "learning_rate": 2.629040468028697e-05,
+      "loss": 0.9481,
+      "step": 24900
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.047115802764893,
+      "learning_rate": 2.6274526865262094e-05,
+      "loss": 1.0517,
+      "step": 24910
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.717897891998291,
+      "learning_rate": 2.6258648534777396e-05,
+      "loss": 1.0285,
+      "step": 24920
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 6.477229118347168,
+      "learning_rate": 2.6242769695254578e-05,
+      "loss": 1.0433,
+      "step": 24930
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.9484124183654785,
+      "learning_rate": 2.6226890353115548e-05,
+      "loss": 0.8286,
+      "step": 24940
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.258402347564697,
+      "learning_rate": 2.621101051478243e-05,
+      "loss": 0.9466,
+      "step": 24950
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 6.534206390380859,
+      "learning_rate": 2.6195130186677552e-05,
+      "loss": 0.9399,
+      "step": 24960
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.956149101257324,
+      "learning_rate": 2.6179249375223418e-05,
+      "loss": 0.8654,
+      "step": 24970
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 7.101629257202148,
+      "learning_rate": 2.6163368086842748e-05,
+      "loss": 0.94,
+      "step": 24980
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.874995231628418,
+      "learning_rate": 2.6147486327958447e-05,
+      "loss": 0.9346,
+      "step": 24990
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.479530334472656,
+      "learning_rate": 2.6131604104993617e-05,
+      "loss": 0.9696,
+      "step": 25000
+    },
+    {
+      "epoch": 0.99,
+      "eval_loss": 0.822442352771759,
+      "eval_runtime": 423.5611,
+      "eval_samples_per_second": 50.101,
+      "eval_steps_per_second": 0.392,
+      "step": 25000
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.742852687835693,
+      "learning_rate": 2.6115721424371532e-05,
+      "loss": 1.0071,
+      "step": 25010
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.792036056518555,
+      "learning_rate": 2.6099838292515667e-05,
+      "loss": 1.1897,
+      "step": 25020
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 6.752234935760498,
+      "learning_rate": 2.6083954715849672e-05,
+      "loss": 1.1183,
+      "step": 25030
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 6.227586269378662,
+      "learning_rate": 2.6068070700797374e-05,
+      "loss": 0.9212,
+      "step": 25040
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 6.784265041351318,
+      "learning_rate": 2.6053774717739467e-05,
+      "loss": 1.0107,
+      "step": 25050
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.866714000701904,
+      "learning_rate": 2.6037889887451493e-05,
+      "loss": 0.9186,
+      "step": 25060
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 6.175454139709473,
+      "learning_rate": 2.6022004637407326e-05,
+      "loss": 1.068,
+      "step": 25070
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 6.915493488311768,
+      "learning_rate": 2.6006118974031474e-05,
+      "loss": 1.1189,
+      "step": 25080
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.7575249671936035,
+      "learning_rate": 2.599023290374861e-05,
+      "loss": 1.0609,
+      "step": 25090
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.929770469665527,
+      "learning_rate": 2.597434643298358e-05,
+      "loss": 0.929,
+      "step": 25100
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 6.256256580352783,
+      "learning_rate": 2.5958459568161376e-05,
+      "loss": 0.9206,
+      "step": 25110
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.565075874328613,
+      "learning_rate": 2.594257231570717e-05,
+      "loss": 1.1953,
+      "step": 25120
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 6.530720233917236,
+      "learning_rate": 2.592668468204627e-05,
+      "loss": 1.1846,
+      "step": 25130
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.470176696777344,
+      "learning_rate": 2.5910796673604152e-05,
+      "loss": 1.0581,
+      "step": 25140
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 7.724809169769287,
+      "learning_rate": 2.5894908296806447e-05,
+      "loss": 1.0698,
+      "step": 25150
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 7.146673202514648,
+      "learning_rate": 2.5879019558078915e-05,
+      "loss": 0.9748,
+      "step": 25160
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.7981722354888916,
+      "learning_rate": 2.5863130463847485e-05,
+      "loss": 0.9278,
+      "step": 25170
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.669990062713623,
+      "learning_rate": 2.5847241020538216e-05,
+      "loss": 0.9247,
+      "step": 25180
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.570769786834717,
+      "learning_rate": 2.5831351234577312e-05,
+      "loss": 1.0668,
+      "step": 25190
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.4049973487854,
+      "learning_rate": 2.581546111239112e-05,
+      "loss": 0.9017,
+      "step": 25200
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 11.65028190612793,
+      "learning_rate": 2.5799570660406108e-05,
+      "loss": 0.8371,
+      "step": 25210
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.09198522567749,
+      "learning_rate": 2.57836798850489e-05,
+      "loss": 0.8744,
+      "step": 25220
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.409224033355713,
+      "learning_rate": 2.5767788792746234e-05,
+      "loss": 0.7572,
+      "step": 25230
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.958026885986328,
+      "learning_rate": 2.5751897389924978e-05,
+      "loss": 0.9142,
+      "step": 25240
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 6.682915210723877,
+      "learning_rate": 2.5736005683012138e-05,
+      "loss": 0.9601,
+      "step": 25250
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 6.520613670349121,
+      "learning_rate": 2.572011367843482e-05,
+      "loss": 0.8349,
+      "step": 25260
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.187546730041504,
+      "learning_rate": 2.5704221382620274e-05,
+      "loss": 1.0163,
+      "step": 25270
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.339337348937988,
+      "learning_rate": 2.5688328801995852e-05,
+      "loss": 0.7157,
+      "step": 25280
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.10283899307251,
+      "learning_rate": 2.567243594298903e-05,
+      "loss": 0.8549,
+      "step": 25290
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 7.378357887268066,
+      "learning_rate": 2.5656542812027395e-05,
+      "loss": 0.9974,
+      "step": 25300
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.321687698364258,
+      "learning_rate": 2.5640649415538643e-05,
+      "loss": 0.8624,
+      "step": 25310
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 7.61053466796875,
+      "learning_rate": 2.5624755759950573e-05,
+      "loss": 0.9174,
+      "step": 25320
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 5.597862720489502,
+      "learning_rate": 2.560886185169109e-05,
+      "loss": 0.9436,
+      "step": 25330
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 6.148399829864502,
+      "learning_rate": 2.5592967697188215e-05,
+      "loss": 0.7815,
+      "step": 25340
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 6.358560562133789,
+      "learning_rate": 2.5577073302870046e-05,
+      "loss": 0.7412,
+      "step": 25350
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 5.995152950286865,
+      "learning_rate": 2.5561178675164797e-05,
+      "loss": 0.7892,
+      "step": 25360
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 6.227392196655273,
+      "learning_rate": 2.5545283820500766e-05,
+      "loss": 1.0994,
+      "step": 25370
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 5.120614528656006,
+      "learning_rate": 2.5529388745306344e-05,
+      "loss": 0.7946,
+      "step": 25380
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 6.2421956062316895,
+      "learning_rate": 2.5513493456010013e-05,
+      "loss": 0.7676,
+      "step": 25390
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 6.1328229904174805,
+      "learning_rate": 2.5497597959040342e-05,
+      "loss": 0.8032,
+      "step": 25400
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 5.464807510375977,
+      "learning_rate": 2.5481702260825984e-05,
+      "loss": 0.7989,
+      "step": 25410
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 3.6834354400634766,
+      "learning_rate": 2.5465806367795674e-05,
+      "loss": 0.7196,
+      "step": 25420
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 5.588569641113281,
+      "learning_rate": 2.544991028637823e-05,
+      "loss": 0.8116,
+      "step": 25430
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 4.825921535491943,
+      "learning_rate": 2.5434014023002524e-05,
+      "loss": 0.9004,
+      "step": 25440
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 4.437895774841309,
+      "learning_rate": 2.5418117584097535e-05,
+      "loss": 0.876,
+      "step": 25450
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 5.876149654388428,
+      "learning_rate": 2.5402220976092285e-05,
+      "loss": 0.8596,
+      "step": 25460
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 5.468808174133301,
+      "learning_rate": 2.5386324205415885e-05,
+      "loss": 0.7693,
+      "step": 25470
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 5.249061107635498,
+      "learning_rate": 2.5370427278497495e-05,
+      "loss": 0.8297,
+      "step": 25480
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 8.5701322555542,
+      "learning_rate": 2.5354530201766353e-05,
+      "loss": 0.8066,
+      "step": 25490
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 7.010660648345947,
+      "learning_rate": 2.5338632981651744e-05,
+      "loss": 0.7207,
+      "step": 25500
+    },
+    {
+      "epoch": 1.01,
+      "eval_loss": 0.8218079209327698,
+      "eval_runtime": 423.4645,
+      "eval_samples_per_second": 50.113,
+      "eval_steps_per_second": 0.392,
+      "step": 25500
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 8.821890830993652,
+      "learning_rate": 2.5322735624583016e-05,
+      "loss": 0.8635,
+      "step": 25510
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 5.41572380065918,
+      "learning_rate": 2.5306838136989582e-05,
+      "loss": 0.8069,
+      "step": 25520
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 6.105928421020508,
+      "learning_rate": 2.5290940525300894e-05,
+      "loss": 0.7652,
+      "step": 25530
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 5.625561237335205,
+      "learning_rate": 2.5275042795946458e-05,
+      "loss": 0.7926,
+      "step": 25540
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 5.759280681610107,
+      "learning_rate": 2.5259144955355834e-05,
+      "loss": 0.6571,
+      "step": 25550
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 6.044548511505127,
+      "learning_rate": 2.5243247009958625e-05,
+      "loss": 0.8736,
+      "step": 25560
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 5.191581726074219,
+      "learning_rate": 2.522734896618446e-05,
+      "loss": 0.7515,
+      "step": 25570
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 6.577776908874512,
+      "learning_rate": 2.5211450830463035e-05,
+      "loss": 0.8593,
+      "step": 25580
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 5.539211273193359,
+      "learning_rate": 2.5197142435012992e-05,
+      "loss": 0.9783,
+      "step": 25590
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 3.6030943393707275,
+      "learning_rate": 2.5181244142305664e-05,
+      "loss": 0.63,
+      "step": 25600
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 6.335265159606934,
+      "learning_rate": 2.5165345776297355e-05,
+      "loss": 0.7076,
+      "step": 25610
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 4.926506042480469,
+      "learning_rate": 2.514944734341787e-05,
+      "loss": 0.7943,
+      "step": 25620
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 6.27077579498291,
+      "learning_rate": 2.5133548850097056e-05,
+      "loss": 0.7876,
+      "step": 25630
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 6.246279239654541,
+      "learning_rate": 2.5117650302764773e-05,
+      "loss": 0.8547,
+      "step": 25640
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 5.265202522277832,
+      "learning_rate": 2.5101751707850902e-05,
+      "loss": 0.7452,
+      "step": 25650
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 4.233214378356934,
+      "learning_rate": 2.508585307178536e-05,
+      "loss": 0.8106,
+      "step": 25660
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 4.825283050537109,
+      "learning_rate": 2.506995440099806e-05,
+      "loss": 0.8445,
+      "step": 25670
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 6.330179691314697,
+      "learning_rate": 2.505405570191894e-05,
+      "loss": 0.8904,
+      "step": 25680
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 8.063553810119629,
+      "learning_rate": 2.5038156980977943e-05,
+      "loss": 0.81,
+      "step": 25690
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 6.087771415710449,
+      "learning_rate": 2.502225824460504e-05,
+      "loss": 0.9086,
+      "step": 25700
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 5.21497106552124,
+      "learning_rate": 2.500635949923018e-05,
+      "loss": 0.7476,
+      "step": 25710
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 4.309562683105469,
+      "learning_rate": 2.4990460751283335e-05,
+      "loss": 0.8674,
+      "step": 25720
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 7.431371688842773,
+      "learning_rate": 2.497456200719447e-05,
+      "loss": 0.9022,
+      "step": 25730
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 7.394750595092773,
+      "learning_rate": 2.4958663273393546e-05,
+      "loss": 0.9295,
+      "step": 25740
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 6.644321918487549,
+      "learning_rate": 2.4942764556310538e-05,
+      "loss": 0.8885,
+      "step": 25750
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 6.062170505523682,
+      "learning_rate": 2.492686586237538e-05,
+      "loss": 0.9228,
+      "step": 25760
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 5.980799674987793,
+      "learning_rate": 2.4910967198018043e-05,
+      "loss": 1.0506,
+      "step": 25770
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 5.726630687713623,
+      "learning_rate": 2.4895068569668438e-05,
+      "loss": 0.7897,
+      "step": 25780
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 6.5464701652526855,
+      "learning_rate": 2.48791699837565e-05,
+      "loss": 0.9827,
+      "step": 25790
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 4.447783470153809,
+      "learning_rate": 2.4863271446712112e-05,
+      "loss": 0.7824,
+      "step": 25800
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 5.8002753257751465,
+      "learning_rate": 2.484737296496517e-05,
+      "loss": 0.8018,
+      "step": 25810
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 6.029834270477295,
+      "learning_rate": 2.4831474544945538e-05,
+      "loss": 0.9521,
+      "step": 25820
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 6.578429222106934,
+      "learning_rate": 2.481557619308304e-05,
+      "loss": 0.8195,
+      "step": 25830
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 5.144342422485352,
+      "learning_rate": 2.4799677915807485e-05,
+      "loss": 1.0032,
+      "step": 25840
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 7.225157737731934,
+      "learning_rate": 2.478377971954865e-05,
+      "loss": 0.842,
+      "step": 25850
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 6.011831760406494,
+      "learning_rate": 2.476788161073629e-05,
+      "loss": 0.7993,
+      "step": 25860
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 4.749063968658447,
+      "learning_rate": 2.475198359580009e-05,
+      "loss": 0.884,
+      "step": 25870
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 5.749574184417725,
+      "learning_rate": 2.4736085681169745e-05,
+      "loss": 0.6652,
+      "step": 25880
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 3.2911605834960938,
+      "learning_rate": 2.4720187873274867e-05,
+      "loss": 0.667,
+      "step": 25890
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 6.496387481689453,
+      "learning_rate": 2.470429017854506e-05,
+      "loss": 0.8754,
+      "step": 25900
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 6.183441162109375,
+      "learning_rate": 2.4688392603409846e-05,
+      "loss": 0.9396,
+      "step": 25910
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 5.723308563232422,
+      "learning_rate": 2.4672495154298738e-05,
+      "loss": 0.6932,
+      "step": 25920
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 6.234645366668701,
+      "learning_rate": 2.4656597837641157e-05,
+      "loss": 0.7469,
+      "step": 25930
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 5.815822601318359,
+      "learning_rate": 2.4640700659866516e-05,
+      "loss": 0.8934,
+      "step": 25940
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 6.904934883117676,
+      "learning_rate": 2.462480362740412e-05,
+      "loss": 1.0636,
+      "step": 25950
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 4.239479064941406,
+      "learning_rate": 2.460890674668326e-05,
+      "loss": 0.7981,
+      "step": 25960
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 3.3169562816619873,
+      "learning_rate": 2.4593010024133145e-05,
+      "loss": 0.7816,
+      "step": 25970
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 6.4765448570251465,
+      "learning_rate": 2.457711346618292e-05,
+      "loss": 0.7303,
+      "step": 25980
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 6.9305338859558105,
+      "learning_rate": 2.456121707926168e-05,
+      "loss": 0.9839,
+      "step": 25990
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 5.555163860321045,
+      "learning_rate": 2.4545320869798406e-05,
+      "loss": 0.7685,
+      "step": 26000
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 0.8134897947311401,
+      "eval_runtime": 424.2718,
+      "eval_samples_per_second": 50.017,
+      "eval_steps_per_second": 0.391,
+      "step": 26000
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 5.668437480926514,
+      "learning_rate": 2.4529424844222077e-05,
+      "loss": 0.7596,
+      "step": 26010
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 6.587895393371582,
+      "learning_rate": 2.4513529008961525e-05,
+      "loss": 1.0133,
+      "step": 26020
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 7.3634467124938965,
+      "learning_rate": 2.4497633370445568e-05,
+      "loss": 0.8162,
+      "step": 26030
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 7.38131046295166,
+      "learning_rate": 2.448173793510289e-05,
+      "loss": 0.7067,
+      "step": 26040
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 5.404721260070801,
+      "learning_rate": 2.446584270936214e-05,
+      "loss": 0.7124,
+      "step": 26050
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 4.657472610473633,
+      "learning_rate": 2.4449947699651847e-05,
+      "loss": 0.908,
+      "step": 26060
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 4.948638439178467,
+      "learning_rate": 2.4434052912400474e-05,
+      "loss": 0.7016,
+      "step": 26070
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 6.4588398933410645,
+      "learning_rate": 2.4418158354036375e-05,
+      "loss": 1.0025,
+      "step": 26080
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 7.385505199432373,
+      "learning_rate": 2.4402264030987843e-05,
+      "loss": 0.9922,
+      "step": 26090
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 7.483156204223633,
+      "learning_rate": 2.4386369949683033e-05,
+      "loss": 0.9323,
+      "step": 26100
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 5.943239688873291,
+      "learning_rate": 2.4370476116550044e-05,
+      "loss": 1.0032,
+      "step": 26110
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 5.027673244476318,
+      "learning_rate": 2.4354582538016842e-05,
+      "loss": 0.9413,
+      "step": 26120
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 6.518718719482422,
+      "learning_rate": 2.4338689220511304e-05,
+      "loss": 0.9336,
+      "step": 26130
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 6.344429016113281,
+      "learning_rate": 2.4322796170461216e-05,
+      "loss": 0.8601,
+      "step": 26140
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 4.818910121917725,
+      "learning_rate": 2.430690339429422e-05,
+      "loss": 0.7398,
+      "step": 26150
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 5.471595764160156,
+      "learning_rate": 2.429101089843789e-05,
+      "loss": 0.8607,
+      "step": 26160
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 5.157017707824707,
+      "learning_rate": 2.4275118689319637e-05,
+      "loss": 0.9052,
+      "step": 26170
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 4.956745147705078,
+      "learning_rate": 2.4259226773366815e-05,
+      "loss": 0.7071,
+      "step": 26180
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 5.456181526184082,
+      "learning_rate": 2.4243335157006598e-05,
+      "loss": 0.7687,
+      "step": 26190
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 3.628636598587036,
+      "learning_rate": 2.42274438466661e-05,
+      "loss": 0.8581,
+      "step": 26200
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 6.662515163421631,
+      "learning_rate": 2.4211552848772244e-05,
+      "loss": 0.7964,
+      "step": 26210
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 6.005852222442627,
+      "learning_rate": 2.41956621697519e-05,
+      "loss": 0.8708,
+      "step": 26220
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 4.818019866943359,
+      "learning_rate": 2.4179771816031744e-05,
+      "loss": 0.7161,
+      "step": 26230
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 4.980724811553955,
+      "learning_rate": 2.4163881794038367e-05,
+      "loss": 0.9444,
+      "step": 26240
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 7.681107044219971,
+      "learning_rate": 2.4147992110198195e-05,
+      "loss": 0.957,
+      "step": 26250
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 6.3678436279296875,
+      "learning_rate": 2.413210277093753e-05,
+      "loss": 0.7574,
+      "step": 26260
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 6.465158462524414,
+      "learning_rate": 2.411621378268254e-05,
+      "loss": 0.8102,
+      "step": 26270
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 5.582615852355957,
+      "learning_rate": 2.4100325151859237e-05,
+      "loss": 0.7812,
+      "step": 26280
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 7.791302680969238,
+      "learning_rate": 2.4084436884893512e-05,
+      "loss": 0.9644,
+      "step": 26290
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 5.091775417327881,
+      "learning_rate": 2.4068548988211072e-05,
+      "loss": 0.8033,
+      "step": 26300
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 5.1609110832214355,
+      "learning_rate": 2.405266146823752e-05,
+      "loss": 0.9437,
+      "step": 26310
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 5.627232551574707,
+      "learning_rate": 2.4036774331398255e-05,
+      "loss": 0.9934,
+      "step": 26320
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 7.561347007751465,
+      "learning_rate": 2.4020887584118572e-05,
+      "loss": 0.6971,
+      "step": 26330
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 6.16213321685791,
+      "learning_rate": 2.4005001232823568e-05,
+      "loss": 0.7843,
+      "step": 26340
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 5.762958526611328,
+      "learning_rate": 2.3989115283938212e-05,
+      "loss": 0.9591,
+      "step": 26350
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 6.284593105316162,
+      "learning_rate": 2.397322974388728e-05,
+      "loss": 0.7908,
+      "step": 26360
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 4.062312602996826,
+      "learning_rate": 2.395734461909541e-05,
+      "loss": 0.7879,
+      "step": 26370
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 6.5525970458984375,
+      "learning_rate": 2.394145991598705e-05,
+      "loss": 0.9345,
+      "step": 26380
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 4.335060119628906,
+      "learning_rate": 2.392557564098649e-05,
+      "loss": 0.8968,
+      "step": 26390
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 7.626873970031738,
+      "learning_rate": 2.3909691800517843e-05,
+      "loss": 0.8632,
+      "step": 26400
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 5.506827354431152,
+      "learning_rate": 2.3893808401005055e-05,
+      "loss": 0.7814,
+      "step": 26410
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 5.59433650970459,
+      "learning_rate": 2.387792544887187e-05,
+      "loss": 0.9638,
+      "step": 26420
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 7.186860084533691,
+      "learning_rate": 2.3862042950541872e-05,
+      "loss": 0.9554,
+      "step": 26430
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 3.7368743419647217,
+      "learning_rate": 2.3846160912438476e-05,
+      "loss": 0.7341,
+      "step": 26440
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 7.189666748046875,
+      "learning_rate": 2.3830279340984863e-05,
+      "loss": 0.7599,
+      "step": 26450
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 4.515231609344482,
+      "learning_rate": 2.3814398242604075e-05,
+      "loss": 0.6771,
+      "step": 26460
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 5.628271102905273,
+      "learning_rate": 2.379851762371892e-05,
+      "loss": 0.7944,
+      "step": 26470
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 5.582614421844482,
+      "learning_rate": 2.3782637490752065e-05,
+      "loss": 0.9336,
+      "step": 26480
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 7.26879358291626,
+      "learning_rate": 2.3766757850125908e-05,
+      "loss": 0.9357,
+      "step": 26490
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 6.00214147567749,
+      "learning_rate": 2.3750878708262726e-05,
+      "loss": 0.8384,
+      "step": 26500
+    },
+    {
+      "epoch": 1.05,
+      "eval_loss": 0.8163621425628662,
+      "eval_runtime": 424.0406,
+      "eval_samples_per_second": 50.045,
+      "eval_steps_per_second": 0.391,
+      "step": 26500
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 7.826951026916504,
+      "learning_rate": 2.3735000071584535e-05,
+      "loss": 0.9924,
+      "step": 26510
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.5513205528259277,
+      "learning_rate": 2.3719121946513174e-05,
+      "loss": 0.8721,
+      "step": 26520
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 8.330363273620605,
+      "learning_rate": 2.3703244339470267e-05,
+      "loss": 0.9224,
+      "step": 26530
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 7.835899353027344,
+      "learning_rate": 2.3687367256877245e-05,
+      "loss": 0.8363,
+      "step": 26540
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 6.3573431968688965,
+      "learning_rate": 2.3671490705155286e-05,
+      "loss": 0.889,
+      "step": 26550
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 6.515617847442627,
+      "learning_rate": 2.365561469072541e-05,
+      "loss": 0.7764,
+      "step": 26560
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 5.223396301269531,
+      "learning_rate": 2.3639739220008365e-05,
+      "loss": 0.7352,
+      "step": 26570
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 6.898634433746338,
+      "learning_rate": 2.3623864299424714e-05,
+      "loss": 0.7099,
+      "step": 26580
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 5.1237030029296875,
+      "learning_rate": 2.36079899353948e-05,
+      "loss": 0.9318,
+      "step": 26590
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 8.609919548034668,
+      "learning_rate": 2.3592116134338708e-05,
+      "loss": 0.9497,
+      "step": 26600
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 7.335936546325684,
+      "learning_rate": 2.3576242902676335e-05,
+      "loss": 0.8113,
+      "step": 26610
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 7.537917137145996,
+      "learning_rate": 2.356037024682731e-05,
+      "loss": 1.1059,
+      "step": 26620
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 3.553316354751587,
+      "learning_rate": 2.3544498173211064e-05,
+      "loss": 0.6356,
+      "step": 26630
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 7.63238000869751,
+      "learning_rate": 2.3528626688246767e-05,
+      "loss": 0.7973,
+      "step": 26640
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 6.204401016235352,
+      "learning_rate": 2.351275579835336e-05,
+      "loss": 0.9121,
+      "step": 26650
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 5.891021251678467,
+      "learning_rate": 2.349688550994955e-05,
+      "loss": 0.8993,
+      "step": 26660
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 6.562130928039551,
+      "learning_rate": 2.348101582945379e-05,
+      "loss": 0.9895,
+      "step": 26670
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 5.667405128479004,
+      "learning_rate": 2.346514676328428e-05,
+      "loss": 0.7913,
+      "step": 26680
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 8.538568496704102,
+      "learning_rate": 2.344927831785901e-05,
+      "loss": 0.8095,
+      "step": 26690
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 5.930050849914551,
+      "learning_rate": 2.3433410499595664e-05,
+      "loss": 0.8035,
+      "step": 26700
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 4.325544357299805,
+      "learning_rate": 2.3417543314911723e-05,
+      "loss": 0.8588,
+      "step": 26710
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 5.191897392272949,
+      "learning_rate": 2.3401676770224366e-05,
+      "loss": 0.9598,
+      "step": 26720
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 6.541290760040283,
+      "learning_rate": 2.338581087195055e-05,
+      "loss": 0.8559,
+      "step": 26730
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 4.471774578094482,
+      "learning_rate": 2.336994562650696e-05,
+      "loss": 0.9726,
+      "step": 26740
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 3.615631580352783,
+      "learning_rate": 2.3354081040310005e-05,
+      "loss": 0.7438,
+      "step": 26750
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 4.3517255783081055,
+      "learning_rate": 2.333821711977585e-05,
+      "loss": 0.868,
+      "step": 26760
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 5.122809886932373,
+      "learning_rate": 2.332235387132036e-05,
+      "loss": 0.8855,
+      "step": 26770
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 6.609111309051514,
+      "learning_rate": 2.330649130135915e-05,
+      "loss": 0.9223,
+      "step": 26780
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 7.1288933753967285,
+      "learning_rate": 2.3290629416307562e-05,
+      "loss": 1.0309,
+      "step": 26790
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 5.815499782562256,
+      "learning_rate": 2.3274768222580647e-05,
+      "loss": 0.7258,
+      "step": 26800
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.1416308879852295,
+      "learning_rate": 2.3258907726593188e-05,
+      "loss": 0.7471,
+      "step": 26810
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 5.40012264251709,
+      "learning_rate": 2.3243047934759688e-05,
+      "loss": 0.8442,
+      "step": 26820
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 6.606177806854248,
+      "learning_rate": 2.322718885349434e-05,
+      "loss": 0.6922,
+      "step": 26830
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 6.906505584716797,
+      "learning_rate": 2.321133048921109e-05,
+      "loss": 1.1521,
+      "step": 26840
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 4.048044681549072,
+      "learning_rate": 2.3195472848323553e-05,
+      "loss": 0.7805,
+      "step": 26850
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 5.731555938720703,
+      "learning_rate": 2.3179615937245095e-05,
+      "loss": 0.7655,
+      "step": 26860
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 8.18224811553955,
+      "learning_rate": 2.3163759762388736e-05,
+      "loss": 0.9756,
+      "step": 26870
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 6.7355217933654785,
+      "learning_rate": 2.3147904330167243e-05,
+      "loss": 0.8416,
+      "step": 26880
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 6.364537715911865,
+      "learning_rate": 2.313204964699307e-05,
+      "loss": 0.7805,
+      "step": 26890
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 5.899649620056152,
+      "learning_rate": 2.3116195719278346e-05,
+      "loss": 0.9004,
+      "step": 26900
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 6.651846408843994,
+      "learning_rate": 2.3100342553434924e-05,
+      "loss": 0.7666,
+      "step": 26910
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 6.383004188537598,
+      "learning_rate": 2.308449015587433e-05,
+      "loss": 0.8207,
+      "step": 26920
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 5.144965171813965,
+      "learning_rate": 2.3068638533007786e-05,
+      "loss": 0.6595,
+      "step": 26930
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 5.40513277053833,
+      "learning_rate": 2.3052787691246202e-05,
+      "loss": 0.6879,
+      "step": 26940
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 5.859258651733398,
+      "learning_rate": 2.3036937637000175e-05,
+      "loss": 0.9038,
+      "step": 26950
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 6.211092472076416,
+      "learning_rate": 2.3021088376679962e-05,
+      "loss": 0.9333,
+      "step": 26960
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 6.845388889312744,
+      "learning_rate": 2.3005239916695544e-05,
+      "loss": 0.8483,
+      "step": 26970
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 7.248631000518799,
+      "learning_rate": 2.2989392263456515e-05,
+      "loss": 0.7706,
+      "step": 26980
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 4.441316604614258,
+      "learning_rate": 2.2973545423372213e-05,
+      "loss": 0.8041,
+      "step": 26990
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 5.478137493133545,
+      "learning_rate": 2.2957699402851583e-05,
+      "loss": 0.7815,
+      "step": 27000
+    },
+    {
+      "epoch": 1.07,
+      "eval_loss": 0.8039602041244507,
+      "eval_runtime": 423.1984,
+      "eval_samples_per_second": 50.144,
+      "eval_steps_per_second": 0.392,
+      "step": 27000
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 4.606276988983154,
+      "learning_rate": 2.2941854208303293e-05,
+      "loss": 0.791,
+      "step": 27010
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 5.879239559173584,
+      "learning_rate": 2.2926009846135626e-05,
+      "loss": 0.9461,
+      "step": 27020
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 7.483617782592773,
+      "learning_rate": 2.2910166322756575e-05,
+      "loss": 0.8627,
+      "step": 27030
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 6.545680046081543,
+      "learning_rate": 2.2894323644573758e-05,
+      "loss": 1.0065,
+      "step": 27040
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 4.946022987365723,
+      "learning_rate": 2.2878481817994475e-05,
+      "loss": 0.989,
+      "step": 27050
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 5.759241580963135,
+      "learning_rate": 2.286264084942567e-05,
+      "loss": 0.8208,
+      "step": 27060
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 8.829256057739258,
+      "learning_rate": 2.2846800745273942e-05,
+      "loss": 0.9144,
+      "step": 27070
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 6.725993633270264,
+      "learning_rate": 2.2830961511945543e-05,
+      "loss": 0.9086,
+      "step": 27080
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 8.805347442626953,
+      "learning_rate": 2.281512315584636e-05,
+      "loss": 0.8256,
+      "step": 27090
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 3.1862964630126953,
+      "learning_rate": 2.2799285683381953e-05,
+      "loss": 0.8983,
+      "step": 27100
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 6.6998395919799805,
+      "learning_rate": 2.2783449100957484e-05,
+      "loss": 1.0174,
+      "step": 27110
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 5.477229118347168,
+      "learning_rate": 2.2767613414977806e-05,
+      "loss": 0.764,
+      "step": 27120
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 5.287775993347168,
+      "learning_rate": 2.275177863184736e-05,
+      "loss": 0.7611,
+      "step": 27130
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 4.781260967254639,
+      "learning_rate": 2.273594475797026e-05,
+      "loss": 0.8696,
+      "step": 27140
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 9.884573936462402,
+      "learning_rate": 2.272011179975022e-05,
+      "loss": 0.9336,
+      "step": 27150
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 6.082416534423828,
+      "learning_rate": 2.2704279763590615e-05,
+      "loss": 0.7102,
+      "step": 27160
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 5.9974775314331055,
+      "learning_rate": 2.2688448655894415e-05,
+      "loss": 0.7835,
+      "step": 27170
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 6.388651371002197,
+      "learning_rate": 2.2672618483064245e-05,
+      "loss": 0.9204,
+      "step": 27180
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 5.64275598526001,
+      "learning_rate": 2.2656789251502337e-05,
+      "loss": 1.0006,
+      "step": 27190
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 6.0904059410095215,
+      "learning_rate": 2.264096096761054e-05,
+      "loss": 0.8826,
+      "step": 27200
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 7.5173845291137695,
+      "learning_rate": 2.262513363779033e-05,
+      "loss": 0.811,
+      "step": 27210
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 4.211666584014893,
+      "learning_rate": 2.260930726844277e-05,
+      "loss": 0.9456,
+      "step": 27220
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 4.181629180908203,
+      "learning_rate": 2.2593481865968583e-05,
+      "loss": 0.955,
+      "step": 27230
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 5.616909503936768,
+      "learning_rate": 2.2577657436768048e-05,
+      "loss": 0.9892,
+      "step": 27240
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 5.494133472442627,
+      "learning_rate": 2.2561833987241096e-05,
+      "loss": 1.0487,
+      "step": 27250
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 5.282896995544434,
+      "learning_rate": 2.254601152378722e-05,
+      "loss": 0.8506,
+      "step": 27260
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 5.415282726287842,
+      "learning_rate": 2.253019005280556e-05,
+      "loss": 0.6722,
+      "step": 27270
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 5.3605756759643555,
+      "learning_rate": 2.251436958069481e-05,
+      "loss": 0.9015,
+      "step": 27280
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.710216760635376,
+      "learning_rate": 2.2498550113853288e-05,
+      "loss": 0.687,
+      "step": 27290
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 4.0933403968811035,
+      "learning_rate": 2.248273165867889e-05,
+      "loss": 0.843,
+      "step": 27300
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 5.501979351043701,
+      "learning_rate": 2.246691422156913e-05,
+      "loss": 0.7038,
+      "step": 27310
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 6.13559627532959,
+      "learning_rate": 2.2451097808921063e-05,
+      "loss": 0.6361,
+      "step": 27320
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 6.584084987640381,
+      "learning_rate": 2.243528242713138e-05,
+      "loss": 0.7152,
+      "step": 27330
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 4.350105285644531,
+      "learning_rate": 2.2419468082596317e-05,
+      "loss": 0.7278,
+      "step": 27340
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 6.251486301422119,
+      "learning_rate": 2.2403654781711712e-05,
+      "loss": 0.8233,
+      "step": 27350
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 4.443666458129883,
+      "learning_rate": 2.238784253087298e-05,
+      "loss": 0.7223,
+      "step": 27360
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 6.349146366119385,
+      "learning_rate": 2.237203133647509e-05,
+      "loss": 0.9775,
+      "step": 27370
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 6.747358798980713,
+      "learning_rate": 2.2356221204912624e-05,
+      "loss": 0.8125,
+      "step": 27380
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 4.399649143218994,
+      "learning_rate": 2.234041214257968e-05,
+      "loss": 0.6594,
+      "step": 27390
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 6.970942974090576,
+      "learning_rate": 2.2324604155869984e-05,
+      "loss": 0.9508,
+      "step": 27400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 4.153322219848633,
+      "learning_rate": 2.2308797251176772e-05,
+      "loss": 0.7113,
+      "step": 27410
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 3.4371299743652344,
+      "learning_rate": 2.2292991434892885e-05,
+      "loss": 0.8152,
+      "step": 27420
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 5.604278564453125,
+      "learning_rate": 2.2277186713410687e-05,
+      "loss": 0.7191,
+      "step": 27430
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 4.119433403015137,
+      "learning_rate": 2.2261383093122142e-05,
+      "loss": 0.9397,
+      "step": 27440
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 7.586767673492432,
+      "learning_rate": 2.2245580580418717e-05,
+      "loss": 0.8429,
+      "step": 27450
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 6.449422836303711,
+      "learning_rate": 2.222977918169148e-05,
+      "loss": 0.816,
+      "step": 27460
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 4.070833683013916,
+      "learning_rate": 2.2213978903331017e-05,
+      "loss": 0.7008,
+      "step": 27470
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 5.02301549911499,
+      "learning_rate": 2.2198179751727473e-05,
+      "loss": 0.7639,
+      "step": 27480
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 3.2170095443725586,
+      "learning_rate": 2.2182381733270535e-05,
+      "loss": 0.8448,
+      "step": 27490
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 4.735213756561279,
+      "learning_rate": 2.2166584854349426e-05,
+      "loss": 0.9966,
+      "step": 27500
+    },
+    {
+      "epoch": 1.09,
+      "eval_loss": 0.8120434284210205,
+      "eval_runtime": 423.5525,
+      "eval_samples_per_second": 50.102,
+      "eval_steps_per_second": 0.392,
+      "step": 27500
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 6.116840362548828,
+      "learning_rate": 2.2150789121352934e-05,
+      "loss": 0.7008,
+      "step": 27510
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 8.083114624023438,
+      "learning_rate": 2.2134994540669334e-05,
+      "loss": 0.9438,
+      "step": 27520
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 5.5293073654174805,
+      "learning_rate": 2.2119201118686493e-05,
+      "loss": 0.791,
+      "step": 27530
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 6.167052745819092,
+      "learning_rate": 2.2103408861791762e-05,
+      "loss": 0.8928,
+      "step": 27540
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 6.839602947235107,
+      "learning_rate": 2.208761777637205e-05,
+      "loss": 0.8532,
+      "step": 27550
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 6.9067182540893555,
+      "learning_rate": 2.2071827868813774e-05,
+      "loss": 0.8446,
+      "step": 27560
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 5.717200756072998,
+      "learning_rate": 2.2056039145502896e-05,
+      "loss": 0.9497,
+      "step": 27570
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 7.686497211456299,
+      "learning_rate": 2.2040251612824863e-05,
+      "loss": 0.8287,
+      "step": 27580
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 5.452780723571777,
+      "learning_rate": 2.202446527716469e-05,
+      "loss": 0.8571,
+      "step": 27590
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 7.625713348388672,
+      "learning_rate": 2.2008680144906863e-05,
+      "loss": 0.7942,
+      "step": 27600
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 4.4305033683776855,
+      "learning_rate": 2.1992896222435402e-05,
+      "loss": 0.7108,
+      "step": 27610
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 5.990145683288574,
+      "learning_rate": 2.1977113516133837e-05,
+      "loss": 0.8994,
+      "step": 27620
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 5.80300235748291,
+      "learning_rate": 2.1961332032385214e-05,
+      "loss": 0.808,
+      "step": 27630
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 5.097789764404297,
+      "learning_rate": 2.194555177757205e-05,
+      "loss": 0.7884,
+      "step": 27640
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 4.728410720825195,
+      "learning_rate": 2.1929772758076407e-05,
+      "loss": 0.7172,
+      "step": 27650
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 6.903634548187256,
+      "learning_rate": 2.1913994980279833e-05,
+      "loss": 0.8995,
+      "step": 27660
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 6.09291934967041,
+      "learning_rate": 2.1898218450563363e-05,
+      "loss": 0.7821,
+      "step": 27670
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 4.25015115737915,
+      "learning_rate": 2.1882443175307543e-05,
+      "loss": 0.8201,
+      "step": 27680
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 7.11330509185791,
+      "learning_rate": 2.186666916089239e-05,
+      "loss": 0.8858,
+      "step": 27690
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 7.739790439605713,
+      "learning_rate": 2.1850896413697448e-05,
+      "loss": 0.7856,
+      "step": 27700
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 6.269674301147461,
+      "learning_rate": 2.1835124940101698e-05,
+      "loss": 0.627,
+      "step": 27710
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 6.729603290557861,
+      "learning_rate": 2.181935474648366e-05,
+      "loss": 0.7841,
+      "step": 27720
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 7.994325637817383,
+      "learning_rate": 2.1803585839221286e-05,
+      "loss": 0.9863,
+      "step": 27730
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 5.520269870758057,
+      "learning_rate": 2.1787818224692045e-05,
+      "loss": 0.8233,
+      "step": 27740
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 6.797540187835693,
+      "learning_rate": 2.1772051909272864e-05,
+      "loss": 0.8018,
+      "step": 27750
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 5.143209934234619,
+      "learning_rate": 2.175628689934016e-05,
+      "loss": 0.8895,
+      "step": 27760
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 5.565918445587158,
+      "learning_rate": 2.1740523201269793e-05,
+      "loss": 0.953,
+      "step": 27770
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 4.588919639587402,
+      "learning_rate": 2.1724760821437136e-05,
+      "loss": 0.5164,
+      "step": 27780
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 5.243714809417725,
+      "learning_rate": 2.170899976621698e-05,
+      "loss": 0.854,
+      "step": 27790
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 4.072144031524658,
+      "learning_rate": 2.169324004198362e-05,
+      "loss": 0.675,
+      "step": 27800
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 12.309389114379883,
+      "learning_rate": 2.1677481655110797e-05,
+      "loss": 0.7198,
+      "step": 27810
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 6.904654502868652,
+      "learning_rate": 2.1661724611971708e-05,
+      "loss": 0.9073,
+      "step": 27820
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 7.754100799560547,
+      "learning_rate": 2.1645968918939012e-05,
+      "loss": 1.0127,
+      "step": 27830
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 4.620045185089111,
+      "learning_rate": 2.1630214582384818e-05,
+      "loss": 0.8581,
+      "step": 27840
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 5.994672775268555,
+      "learning_rate": 2.1614461608680696e-05,
+      "loss": 0.8304,
+      "step": 27850
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 8.081524848937988,
+      "learning_rate": 2.1598710004197646e-05,
+      "loss": 0.9181,
+      "step": 27860
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 5.951777935028076,
+      "learning_rate": 2.158295977530613e-05,
+      "loss": 0.8525,
+      "step": 27870
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 5.592679023742676,
+      "learning_rate": 2.1567210928376052e-05,
+      "loss": 0.8406,
+      "step": 27880
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 4.193765163421631,
+      "learning_rate": 2.1551463469776752e-05,
+      "loss": 0.8025,
+      "step": 27890
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 4.673293113708496,
+      "learning_rate": 2.153571740587701e-05,
+      "loss": 0.8581,
+      "step": 27900
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 4.973916053771973,
+      "learning_rate": 2.1519972743045054e-05,
+      "loss": 0.857,
+      "step": 27910
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 6.531762599945068,
+      "learning_rate": 2.150422948764851e-05,
+      "loss": 0.9002,
+      "step": 27920
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 6.391998767852783,
+      "learning_rate": 2.1488487646054482e-05,
+      "loss": 0.9113,
+      "step": 27930
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 5.798454761505127,
+      "learning_rate": 2.1472747224629462e-05,
+      "loss": 0.8432,
+      "step": 27940
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 8.478912353515625,
+      "learning_rate": 2.1457008229739394e-05,
+      "loss": 0.8128,
+      "step": 27950
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 4.298527240753174,
+      "learning_rate": 2.1441270667749647e-05,
+      "loss": 0.7071,
+      "step": 27960
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 7.143258094787598,
+      "learning_rate": 2.142553454502498e-05,
+      "loss": 1.0005,
+      "step": 27970
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 4.7733049392700195,
+      "learning_rate": 2.14097998679296e-05,
+      "loss": 0.7132,
+      "step": 27980
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 6.2097673416137695,
+      "learning_rate": 2.1394066642827116e-05,
+      "loss": 0.9356,
+      "step": 27990
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 4.75264310836792,
+      "learning_rate": 2.137833487608055e-05,
+      "loss": 0.9326,
+      "step": 28000
+    },
+    {
+      "epoch": 1.11,
+      "eval_loss": 0.7957361340522766,
+      "eval_runtime": 423.9055,
+      "eval_samples_per_second": 50.061,
+      "eval_steps_per_second": 0.392,
+      "step": 28000
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 6.989806175231934,
+      "learning_rate": 2.136260457405234e-05,
+      "loss": 0.6873,
+      "step": 28010
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 4.219059467315674,
+      "learning_rate": 2.1346875743104328e-05,
+      "loss": 0.8081,
+      "step": 28020
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 7.7409257888793945,
+      "learning_rate": 2.133114838959776e-05,
+      "loss": 0.8512,
+      "step": 28030
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 4.118712902069092,
+      "learning_rate": 2.1315422519893284e-05,
+      "loss": 0.8059,
+      "step": 28040
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 4.194961071014404,
+      "learning_rate": 2.129969814035095e-05,
+      "loss": 0.8236,
+      "step": 28050
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 5.040365219116211,
+      "learning_rate": 2.1283975257330212e-05,
+      "loss": 0.6846,
+      "step": 28060
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 6.683485507965088,
+      "learning_rate": 2.12682538771899e-05,
+      "loss": 0.8424,
+      "step": 28070
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 5.844532489776611,
+      "learning_rate": 2.1252534006288264e-05,
+      "loss": 0.7068,
+      "step": 28080
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 3.9057223796844482,
+      "learning_rate": 2.1236815650982907e-05,
+      "loss": 0.805,
+      "step": 28090
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 5.113594055175781,
+      "learning_rate": 2.122109881763085e-05,
+      "loss": 0.7623,
+      "step": 28100
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 7.395715236663818,
+      "learning_rate": 2.1205383512588495e-05,
+      "loss": 0.6958,
+      "step": 28110
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 9.980669975280762,
+      "learning_rate": 2.118966974221161e-05,
+      "loss": 0.9412,
+      "step": 28120
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 7.687812805175781,
+      "learning_rate": 2.1173957512855356e-05,
+      "loss": 0.8665,
+      "step": 28130
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 4.53033971786499,
+      "learning_rate": 2.115824683087426e-05,
+      "loss": 0.9047,
+      "step": 28140
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 4.298343181610107,
+      "learning_rate": 2.1142537702622233e-05,
+      "loss": 0.6928,
+      "step": 28150
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 8.962779998779297,
+      "learning_rate": 2.1126830134452556e-05,
+      "loss": 0.9243,
+      "step": 28160
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 6.940954208374023,
+      "learning_rate": 2.1111124132717883e-05,
+      "loss": 0.814,
+      "step": 28170
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 6.626890659332275,
+      "learning_rate": 2.1095419703770206e-05,
+      "loss": 1.0028,
+      "step": 28180
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 3.1184871196746826,
+      "learning_rate": 2.107971685396093e-05,
+      "loss": 0.6425,
+      "step": 28190
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 4.596141815185547,
+      "learning_rate": 2.1064015589640768e-05,
+      "loss": 0.9146,
+      "step": 28200
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 7.643907070159912,
+      "learning_rate": 2.1048315917159846e-05,
+      "loss": 0.8848,
+      "step": 28210
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 3.3073270320892334,
+      "learning_rate": 2.103261784286759e-05,
+      "loss": 0.6529,
+      "step": 28220
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 8.464675903320312,
+      "learning_rate": 2.1016921373112842e-05,
+      "loss": 0.8443,
+      "step": 28230
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 4.538732051849365,
+      "learning_rate": 2.1001226514243727e-05,
+      "loss": 0.7508,
+      "step": 28240
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 5.070686340332031,
+      "learning_rate": 2.098553327260777e-05,
+      "loss": 0.905,
+      "step": 28250
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 4.474721908569336,
+      "learning_rate": 2.0969841654551834e-05,
+      "loss": 0.7506,
+      "step": 28260
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 6.657362461090088,
+      "learning_rate": 2.09541516664221e-05,
+      "loss": 0.7138,
+      "step": 28270
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 5.7071380615234375,
+      "learning_rate": 2.0938463314564115e-05,
+      "loss": 0.7417,
+      "step": 28280
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 6.516452312469482,
+      "learning_rate": 2.0922776605322748e-05,
+      "loss": 0.7997,
+      "step": 28290
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 7.425762176513672,
+      "learning_rate": 2.090709154504222e-05,
+      "loss": 0.8945,
+      "step": 28300
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 5.220942974090576,
+      "learning_rate": 2.0891408140066063e-05,
+      "loss": 0.9382,
+      "step": 28310
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 6.388967514038086,
+      "learning_rate": 2.087572639673717e-05,
+      "loss": 0.8898,
+      "step": 28320
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 6.784420490264893,
+      "learning_rate": 2.0860046321397723e-05,
+      "loss": 0.9121,
+      "step": 28330
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 7.733227252960205,
+      "learning_rate": 2.0844367920389273e-05,
+      "loss": 0.7338,
+      "step": 28340
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 5.887832164764404,
+      "learning_rate": 2.0828691200052653e-05,
+      "loss": 0.8239,
+      "step": 28350
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 5.135347843170166,
+      "learning_rate": 2.0813016166728054e-05,
+      "loss": 0.9003,
+      "step": 28360
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 4.98162841796875,
+      "learning_rate": 2.079734282675494e-05,
+      "loss": 0.867,
+      "step": 28370
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 6.192076683044434,
+      "learning_rate": 2.0781671186472147e-05,
+      "loss": 0.7835,
+      "step": 28380
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 4.777847766876221,
+      "learning_rate": 2.076600125221777e-05,
+      "loss": 0.9851,
+      "step": 28390
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 4.628394603729248,
+      "learning_rate": 2.075033303032924e-05,
+      "loss": 0.8748,
+      "step": 28400
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 5.411494731903076,
+      "learning_rate": 2.07346665271433e-05,
+      "loss": 0.8238,
+      "step": 28410
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 7.587833881378174,
+      "learning_rate": 2.0719001748995987e-05,
+      "loss": 0.6896,
+      "step": 28420
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 15.300435066223145,
+      "learning_rate": 2.0704904928807602e-05,
+      "loss": 1.1702,
+      "step": 28430
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 7.078420162200928,
+      "learning_rate": 2.0689243445686964e-05,
+      "loss": 0.837,
+      "step": 28440
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 6.704090595245361,
+      "learning_rate": 2.0673583705975512e-05,
+      "loss": 0.9921,
+      "step": 28450
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 9.670187950134277,
+      "learning_rate": 2.0657925716006558e-05,
+      "loss": 0.9566,
+      "step": 28460
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 6.644848346710205,
+      "learning_rate": 2.064226948211268e-05,
+      "loss": 0.8137,
+      "step": 28470
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 4.828913688659668,
+      "learning_rate": 2.0626615010625792e-05,
+      "loss": 0.9829,
+      "step": 28480
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 6.244342803955078,
+      "learning_rate": 2.0610962307877034e-05,
+      "loss": 0.9614,
+      "step": 28490
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 6.316976547241211,
+      "learning_rate": 2.05953113801969e-05,
+      "loss": 0.8633,
+      "step": 28500
+    },
+    {
+      "epoch": 1.13,
+      "eval_loss": 0.8211910128593445,
+      "eval_runtime": 423.3843,
+      "eval_samples_per_second": 50.122,
+      "eval_steps_per_second": 0.392,
+      "step": 28500
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 9.63792610168457,
+      "learning_rate": 2.0579662233915102e-05,
+      "loss": 0.8989,
+      "step": 28510
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 6.637293338775635,
+      "learning_rate": 2.0564014875360683e-05,
+      "loss": 0.9424,
+      "step": 28520
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 3.8526394367218018,
+      "learning_rate": 2.0548369310861915e-05,
+      "loss": 0.8532,
+      "step": 28530
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 4.357787609100342,
+      "learning_rate": 2.0532725546746394e-05,
+      "loss": 0.8687,
+      "step": 28540
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 6.075691223144531,
+      "learning_rate": 2.0517083589340948e-05,
+      "loss": 0.929,
+      "step": 28550
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 6.322478771209717,
+      "learning_rate": 2.05014434449717e-05,
+      "loss": 0.8247,
+      "step": 28560
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 9.390968322753906,
+      "learning_rate": 2.0485805119964026e-05,
+      "loss": 0.9507,
+      "step": 28570
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 8.567316055297852,
+      "learning_rate": 2.0470168620642564e-05,
+      "loss": 0.8729,
+      "step": 28580
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 5.972450256347656,
+      "learning_rate": 2.0454533953331227e-05,
+      "loss": 0.6948,
+      "step": 28590
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 5.87274694442749,
+      "learning_rate": 2.0438901124353172e-05,
+      "loss": 0.7915,
+      "step": 28600
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 14.516148567199707,
+      "learning_rate": 2.0423270140030836e-05,
+      "loss": 0.7066,
+      "step": 28610
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 6.155118465423584,
+      "learning_rate": 2.0407641006685865e-05,
+      "loss": 0.8352,
+      "step": 28620
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 4.854623317718506,
+      "learning_rate": 2.039201373063922e-05,
+      "loss": 0.7912,
+      "step": 28630
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 5.83957052230835,
+      "learning_rate": 2.037638831821104e-05,
+      "loss": 0.7978,
+      "step": 28640
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 4.001251697540283,
+      "learning_rate": 2.0360764775720784e-05,
+      "loss": 1.1015,
+      "step": 28650
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 6.929150581359863,
+      "learning_rate": 2.0345143109487083e-05,
+      "loss": 0.942,
+      "step": 28660
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 5.4490790367126465,
+      "learning_rate": 2.032952332582787e-05,
+      "loss": 0.6576,
+      "step": 28670
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 5.2259931564331055,
+      "learning_rate": 2.031390543106027e-05,
+      "loss": 0.8493,
+      "step": 28680
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 7.039891719818115,
+      "learning_rate": 2.029828943150068e-05,
+      "loss": 0.7428,
+      "step": 28690
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 6.740241050720215,
+      "learning_rate": 2.0282675333464704e-05,
+      "loss": 0.8117,
+      "step": 28700
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 5.39209508895874,
+      "learning_rate": 2.0267063143267193e-05,
+      "loss": 1.0297,
+      "step": 28710
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 6.868276119232178,
+      "learning_rate": 2.0251452867222224e-05,
+      "loss": 0.6847,
+      "step": 28720
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 6.516077518463135,
+      "learning_rate": 2.0235844511643083e-05,
+      "loss": 0.8183,
+      "step": 28730
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 4.910519123077393,
+      "learning_rate": 2.022023808284231e-05,
+      "loss": 0.7989,
+      "step": 28740
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 7.043450355529785,
+      "learning_rate": 2.0204633587131634e-05,
+      "loss": 0.8918,
+      "step": 28750
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 5.686897277832031,
+      "learning_rate": 2.0189031030822032e-05,
+      "loss": 0.8069,
+      "step": 28760
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 4.496852874755859,
+      "learning_rate": 2.017343042022366e-05,
+      "loss": 0.7016,
+      "step": 28770
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 7.174291133880615,
+      "learning_rate": 2.015783176164594e-05,
+      "loss": 0.9268,
+      "step": 28780
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 7.180439472198486,
+      "learning_rate": 2.014223506139744e-05,
+      "loss": 0.7313,
+      "step": 28790
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 5.850827693939209,
+      "learning_rate": 2.0126640325786e-05,
+      "loss": 0.8114,
+      "step": 28800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 6.6945295333862305,
+      "learning_rate": 2.011104756111861e-05,
+      "loss": 1.0083,
+      "step": 28810
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 7.274098873138428,
+      "learning_rate": 2.0095456773701508e-05,
+      "loss": 0.8759,
+      "step": 28820
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 5.3831071853637695,
+      "learning_rate": 2.0079867969840087e-05,
+      "loss": 0.8433,
+      "step": 28830
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 6.4197869300842285,
+      "learning_rate": 2.0064281155838994e-05,
+      "loss": 0.7917,
+      "step": 28840
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 7.371774673461914,
+      "learning_rate": 2.0048696338002015e-05,
+      "loss": 0.9024,
+      "step": 28850
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 6.69523286819458,
+      "learning_rate": 2.0033113522632158e-05,
+      "loss": 0.9533,
+      "step": 28860
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 4.825892448425293,
+      "learning_rate": 2.001753271603163e-05,
+      "loss": 0.9519,
+      "step": 28870
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 8.741418838500977,
+      "learning_rate": 2.0001953924501795e-05,
+      "loss": 0.8733,
+      "step": 28880
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 8.151668548583984,
+      "learning_rate": 1.998637715434324e-05,
+      "loss": 1.0583,
+      "step": 28890
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 6.010997295379639,
+      "learning_rate": 1.997080241185569e-05,
+      "loss": 0.7151,
+      "step": 28900
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 4.502925872802734,
+      "learning_rate": 1.9955229703338095e-05,
+      "loss": 0.9183,
+      "step": 28910
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 7.09443473815918,
+      "learning_rate": 1.9939659035088546e-05,
+      "loss": 0.7754,
+      "step": 28920
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 7.443140506744385,
+      "learning_rate": 1.992409041340434e-05,
+      "loss": 1.0023,
+      "step": 28930
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 5.736412048339844,
+      "learning_rate": 1.990852384458191e-05,
+      "loss": 0.8287,
+      "step": 28940
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 5.1260294914245605,
+      "learning_rate": 1.98929593349169e-05,
+      "loss": 0.6646,
+      "step": 28950
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 7.711024761199951,
+      "learning_rate": 1.9877396890704082e-05,
+      "loss": 0.8175,
+      "step": 28960
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 6.406548976898193,
+      "learning_rate": 1.9861836518237426e-05,
+      "loss": 0.8494,
+      "step": 28970
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 7.0874924659729,
+      "learning_rate": 1.984627822381004e-05,
+      "loss": 0.9167,
+      "step": 28980
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 5.650824069976807,
+      "learning_rate": 1.9830722013714205e-05,
+      "loss": 1.0478,
+      "step": 28990
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 7.979854106903076,
+      "learning_rate": 1.9815167894241356e-05,
+      "loss": 1.0098,
+      "step": 29000
+    },
+    {
+      "epoch": 1.15,
+      "eval_loss": 0.8063564896583557,
+      "eval_runtime": 423.6533,
+      "eval_samples_per_second": 50.09,
+      "eval_steps_per_second": 0.392,
+      "step": 29000
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 4.561464786529541,
+      "learning_rate": 1.979961587168207e-05,
+      "loss": 0.912,
+      "step": 29010
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 3.882627010345459,
+      "learning_rate": 1.978406595232611e-05,
+      "loss": 0.8826,
+      "step": 29020
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 7.464628219604492,
+      "learning_rate": 1.9768518142462337e-05,
+      "loss": 0.8106,
+      "step": 29030
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 8.579936981201172,
+      "learning_rate": 1.9752972448378814e-05,
+      "loss": 0.8239,
+      "step": 29040
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 5.6800456047058105,
+      "learning_rate": 1.97374288763627e-05,
+      "loss": 0.8382,
+      "step": 29050
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 8.313254356384277,
+      "learning_rate": 1.9721887432700343e-05,
+      "loss": 0.8901,
+      "step": 29060
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 7.248850345611572,
+      "learning_rate": 1.9706348123677174e-05,
+      "loss": 0.6893,
+      "step": 29070
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 4.559626579284668,
+      "learning_rate": 1.969081095557782e-05,
+      "loss": 0.8996,
+      "step": 29080
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 6.618696689605713,
+      "learning_rate": 1.9675275934685986e-05,
+      "loss": 0.7863,
+      "step": 29090
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 7.138997554779053,
+      "learning_rate": 1.9659743067284565e-05,
+      "loss": 0.8898,
+      "step": 29100
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 4.046321868896484,
+      "learning_rate": 1.964421235965553e-05,
+      "loss": 0.8379,
+      "step": 29110
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 4.938464164733887,
+      "learning_rate": 1.9628683818080007e-05,
+      "loss": 0.7612,
+      "step": 29120
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 6.63413667678833,
+      "learning_rate": 1.9613157448838242e-05,
+      "loss": 0.8763,
+      "step": 29130
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 5.615811347961426,
+      "learning_rate": 1.95976332582096e-05,
+      "loss": 0.9577,
+      "step": 29140
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 4.839416027069092,
+      "learning_rate": 1.9582111252472553e-05,
+      "loss": 0.8236,
+      "step": 29150
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 6.097073554992676,
+      "learning_rate": 1.956659143790471e-05,
+      "loss": 1.0098,
+      "step": 29160
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 5.599035263061523,
+      "learning_rate": 1.95510738207828e-05,
+      "loss": 0.7876,
+      "step": 29170
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 6.090539455413818,
+      "learning_rate": 1.9535558407382625e-05,
+      "loss": 0.9625,
+      "step": 29180
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 8.603718757629395,
+      "learning_rate": 1.952004520397914e-05,
+      "loss": 0.7671,
+      "step": 29190
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 5.015973091125488,
+      "learning_rate": 1.9504534216846366e-05,
+      "loss": 0.7476,
+      "step": 29200
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 6.233822822570801,
+      "learning_rate": 1.9489025452257465e-05,
+      "loss": 0.6849,
+      "step": 29210
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 6.099392890930176,
+      "learning_rate": 1.9473518916484663e-05,
+      "loss": 0.9901,
+      "step": 29220
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 4.486176013946533,
+      "learning_rate": 1.9458014615799326e-05,
+      "loss": 0.8393,
+      "step": 29230
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 6.766070365905762,
+      "learning_rate": 1.944251255647188e-05,
+      "loss": 0.6504,
+      "step": 29240
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 6.20212459564209,
+      "learning_rate": 1.9427012744771857e-05,
+      "loss": 0.7169,
+      "step": 29250
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 5.461494445800781,
+      "learning_rate": 1.941151518696789e-05,
+      "loss": 0.9376,
+      "step": 29260
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 8.070244789123535,
+      "learning_rate": 1.9396019889327688e-05,
+      "loss": 0.902,
+      "step": 29270
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 5.939727783203125,
+      "learning_rate": 1.9380526858118043e-05,
+      "loss": 0.8162,
+      "step": 29280
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 7.3278279304504395,
+      "learning_rate": 1.936503609960485e-05,
+      "loss": 0.9636,
+      "step": 29290
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 5.049187183380127,
+      "learning_rate": 1.934954762005306e-05,
+      "loss": 0.8147,
+      "step": 29300
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 6.915448188781738,
+      "learning_rate": 1.933406142572672e-05,
+      "loss": 0.8542,
+      "step": 29310
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 6.744935512542725,
+      "learning_rate": 1.9318577522888955e-05,
+      "loss": 1.0523,
+      "step": 29320
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 6.081545352935791,
+      "learning_rate": 1.9303095917801938e-05,
+      "loss": 0.7159,
+      "step": 29330
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 7.755534648895264,
+      "learning_rate": 1.928761661672695e-05,
+      "loss": 1.0563,
+      "step": 29340
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 6.315848350524902,
+      "learning_rate": 1.92721396259243e-05,
+      "loss": 0.832,
+      "step": 29350
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 5.882720947265625,
+      "learning_rate": 1.9256664951653408e-05,
+      "loss": 0.7928,
+      "step": 29360
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 7.314845085144043,
+      "learning_rate": 1.924119260017271e-05,
+      "loss": 0.9678,
+      "step": 29370
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 5.904716491699219,
+      "learning_rate": 1.922572257773973e-05,
+      "loss": 0.828,
+      "step": 29380
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 6.712629795074463,
+      "learning_rate": 1.9210254890611046e-05,
+      "loss": 0.8875,
+      "step": 29390
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 3.4837889671325684,
+      "learning_rate": 1.91947895450423e-05,
+      "loss": 0.7617,
+      "step": 29400
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 4.6578497886657715,
+      "learning_rate": 1.9179326547288156e-05,
+      "loss": 0.8712,
+      "step": 29410
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 4.334661960601807,
+      "learning_rate": 1.9163865903602374e-05,
+      "loss": 0.6712,
+      "step": 29420
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 5.238563060760498,
+      "learning_rate": 1.9148407620237714e-05,
+      "loss": 0.6902,
+      "step": 29430
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 4.212186813354492,
+      "learning_rate": 1.9132951703446027e-05,
+      "loss": 0.8808,
+      "step": 29440
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 7.599255561828613,
+      "learning_rate": 1.9117498159478163e-05,
+      "loss": 0.7874,
+      "step": 29450
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 5.931928634643555,
+      "learning_rate": 1.910204699458405e-05,
+      "loss": 0.9459,
+      "step": 29460
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 6.603538990020752,
+      "learning_rate": 1.9086598215012645e-05,
+      "loss": 0.6259,
+      "step": 29470
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 7.13905668258667,
+      "learning_rate": 1.907115182701191e-05,
+      "loss": 0.7793,
+      "step": 29480
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 6.705167293548584,
+      "learning_rate": 1.9055707836828887e-05,
+      "loss": 0.9683,
+      "step": 29490
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.421893835067749,
+      "learning_rate": 1.9040266250709608e-05,
+      "loss": 0.7203,
+      "step": 29500
+    },
+    {
+      "epoch": 1.17,
+      "eval_loss": 0.7918692827224731,
+      "eval_runtime": 423.1965,
+      "eval_samples_per_second": 50.145,
+      "eval_steps_per_second": 0.392,
+      "step": 29500
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 6.645442485809326,
+      "learning_rate": 1.9024827074899154e-05,
+      "loss": 0.8229,
+      "step": 29510
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 6.296727657318115,
+      "learning_rate": 1.9009390315641628e-05,
+      "loss": 0.8639,
+      "step": 29520
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 5.694365501403809,
+      "learning_rate": 1.8993955979180155e-05,
+      "loss": 0.8879,
+      "step": 29530
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 4.518229961395264,
+      "learning_rate": 1.8978524071756874e-05,
+      "loss": 0.8397,
+      "step": 29540
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 4.390954494476318,
+      "learning_rate": 1.8963094599612958e-05,
+      "loss": 0.8583,
+      "step": 29550
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 5.893826007843018,
+      "learning_rate": 1.8947667568988563e-05,
+      "loss": 0.8766,
+      "step": 29560
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 9.185746192932129,
+      "learning_rate": 1.8932242986122906e-05,
+      "loss": 0.942,
+      "step": 29570
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 5.973138809204102,
+      "learning_rate": 1.8916820857254158e-05,
+      "loss": 0.7734,
+      "step": 29580
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 7.721798419952393,
+      "learning_rate": 1.890140118861955e-05,
+      "loss": 0.8759,
+      "step": 29590
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 7.129239559173584,
+      "learning_rate": 1.8885983986455277e-05,
+      "loss": 0.8619,
+      "step": 29600
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 8.963841438293457,
+      "learning_rate": 1.887056925699656e-05,
+      "loss": 0.8665,
+      "step": 29610
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 5.192595481872559,
+      "learning_rate": 1.885515700647762e-05,
+      "loss": 0.7732,
+      "step": 29620
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 3.9450995922088623,
+      "learning_rate": 1.8839747241131663e-05,
+      "loss": 0.8338,
+      "step": 29630
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 7.913498401641846,
+      "learning_rate": 1.8824339967190895e-05,
+      "loss": 0.8,
+      "step": 29640
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 6.026926517486572,
+      "learning_rate": 1.8808935190886516e-05,
+      "loss": 0.9051,
+      "step": 29650
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 4.25316047668457,
+      "learning_rate": 1.8793532918448714e-05,
+      "loss": 0.7872,
+      "step": 29660
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 4.780691146850586,
+      "learning_rate": 1.8778133156106667e-05,
+      "loss": 0.7659,
+      "step": 29670
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 5.362725734710693,
+      "learning_rate": 1.876273591008854e-05,
+      "loss": 0.9072,
+      "step": 29680
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 7.017155647277832,
+      "learning_rate": 1.8747341186621465e-05,
+      "loss": 0.7351,
+      "step": 29690
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 6.47342586517334,
+      "learning_rate": 1.873194899193158e-05,
+      "loss": 0.7983,
+      "step": 29700
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 4.059345245361328,
+      "learning_rate": 1.871655933224397e-05,
+      "loss": 0.7833,
+      "step": 29710
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 6.868967533111572,
+      "learning_rate": 1.870117221378273e-05,
+      "loss": 0.8934,
+      "step": 29720
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 6.477766036987305,
+      "learning_rate": 1.868578764277089e-05,
+      "loss": 0.7096,
+      "step": 29730
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 6.852750778198242,
+      "learning_rate": 1.867040562543048e-05,
+      "loss": 0.8304,
+      "step": 29740
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 7.0291242599487305,
+      "learning_rate": 1.865502616798247e-05,
+      "loss": 0.8272,
+      "step": 29750
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 5.525005340576172,
+      "learning_rate": 1.863964927664682e-05,
+      "loss": 1.0729,
+      "step": 29760
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 7.355740547180176,
+      "learning_rate": 1.8624274957642446e-05,
+      "loss": 0.9006,
+      "step": 29770
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 5.079577445983887,
+      "learning_rate": 1.8608903217187205e-05,
+      "loss": 0.7143,
+      "step": 29780
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 8.25572681427002,
+      "learning_rate": 1.859353406149794e-05,
+      "loss": 0.8158,
+      "step": 29790
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 4.564329624176025,
+      "learning_rate": 1.857816749679043e-05,
+      "loss": 0.9332,
+      "step": 29800
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 4.347033977508545,
+      "learning_rate": 1.8562803529279406e-05,
+      "loss": 0.7824,
+      "step": 29810
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 7.53162956237793,
+      "learning_rate": 1.8547442165178546e-05,
+      "loss": 0.9971,
+      "step": 29820
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 6.866411209106445,
+      "learning_rate": 1.85320834107005e-05,
+      "loss": 0.7005,
+      "step": 29830
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 5.686083793640137,
+      "learning_rate": 1.851672727205682e-05,
+      "loss": 0.8693,
+      "step": 29840
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 4.767289161682129,
+      "learning_rate": 1.8501373755458055e-05,
+      "loss": 0.6999,
+      "step": 29850
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 5.505954265594482,
+      "learning_rate": 1.8486022867113638e-05,
+      "loss": 0.7507,
+      "step": 29860
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 5.4746551513671875,
+      "learning_rate": 1.847067461323198e-05,
+      "loss": 0.8114,
+      "step": 29870
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 5.502439498901367,
+      "learning_rate": 1.8455329000020395e-05,
+      "loss": 0.9344,
+      "step": 29880
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 6.090731620788574,
+      "learning_rate": 1.8439986033685157e-05,
+      "loss": 0.9626,
+      "step": 29890
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 7.508373260498047,
+      "learning_rate": 1.842464572043145e-05,
+      "loss": 1.0854,
+      "step": 29900
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 5.483656883239746,
+      "learning_rate": 1.8409308066463398e-05,
+      "loss": 0.832,
+      "step": 29910
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 5.950078964233398,
+      "learning_rate": 1.8393973077984043e-05,
+      "loss": 0.8218,
+      "step": 29920
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 5.272325038909912,
+      "learning_rate": 1.8378640761195347e-05,
+      "loss": 0.7641,
+      "step": 29930
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 7.2701616287231445,
+      "learning_rate": 1.83633111222982e-05,
+      "loss": 0.8439,
+      "step": 29940
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 4.734098434448242,
+      "learning_rate": 1.8347984167492384e-05,
+      "loss": 0.9095,
+      "step": 29950
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 7.393209934234619,
+      "learning_rate": 1.8332659902976647e-05,
+      "loss": 0.9594,
+      "step": 29960
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 7.869496822357178,
+      "learning_rate": 1.831733833494858e-05,
+      "loss": 0.9008,
+      "step": 29970
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 4.70518684387207,
+      "learning_rate": 1.8302019469604753e-05,
+      "loss": 0.8089,
+      "step": 29980
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 4.11741304397583,
+      "learning_rate": 1.828670331314058e-05,
+      "loss": 0.6794,
+      "step": 29990
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 4.8165693283081055,
+      "learning_rate": 1.827138987175044e-05,
+      "loss": 0.9994,
+      "step": 30000
+    },
+    {
+      "epoch": 1.19,
+      "eval_loss": 0.7973827719688416,
+      "eval_runtime": 423.3138,
+      "eval_samples_per_second": 50.131,
+      "eval_steps_per_second": 0.392,
+      "step": 30000
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 6.318603038787842,
+      "learning_rate": 1.8256079151627556e-05,
+      "loss": 0.7755,
+      "step": 30010
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 5.472081661224365,
+      "learning_rate": 1.8240771158964097e-05,
+      "loss": 0.7104,
+      "step": 30020
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 5.562380790710449,
+      "learning_rate": 1.8225465899951098e-05,
+      "loss": 0.8252,
+      "step": 30030
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 5.587439060211182,
+      "learning_rate": 1.821016338077851e-05,
+      "loss": 0.8105,
+      "step": 30040
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 5.049861907958984,
+      "learning_rate": 1.819486360763515e-05,
+      "loss": 0.8641,
+      "step": 30050
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 4.803211212158203,
+      "learning_rate": 1.8179566586708763e-05,
+      "loss": 0.837,
+      "step": 30060
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 6.037726879119873,
+      "learning_rate": 1.816427232418594e-05,
+      "loss": 0.8909,
+      "step": 30070
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 7.569865703582764,
+      "learning_rate": 1.8148980826252177e-05,
+      "loss": 0.9072,
+      "step": 30080
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 5.122420787811279,
+      "learning_rate": 1.8133692099091864e-05,
+      "loss": 0.7398,
+      "step": 30090
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 7.280158519744873,
+      "learning_rate": 1.8118406148888235e-05,
+      "loss": 0.9083,
+      "step": 30100
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 12.491714477539062,
+      "learning_rate": 1.8103122981823438e-05,
+      "loss": 0.9428,
+      "step": 30110
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 5.313802719116211,
+      "learning_rate": 1.808784260407847e-05,
+      "loss": 0.7512,
+      "step": 30120
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 8.468059539794922,
+      "learning_rate": 1.807256502183322e-05,
+      "loss": 0.8078,
+      "step": 30130
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 4.894462585449219,
+      "learning_rate": 1.8057290241266422e-05,
+      "loss": 0.7025,
+      "step": 30140
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 6.272170543670654,
+      "learning_rate": 1.8042018268555706e-05,
+      "loss": 0.7609,
+      "step": 30150
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 6.927836894989014,
+      "learning_rate": 1.8026749109877533e-05,
+      "loss": 1.0926,
+      "step": 30160
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 8.739383697509766,
+      "learning_rate": 1.8011482771407263e-05,
+      "loss": 0.8705,
+      "step": 30170
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 6.538280487060547,
+      "learning_rate": 1.799621925931908e-05,
+      "loss": 0.8372,
+      "step": 30180
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 8.374919891357422,
+      "learning_rate": 1.7980958579786054e-05,
+      "loss": 0.7787,
+      "step": 30190
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 4.18115758895874,
+      "learning_rate": 1.796570073898009e-05,
+      "loss": 0.9274,
+      "step": 30200
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 6.322559356689453,
+      "learning_rate": 1.7950445743071952e-05,
+      "loss": 0.8524,
+      "step": 30210
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 8.24243450164795,
+      "learning_rate": 1.7935193598231254e-05,
+      "loss": 0.9719,
+      "step": 30220
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 7.763868808746338,
+      "learning_rate": 1.7919944310626448e-05,
+      "loss": 0.9227,
+      "step": 30230
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 6.839191913604736,
+      "learning_rate": 1.7904697886424858e-05,
+      "loss": 1.0646,
+      "step": 30240
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 3.6695046424865723,
+      "learning_rate": 1.788945433179261e-05,
+      "loss": 0.7005,
+      "step": 30250
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 7.458666801452637,
+      "learning_rate": 1.7874213652894708e-05,
+      "loss": 1.1009,
+      "step": 30260
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 8.100018501281738,
+      "learning_rate": 1.7858975855894956e-05,
+      "loss": 0.8161,
+      "step": 30270
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 4.773386478424072,
+      "learning_rate": 1.7843740946956027e-05,
+      "loss": 0.7935,
+      "step": 30280
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 6.907776832580566,
+      "learning_rate": 1.78285089322394e-05,
+      "loss": 0.7794,
+      "step": 30290
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.9532277584075928,
+      "learning_rate": 1.7813279817905406e-05,
+      "loss": 0.8231,
+      "step": 30300
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 8.076249122619629,
+      "learning_rate": 1.779805361011317e-05,
+      "loss": 0.7913,
+      "step": 30310
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 7.5910258293151855,
+      "learning_rate": 1.778283031502068e-05,
+      "loss": 0.7813,
+      "step": 30320
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 6.7611985206604,
+      "learning_rate": 1.7767609938784725e-05,
+      "loss": 0.7992,
+      "step": 30330
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 5.98903751373291,
+      "learning_rate": 1.7752392487560914e-05,
+      "loss": 0.9384,
+      "step": 30340
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 5.45658540725708,
+      "learning_rate": 1.7737177967503676e-05,
+      "loss": 0.8053,
+      "step": 30350
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 6.881286144256592,
+      "learning_rate": 1.772196638476626e-05,
+      "loss": 0.8636,
+      "step": 30360
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 5.958700656890869,
+      "learning_rate": 1.7706757745500704e-05,
+      "loss": 0.9702,
+      "step": 30370
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 6.530422687530518,
+      "learning_rate": 1.7691552055857886e-05,
+      "loss": 0.9894,
+      "step": 30380
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 7.243199348449707,
+      "learning_rate": 1.7676349321987486e-05,
+      "loss": 0.8446,
+      "step": 30390
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 6.714419364929199,
+      "learning_rate": 1.7661149550037965e-05,
+      "loss": 0.852,
+      "step": 30400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 6.448650360107422,
+      "learning_rate": 1.7645952746156614e-05,
+      "loss": 1.0304,
+      "step": 30410
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 6.8941144943237305,
+      "learning_rate": 1.7630758916489502e-05,
+      "loss": 1.05,
+      "step": 30420
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 6.770958423614502,
+      "learning_rate": 1.7615568067181514e-05,
+      "loss": 0.8025,
+      "step": 30430
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 8.481040000915527,
+      "learning_rate": 1.7600380204376304e-05,
+      "loss": 1.0195,
+      "step": 30440
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 6.762326717376709,
+      "learning_rate": 1.7585195334216357e-05,
+      "loss": 0.8358,
+      "step": 30450
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 5.718352317810059,
+      "learning_rate": 1.7570013462842904e-05,
+      "loss": 0.941,
+      "step": 30460
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 6.6078715324401855,
+      "learning_rate": 1.7554834596395996e-05,
+      "loss": 0.8578,
+      "step": 30470
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 7.267546653747559,
+      "learning_rate": 1.7539658741014454e-05,
+      "loss": 0.8016,
+      "step": 30480
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 4.832851886749268,
+      "learning_rate": 1.7524485902835888e-05,
+      "loss": 0.7426,
+      "step": 30490
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 7.524191856384277,
+      "learning_rate": 1.750931608799667e-05,
+      "loss": 0.679,
+      "step": 30500
+    },
+    {
+      "epoch": 1.21,
+      "eval_loss": 0.8014847040176392,
+      "eval_runtime": 423.598,
+      "eval_samples_per_second": 50.097,
+      "eval_steps_per_second": 0.392,
+      "step": 30500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 50400,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}