diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7034 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001,
+      "grad_norm": 28.75,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 6.1232,
+      "step": 10
+    },
+    {
+      "epoch": 0.002,
+      "grad_norm": 26.5,
+      "learning_rate": 2.1111111111111113e-07,
+      "loss": 5.9908,
+      "step": 20
+    },
+    {
+      "epoch": 0.003,
+      "grad_norm": 27.625,
+      "learning_rate": 3.2222222222222227e-07,
+      "loss": 6.8032,
+      "step": 30
+    },
+    {
+      "epoch": 0.004,
+      "grad_norm": 28.5,
+      "learning_rate": 4.333333333333334e-07,
+      "loss": 6.3989,
+      "step": 40
+    },
+    {
+      "epoch": 0.005,
+      "grad_norm": 34.25,
+      "learning_rate": 5.444444444444444e-07,
+      "loss": 6.5616,
+      "step": 50
+    },
+    {
+      "epoch": 0.006,
+      "grad_norm": 25.5,
+      "learning_rate": 6.555555555555556e-07,
+      "loss": 5.6445,
+      "step": 60
+    },
+    {
+      "epoch": 0.007,
+      "grad_norm": 29.75,
+      "learning_rate": 7.666666666666667e-07,
+      "loss": 7.1952,
+      "step": 70
+    },
+    {
+      "epoch": 0.008,
+      "grad_norm": 43.0,
+      "learning_rate": 8.777777777777778e-07,
+      "loss": 6.3381,
+      "step": 80
+    },
+    {
+      "epoch": 0.009,
+      "grad_norm": 28.625,
+      "learning_rate": 9.88888888888889e-07,
+      "loss": 6.4349,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 31.125,
+      "learning_rate": 1.1e-06,
+      "loss": 5.9032,
+      "step": 100
+    },
+    {
+      "epoch": 0.011,
+      "grad_norm": 27.0,
+      "learning_rate": 1.2111111111111111e-06,
+      "loss": 6.0509,
+      "step": 110
+    },
+    {
+      "epoch": 0.012,
+      "grad_norm": 25.125,
+      "learning_rate": 1.3222222222222222e-06,
+      "loss": 6.1967,
+      "step": 120
+    },
+    {
+      "epoch": 0.013,
+      "grad_norm": 28.5,
+      "learning_rate": 1.4333333333333335e-06,
+      "loss": 6.5403,
+      "step": 130
+    },
+    {
+      "epoch": 0.014,
+      "grad_norm": 23.0,
+      "learning_rate": 1.5444444444444446e-06,
+      "loss": 6.4831,
+      "step": 140
+    },
+    {
+      "epoch": 0.015,
+      "grad_norm": 23.25,
+      "learning_rate": 1.6555555555555559e-06,
+      "loss": 6.0029,
+      "step": 150
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 27.125,
+      "learning_rate": 1.7666666666666668e-06,
+      "loss": 5.4189,
+      "step": 160
+    },
+    {
+      "epoch": 0.017,
+      "grad_norm": 21.375,
+      "learning_rate": 1.8777777777777778e-06,
+      "loss": 6.2558,
+      "step": 170
+    },
+    {
+      "epoch": 0.018,
+      "grad_norm": 23.625,
+      "learning_rate": 1.988888888888889e-06,
+      "loss": 6.1538,
+      "step": 180
+    },
+    {
+      "epoch": 0.019,
+      "grad_norm": 21.0,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 5.973,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 27.625,
+      "learning_rate": 2.2111111111111113e-06,
+      "loss": 6.0803,
+      "step": 200
+    },
+    {
+      "epoch": 0.021,
+      "grad_norm": 25.375,
+      "learning_rate": 2.3222222222222224e-06,
+      "loss": 5.5419,
+      "step": 210
+    },
+    {
+      "epoch": 0.022,
+      "grad_norm": 25.75,
+      "learning_rate": 2.4333333333333335e-06,
+      "loss": 5.7562,
+      "step": 220
+    },
+    {
+      "epoch": 0.023,
+      "grad_norm": 26.25,
+      "learning_rate": 2.5444444444444446e-06,
+      "loss": 5.7623,
+      "step": 230
+    },
+    {
+      "epoch": 0.024,
+      "grad_norm": 29.125,
+      "learning_rate": 2.6555555555555556e-06,
+      "loss": 4.8377,
+      "step": 240
+    },
+    {
+      "epoch": 0.025,
+      "grad_norm": 18.25,
+      "learning_rate": 2.766666666666667e-06,
+      "loss": 4.6085,
+      "step": 250
+    },
+    {
+      "epoch": 0.026,
+      "grad_norm": 16.75,
+      "learning_rate": 2.8777777777777782e-06,
+      "loss": 4.6791,
+      "step": 260
+    },
+    {
+      "epoch": 0.027,
+      "grad_norm": 18.5,
+      "learning_rate": 2.988888888888889e-06,
+      "loss": 5.1719,
+      "step": 270
+    },
+    {
+      "epoch": 0.028,
+      "grad_norm": 15.25,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 4.7122,
+      "step": 280
+    },
+    {
+      "epoch": 0.029,
+      "grad_norm": 16.875,
+      "learning_rate": 3.2111111111111115e-06,
+      "loss": 4.2327,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 15.6875,
+      "learning_rate": 3.322222222222222e-06,
+      "loss": 4.3507,
+      "step": 300
+    },
+    {
+      "epoch": 0.031,
+      "grad_norm": 10.875,
+      "learning_rate": 3.4333333333333336e-06,
+      "loss": 3.6952,
+      "step": 310
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 14.3125,
+      "learning_rate": 3.5444444444444447e-06,
+      "loss": 3.9662,
+      "step": 320
+    },
+    {
+      "epoch": 0.033,
+      "grad_norm": 14.3125,
+      "learning_rate": 3.6555555555555562e-06,
+      "loss": 4.1358,
+      "step": 330
+    },
+    {
+      "epoch": 0.034,
+      "grad_norm": 12.625,
+      "learning_rate": 3.766666666666667e-06,
+      "loss": 3.9546,
+      "step": 340
+    },
+    {
+      "epoch": 0.035,
+      "grad_norm": 16.625,
+      "learning_rate": 3.877777777777778e-06,
+      "loss": 3.9011,
+      "step": 350
+    },
+    {
+      "epoch": 0.036,
+      "grad_norm": 23.5,
+      "learning_rate": 3.9888888888888895e-06,
+      "loss": 4.3082,
+      "step": 360
+    },
+    {
+      "epoch": 0.037,
+      "grad_norm": 19.375,
+      "learning_rate": 4.1e-06,
+      "loss": 4.4547,
+      "step": 370
+    },
+    {
+      "epoch": 0.038,
+      "grad_norm": 15.375,
+      "learning_rate": 4.211111111111112e-06,
+      "loss": 3.6685,
+      "step": 380
+    },
+    {
+      "epoch": 0.039,
+      "grad_norm": 20.125,
+      "learning_rate": 4.322222222222223e-06,
+      "loss": 3.739,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 10.4375,
+      "learning_rate": 4.433333333333334e-06,
+      "loss": 3.3964,
+      "step": 400
+    },
+    {
+      "epoch": 0.041,
+      "grad_norm": 12.75,
+      "learning_rate": 4.544444444444445e-06,
+      "loss": 5.7306,
+      "step": 410
+    },
+    {
+      "epoch": 0.042,
+      "grad_norm": 11.6875,
+      "learning_rate": 4.655555555555556e-06,
+      "loss": 3.6601,
+      "step": 420
+    },
+    {
+      "epoch": 0.043,
+      "grad_norm": 11.25,
+      "learning_rate": 4.766666666666667e-06,
+      "loss": 3.1638,
+      "step": 430
+    },
+    {
+      "epoch": 0.044,
+      "grad_norm": 16.625,
+      "learning_rate": 4.877777777777778e-06,
+      "loss": 3.5302,
+      "step": 440
+    },
+    {
+      "epoch": 0.045,
+      "grad_norm": 9.5625,
+      "learning_rate": 4.988888888888889e-06,
+      "loss": 2.9029,
+      "step": 450
+    },
+    {
+      "epoch": 0.046,
+      "grad_norm": 15.9375,
+      "learning_rate": 5.1e-06,
+      "loss": 3.3652,
+      "step": 460
+    },
+    {
+      "epoch": 0.047,
+      "grad_norm": 16.75,
+      "learning_rate": 5.211111111111111e-06,
+      "loss": 3.267,
+      "step": 470
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 11.25,
+      "learning_rate": 5.322222222222223e-06,
+      "loss": 3.5716,
+      "step": 480
+    },
+    {
+      "epoch": 0.049,
+      "grad_norm": 11.9375,
+      "learning_rate": 5.4333333333333335e-06,
+      "loss": 3.0513,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 10.1875,
+      "learning_rate": 5.544444444444445e-06,
+      "loss": 2.7017,
+      "step": 500
+    },
+    {
+      "epoch": 0.051,
+      "grad_norm": 17.0,
+      "learning_rate": 5.6555555555555566e-06,
+      "loss": 3.0088,
+      "step": 510
+    },
+    {
+      "epoch": 0.052,
+      "grad_norm": 10.25,
+      "learning_rate": 5.766666666666667e-06,
+      "loss": 2.8785,
+      "step": 520
+    },
+    {
+      "epoch": 0.053,
+      "grad_norm": 14.25,
+      "learning_rate": 5.877777777777778e-06,
+      "loss": 3.5989,
+      "step": 530
+    },
+    {
+      "epoch": 0.054,
+      "grad_norm": 14.0,
+      "learning_rate": 5.98888888888889e-06,
+      "loss": 2.866,
+      "step": 540
+    },
+    {
+      "epoch": 0.055,
+      "grad_norm": 16.125,
+      "learning_rate": 6.1e-06,
+      "loss": 3.1185,
+      "step": 550
+    },
+    {
+      "epoch": 0.056,
+      "grad_norm": 10.0625,
+      "learning_rate": 6.211111111111111e-06,
+      "loss": 2.942,
+      "step": 560
+    },
+    {
+      "epoch": 0.057,
+      "grad_norm": 14.5,
+      "learning_rate": 6.322222222222223e-06,
+      "loss": 2.7497,
+      "step": 570
+    },
+    {
+      "epoch": 0.058,
+      "grad_norm": 13.0625,
+      "learning_rate": 6.433333333333333e-06,
+      "loss": 2.774,
+      "step": 580
+    },
+    {
+      "epoch": 0.059,
+      "grad_norm": 14.5625,
+      "learning_rate": 6.544444444444445e-06,
+      "loss": 2.8859,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 17.25,
+      "learning_rate": 6.655555555555556e-06,
+      "loss": 2.7014,
+      "step": 600
+    },
+    {
+      "epoch": 0.061,
+      "grad_norm": 10.875,
+      "learning_rate": 6.7666666666666665e-06,
+      "loss": 2.2977,
+      "step": 610
+    },
+    {
+      "epoch": 0.062,
+      "grad_norm": 12.0625,
+      "learning_rate": 6.8777777777777785e-06,
+      "loss": 2.5887,
+      "step": 620
+    },
+    {
+      "epoch": 0.063,
+      "grad_norm": 10.5,
+      "learning_rate": 6.9888888888888895e-06,
+      "loss": 2.4003,
+      "step": 630
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 10.6875,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 2.6738,
+      "step": 640
+    },
+    {
+      "epoch": 0.065,
+      "grad_norm": 15.0625,
+      "learning_rate": 7.211111111111112e-06,
+      "loss": 2.3186,
+      "step": 650
+    },
+    {
+      "epoch": 0.066,
+      "grad_norm": 12.5,
+      "learning_rate": 7.322222222222223e-06,
+      "loss": 2.3446,
+      "step": 660
+    },
+    {
+      "epoch": 0.067,
+      "grad_norm": 21.75,
+      "learning_rate": 7.433333333333334e-06,
+      "loss": 2.4929,
+      "step": 670
+    },
+    {
+      "epoch": 0.068,
+      "grad_norm": 11.4375,
+      "learning_rate": 7.544444444444445e-06,
+      "loss": 2.4428,
+      "step": 680
+    },
+    {
+      "epoch": 0.069,
+      "grad_norm": 10.125,
+      "learning_rate": 7.655555555555556e-06,
+      "loss": 3.7158,
+      "step": 690
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 15.375,
+      "learning_rate": 7.766666666666666e-06,
+      "loss": 2.0719,
+      "step": 700
+    },
+    {
+      "epoch": 0.071,
+      "grad_norm": 12.5625,
+      "learning_rate": 7.877777777777778e-06,
+      "loss": 2.3798,
+      "step": 710
+    },
+    {
+      "epoch": 0.072,
+      "grad_norm": 11.25,
+      "learning_rate": 7.98888888888889e-06,
+      "loss": 2.039,
+      "step": 720
+    },
+    {
+      "epoch": 0.073,
+      "grad_norm": 14.75,
+      "learning_rate": 8.1e-06,
+      "loss": 2.2863,
+      "step": 730
+    },
+    {
+      "epoch": 0.074,
+      "grad_norm": 24.625,
+      "learning_rate": 8.211111111111112e-06,
+      "loss": 2.5293,
+      "step": 740
+    },
+    {
+      "epoch": 0.075,
+      "grad_norm": 15.875,
+      "learning_rate": 8.322222222222223e-06,
+      "loss": 2.1529,
+      "step": 750
+    },
+    {
+      "epoch": 0.076,
+      "grad_norm": 19.375,
+      "learning_rate": 8.433333333333334e-06,
+      "loss": 2.3109,
+      "step": 760
+    },
+    {
+      "epoch": 0.077,
+      "grad_norm": 12.125,
+      "learning_rate": 8.544444444444445e-06,
+      "loss": 2.1811,
+      "step": 770
+    },
+    {
+      "epoch": 0.078,
+      "grad_norm": 18.0,
+      "learning_rate": 8.655555555555557e-06,
+      "loss": 2.1632,
+      "step": 780
+    },
+    {
+      "epoch": 0.079,
+      "grad_norm": 11.1875,
+      "learning_rate": 8.766666666666669e-06,
+      "loss": 2.3427,
+      "step": 790
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 8.5,
+      "learning_rate": 8.877777777777779e-06,
+      "loss": 2.4968,
+      "step": 800
+    },
+    {
+      "epoch": 0.081,
+      "grad_norm": 9.75,
+      "learning_rate": 8.988888888888889e-06,
+      "loss": 2.8265,
+      "step": 810
+    },
+    {
+      "epoch": 0.082,
+      "grad_norm": 12.625,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 2.2349,
+      "step": 820
+    },
+    {
+      "epoch": 0.083,
+      "grad_norm": 12.3125,
+      "learning_rate": 9.211111111111111e-06,
+      "loss": 2.3352,
+      "step": 830
+    },
+    {
+      "epoch": 0.084,
+      "grad_norm": 8.9375,
+      "learning_rate": 9.322222222222223e-06,
+      "loss": 1.9891,
+      "step": 840
+    },
+    {
+      "epoch": 0.085,
+      "grad_norm": 20.25,
+      "learning_rate": 9.433333333333335e-06,
+      "loss": 2.4215,
+      "step": 850
+    },
+    {
+      "epoch": 0.086,
+      "grad_norm": 23.875,
+      "learning_rate": 9.544444444444445e-06,
+      "loss": 2.6832,
+      "step": 860
+    },
+    {
+      "epoch": 0.087,
+      "grad_norm": 16.0,
+      "learning_rate": 9.655555555555556e-06,
+      "loss": 2.1498,
+      "step": 870
+    },
+    {
+      "epoch": 0.088,
+      "grad_norm": 15.6875,
+      "learning_rate": 9.766666666666667e-06,
+      "loss": 2.1489,
+      "step": 880
+    },
+    {
+      "epoch": 0.089,
+      "grad_norm": 13.625,
+      "learning_rate": 9.877777777777778e-06,
+      "loss": 2.2199,
+      "step": 890
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 10.3125,
+      "learning_rate": 9.98888888888889e-06,
+      "loss": 1.9739,
+      "step": 900
+    },
+    {
+      "epoch": 0.091,
+      "grad_norm": 15.875,
+      "learning_rate": 9.999997639854591e-06,
+      "loss": 2.2215,
+      "step": 910
+    },
+    {
+      "epoch": 0.092,
+      "grad_norm": 15.0625,
+      "learning_rate": 9.999989481330112e-06,
+      "loss": 2.2376,
+      "step": 920
+    },
+    {
+      "epoch": 0.093,
+      "grad_norm": 16.0,
+      "learning_rate": 9.99997549529847e-06,
+      "loss": 2.2546,
+      "step": 930
+    },
+    {
+      "epoch": 0.094,
+      "grad_norm": 20.875,
+      "learning_rate": 9.999955681775966e-06,
+      "loss": 2.1029,
+      "step": 940
+    },
+    {
+      "epoch": 0.095,
+      "grad_norm": 11.0625,
+      "learning_rate": 9.999930040785695e-06,
+      "loss": 2.3505,
+      "step": 950
+    },
+    {
+      "epoch": 0.096,
+      "grad_norm": 17.625,
+      "learning_rate": 9.99989857235754e-06,
+      "loss": 2.2428,
+      "step": 960
+    },
+    {
+      "epoch": 0.097,
+      "grad_norm": 16.875,
+      "learning_rate": 9.999861276528177e-06,
+      "loss": 2.19,
+      "step": 970
+    },
+    {
+      "epoch": 0.098,
+      "grad_norm": 14.8125,
+      "learning_rate": 9.999818153341078e-06,
+      "loss": 2.2612,
+      "step": 980
+    },
+    {
+      "epoch": 0.099,
+      "grad_norm": 8.75,
+      "learning_rate": 9.999769202846499e-06,
+      "loss": 1.8096,
+      "step": 990
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 14.4375,
+      "learning_rate": 9.999714425101493e-06,
+      "loss": 2.3079,
+      "step": 1000
+    },
+    {
+      "epoch": 0.101,
+      "grad_norm": 13.5625,
+      "learning_rate": 9.999653820169904e-06,
+      "loss": 2.4223,
+      "step": 1010
+    },
+    {
+      "epoch": 0.102,
+      "grad_norm": 13.8125,
+      "learning_rate": 9.99958738812237e-06,
+      "loss": 2.2387,
+      "step": 1020
+    },
+    {
+      "epoch": 0.103,
+      "grad_norm": 33.75,
+      "learning_rate": 9.999515129036312e-06,
+      "loss": 2.3369,
+      "step": 1030
+    },
+    {
+      "epoch": 0.104,
+      "grad_norm": 17.25,
+      "learning_rate": 9.999437042995954e-06,
+      "loss": 2.0494,
+      "step": 1040
+    },
+    {
+      "epoch": 0.105,
+      "grad_norm": 9.1875,
+      "learning_rate": 9.999353130092302e-06,
+      "loss": 1.8708,
+      "step": 1050
+    },
+    {
+      "epoch": 0.106,
+      "grad_norm": 13.9375,
+      "learning_rate": 9.999263390423157e-06,
+      "loss": 2.128,
+      "step": 1060
+    },
+    {
+      "epoch": 0.107,
+      "grad_norm": 17.5,
+      "learning_rate": 9.999167824093114e-06,
+      "loss": 2.1933,
+      "step": 1070
+    },
+    {
+      "epoch": 0.108,
+      "grad_norm": 13.25,
+      "learning_rate": 9.999066431213553e-06,
+      "loss": 1.8922,
+      "step": 1080
+    },
+    {
+      "epoch": 0.109,
+      "grad_norm": 13.75,
+      "learning_rate": 9.998959211902647e-06,
+      "loss": 2.1718,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 13.375,
+      "learning_rate": 9.998846166285365e-06,
+      "loss": 2.0742,
+      "step": 1100
+    },
+    {
+      "epoch": 0.111,
+      "grad_norm": 12.75,
+      "learning_rate": 9.998727294493457e-06,
+      "loss": 1.8825,
+      "step": 1110
+    },
+    {
+      "epoch": 0.112,
+      "grad_norm": 14.4375,
+      "learning_rate": 9.998602596665471e-06,
+      "loss": 2.0506,
+      "step": 1120
+    },
+    {
+      "epoch": 0.113,
+      "grad_norm": 13.3125,
+      "learning_rate": 9.998472072946746e-06,
+      "loss": 2.3854,
+      "step": 1130
+    },
+    {
+      "epoch": 0.114,
+      "grad_norm": 10.5625,
+      "learning_rate": 9.9983357234894e-06,
+      "loss": 1.7009,
+      "step": 1140
+    },
+    {
+      "epoch": 0.115,
+      "grad_norm": 13.1875,
+      "learning_rate": 9.998193548452356e-06,
+      "loss": 2.2576,
+      "step": 1150
+    },
+    {
+      "epoch": 0.116,
+      "grad_norm": 12.375,
+      "learning_rate": 9.998045548001316e-06,
+      "loss": 1.7733,
+      "step": 1160
+    },
+    {
+      "epoch": 0.117,
+      "grad_norm": 16.625,
+      "learning_rate": 9.997891722308776e-06,
+      "loss": 2.2163,
+      "step": 1170
+    },
+    {
+      "epoch": 0.118,
+      "grad_norm": 21.25,
+      "learning_rate": 9.99773207155402e-06,
+      "loss": 2.5188,
+      "step": 1180
+    },
+    {
+      "epoch": 0.119,
+      "grad_norm": 11.9375,
+      "learning_rate": 9.997566595923123e-06,
+      "loss": 2.1723,
+      "step": 1190
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 10.9375,
+      "learning_rate": 9.997395295608945e-06,
+      "loss": 1.9266,
+      "step": 1200
+    },
+    {
+      "epoch": 0.121,
+      "grad_norm": 16.875,
+      "learning_rate": 9.99721817081114e-06,
+      "loss": 2.0924,
+      "step": 1210
+    },
+    {
+      "epoch": 0.122,
+      "grad_norm": 9.25,
+      "learning_rate": 9.997035221736146e-06,
+      "loss": 1.7556,
+      "step": 1220
+    },
+    {
+      "epoch": 0.123,
+      "grad_norm": 18.5,
+      "learning_rate": 9.99684644859719e-06,
+      "loss": 2.2428,
+      "step": 1230
+    },
+    {
+      "epoch": 0.124,
+      "grad_norm": 16.375,
+      "learning_rate": 9.996651851614292e-06,
+      "loss": 2.4478,
+      "step": 1240
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 9.625,
+      "learning_rate": 9.99645143101425e-06,
+      "loss": 2.0832,
+      "step": 1250
+    },
+    {
+      "epoch": 0.126,
+      "grad_norm": 13.625,
+      "learning_rate": 9.996245187030657e-06,
+      "loss": 2.9663,
+      "step": 1260
+    },
+    {
+      "epoch": 0.127,
+      "grad_norm": 10.875,
+      "learning_rate": 9.996033119903894e-06,
+      "loss": 2.0538,
+      "step": 1270
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 15.125,
+      "learning_rate": 9.995815229881123e-06,
+      "loss": 2.0506,
+      "step": 1280
+    },
+    {
+      "epoch": 0.129,
+      "grad_norm": 13.4375,
+      "learning_rate": 9.995591517216295e-06,
+      "loss": 2.0531,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.995361982170152e-06,
+      "loss": 2.1295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.131,
+      "grad_norm": 17.875,
+      "learning_rate": 9.995126625010212e-06,
+      "loss": 2.0139,
+      "step": 1310
+    },
+    {
+      "epoch": 0.132,
+      "grad_norm": 14.8125,
+      "learning_rate": 9.994885446010791e-06,
+      "loss": 2.0254,
+      "step": 1320
+    },
+    {
+      "epoch": 0.133,
+      "grad_norm": 15.25,
+      "learning_rate": 9.994638445452978e-06,
+      "loss": 2.2371,
+      "step": 1330
+    },
+    {
+      "epoch": 0.134,
+      "grad_norm": 15.4375,
+      "learning_rate": 9.994385623624658e-06,
+      "loss": 2.0127,
+      "step": 1340
+    },
+    {
+      "epoch": 0.135,
+      "grad_norm": 10.625,
+      "learning_rate": 9.994126980820496e-06,
+      "loss": 1.8109,
+      "step": 1350
+    },
+    {
+      "epoch": 0.136,
+      "grad_norm": 21.875,
+      "learning_rate": 9.993862517341937e-06,
+      "loss": 1.9911,
+      "step": 1360
+    },
+    {
+      "epoch": 0.137,
+      "grad_norm": 11.875,
+      "learning_rate": 9.993592233497216e-06,
+      "loss": 2.2054,
+      "step": 1370
+    },
+    {
+      "epoch": 0.138,
+      "grad_norm": 16.0,
+      "learning_rate": 9.993316129601352e-06,
+      "loss": 1.8838,
+      "step": 1380
+    },
+    {
+      "epoch": 0.139,
+      "grad_norm": 13.625,
+      "learning_rate": 9.993034205976141e-06,
+      "loss": 2.1867,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 11.375,
+      "learning_rate": 9.992746462950171e-06,
+      "loss": 2.0658,
+      "step": 1400
+    },
+    {
+      "epoch": 0.141,
+      "grad_norm": 13.9375,
+      "learning_rate": 9.992452900858804e-06,
+      "loss": 1.9625,
+      "step": 1410
+    },
+    {
+      "epoch": 0.142,
+      "grad_norm": 12.5,
+      "learning_rate": 9.99215352004419e-06,
+      "loss": 2.1101,
+      "step": 1420
+    },
+    {
+      "epoch": 0.143,
+      "grad_norm": 13.5625,
+      "learning_rate": 9.991848320855258e-06,
+      "loss": 1.9158,
+      "step": 1430
+    },
+    {
+      "epoch": 0.144,
+      "grad_norm": 13.875,
+      "learning_rate": 9.991537303647717e-06,
+      "loss": 1.9145,
+      "step": 1440
+    },
+    {
+      "epoch": 0.145,
+      "grad_norm": 15.5,
+      "learning_rate": 9.991220468784059e-06,
+      "loss": 1.9446,
+      "step": 1450
+    },
+    {
+      "epoch": 0.146,
+      "grad_norm": 18.75,
+      "learning_rate": 9.99089781663356e-06,
+      "loss": 1.9702,
+      "step": 1460
+    },
+    {
+      "epoch": 0.147,
+      "grad_norm": 11.25,
+      "learning_rate": 9.990569347572268e-06,
+      "loss": 2.0381,
+      "step": 1470
+    },
+    {
+      "epoch": 0.148,
+      "grad_norm": 28.125,
+      "learning_rate": 9.990235061983015e-06,
+      "loss": 2.1657,
+      "step": 1480
+    },
+    {
+      "epoch": 0.149,
+      "grad_norm": 16.625,
+      "learning_rate": 9.989894960255417e-06,
+      "loss": 1.7358,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 17.375,
+      "learning_rate": 9.98954904278586e-06,
+      "loss": 1.9178,
+      "step": 1500
+    },
+    {
+      "epoch": 0.151,
+      "grad_norm": 16.875,
+      "learning_rate": 9.989197309977512e-06,
+      "loss": 1.9564,
+      "step": 1510
+    },
+    {
+      "epoch": 0.152,
+      "grad_norm": 13.8125,
+      "learning_rate": 9.98883976224032e-06,
+      "loss": 2.2242,
+      "step": 1520
+    },
+    {
+      "epoch": 0.153,
+      "grad_norm": 24.625,
+      "learning_rate": 9.988476399991006e-06,
+      "loss": 1.9153,
+      "step": 1530
+    },
+    {
+      "epoch": 0.154,
+      "grad_norm": 21.125,
+      "learning_rate": 9.988107223653072e-06,
+      "loss": 1.9887,
+      "step": 1540
+    },
+    {
+      "epoch": 0.155,
+      "grad_norm": 13.0,
+      "learning_rate": 9.987732233656793e-06,
+      "loss": 1.8001,
+      "step": 1550
+    },
+    {
+      "epoch": 0.156,
+      "grad_norm": 22.125,
+      "learning_rate": 9.987351430439223e-06,
+      "loss": 2.3175,
+      "step": 1560
+    },
+    {
+      "epoch": 0.157,
+      "grad_norm": 8.9375,
+      "learning_rate": 9.986964814444189e-06,
+      "loss": 1.7252,
+      "step": 1570
+    },
+    {
+      "epoch": 0.158,
+      "grad_norm": 12.9375,
+      "learning_rate": 9.98657238612229e-06,
+      "loss": 2.1488,
+      "step": 1580
+    },
+    {
+      "epoch": 0.159,
+      "grad_norm": 11.8125,
+      "learning_rate": 9.986174145930909e-06,
+      "loss": 1.7074,
+      "step": 1590
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 16.875,
+      "learning_rate": 9.985770094334193e-06,
+      "loss": 2.0617,
+      "step": 1600
+    },
+    {
+      "epoch": 0.161,
+      "grad_norm": 14.1875,
+      "learning_rate": 9.985360231803065e-06,
+      "loss": 1.7983,
+      "step": 1610
+    },
+    {
+      "epoch": 0.162,
+      "grad_norm": 9.4375,
+      "learning_rate": 9.984944558815222e-06,
+      "loss": 1.7604,
+      "step": 1620
+    },
+    {
+      "epoch": 0.163,
+      "grad_norm": 18.25,
+      "learning_rate": 9.984523075855133e-06,
+      "loss": 1.9584,
+      "step": 1630
+    },
+    {
+      "epoch": 0.164,
+      "grad_norm": 12.4375,
+      "learning_rate": 9.984095783414037e-06,
+      "loss": 1.882,
+      "step": 1640
+    },
+    {
+      "epoch": 0.165,
+      "grad_norm": 14.0,
+      "learning_rate": 9.983662681989946e-06,
+      "loss": 1.9021,
+      "step": 1650
+    },
+    {
+      "epoch": 0.166,
+      "grad_norm": 13.125,
+      "learning_rate": 9.983223772087642e-06,
+      "loss": 1.9511,
+      "step": 1660
+    },
+    {
+      "epoch": 0.167,
+      "grad_norm": 12.375,
+      "learning_rate": 9.982779054218673e-06,
+      "loss": 1.8299,
+      "step": 1670
+    },
+    {
+      "epoch": 0.168,
+      "grad_norm": 22.125,
+      "learning_rate": 9.982328528901366e-06,
+      "loss": 1.9774,
+      "step": 1680
+    },
+    {
+      "epoch": 0.169,
+      "grad_norm": 12.25,
+      "learning_rate": 9.981872196660803e-06,
+      "loss": 1.8432,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 15.5,
+      "learning_rate": 9.981410058028845e-06,
+      "loss": 1.9688,
+      "step": 1700
+    },
+    {
+      "epoch": 0.171,
+      "grad_norm": 15.25,
+      "learning_rate": 9.980942113544114e-06,
+      "loss": 2.1108,
+      "step": 1710
+    },
+    {
+      "epoch": 0.172,
+      "grad_norm": 8.8125,
+      "learning_rate": 9.980468363752004e-06,
+      "loss": 1.7164,
+      "step": 1720
+    },
+    {
+      "epoch": 0.173,
+      "grad_norm": 11.3125,
+      "learning_rate": 9.979988809204667e-06,
+      "loss": 1.6531,
+      "step": 1730
+    },
+    {
+      "epoch": 0.174,
+      "grad_norm": 14.125,
+      "learning_rate": 9.979503450461034e-06,
+      "loss": 2.0261,
+      "step": 1740
+    },
+    {
+      "epoch": 0.175,
+      "grad_norm": 18.75,
+      "learning_rate": 9.979012288086785e-06,
+      "loss": 1.8215,
+      "step": 1750
+    },
+    {
+      "epoch": 0.176,
+      "grad_norm": 12.8125,
+      "learning_rate": 9.978515322654374e-06,
+      "loss": 2.3499,
+      "step": 1760
+    },
+    {
+      "epoch": 0.177,
+      "grad_norm": 14.75,
+      "learning_rate": 9.978012554743017e-06,
+      "loss": 1.9989,
+      "step": 1770
+    },
+    {
+      "epoch": 0.178,
+      "grad_norm": 14.5,
+      "learning_rate": 9.977503984938692e-06,
+      "loss": 1.9452,
+      "step": 1780
+    },
+    {
+      "epoch": 0.179,
+      "grad_norm": 14.5,
+      "learning_rate": 9.976989613834138e-06,
+      "loss": 1.9518,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 11.8125,
+      "learning_rate": 9.976469442028857e-06,
+      "loss": 1.5835,
+      "step": 1800
+    },
+    {
+      "epoch": 0.181,
+      "grad_norm": 13.4375,
+      "learning_rate": 9.97594347012911e-06,
+      "loss": 1.8152,
+      "step": 1810
+    },
+    {
+      "epoch": 0.182,
+      "grad_norm": 15.5,
+      "learning_rate": 9.975411698747923e-06,
+      "loss": 2.4346,
+      "step": 1820
+    },
+    {
+      "epoch": 0.183,
+      "grad_norm": 13.25,
+      "learning_rate": 9.974874128505075e-06,
+      "loss": 1.8123,
+      "step": 1830
+    },
+    {
+      "epoch": 0.184,
+      "grad_norm": 15.5,
+      "learning_rate": 9.974330760027104e-06,
+      "loss": 2.0298,
+      "step": 1840
+    },
+    {
+      "epoch": 0.185,
+      "grad_norm": 11.25,
+      "learning_rate": 9.973781593947311e-06,
+      "loss": 2.0189,
+      "step": 1850
+    },
+    {
+      "epoch": 0.186,
+      "grad_norm": 12.875,
+      "learning_rate": 9.973226630905753e-06,
+      "loss": 2.0545,
+      "step": 1860
+    },
+    {
+      "epoch": 0.187,
+      "grad_norm": 9.5,
+      "learning_rate": 9.972665871549238e-06,
+      "loss": 1.6011,
+      "step": 1870
+    },
+    {
+      "epoch": 0.188,
+      "grad_norm": 15.1875,
+      "learning_rate": 9.972099316531335e-06,
+      "loss": 2.0133,
+      "step": 1880
+    },
+    {
+      "epoch": 0.189,
+      "grad_norm": 16.25,
+      "learning_rate": 9.971526966512364e-06,
+      "loss": 1.902,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 12.4375,
+      "learning_rate": 9.970948822159402e-06,
+      "loss": 1.8698,
+      "step": 1900
+    },
+    {
+      "epoch": 0.191,
+      "grad_norm": 13.125,
+      "learning_rate": 9.970364884146282e-06,
+      "loss": 1.768,
+      "step": 1910
+    },
+    {
+      "epoch": 0.192,
+      "grad_norm": 7.5,
+      "learning_rate": 9.969775153153583e-06,
+      "loss": 1.8125,
+      "step": 1920
+    },
+    {
+      "epoch": 0.193,
+      "grad_norm": 17.25,
+      "learning_rate": 9.969179629868636e-06,
+      "loss": 1.8715,
+      "step": 1930
+    },
+    {
+      "epoch": 0.194,
+      "grad_norm": 11.5,
+      "learning_rate": 9.968578314985533e-06,
+      "loss": 1.7289,
+      "step": 1940
+    },
+    {
+      "epoch": 0.195,
+      "grad_norm": 19.25,
+      "learning_rate": 9.967971209205102e-06,
+      "loss": 1.8436,
+      "step": 1950
+    },
+    {
+      "epoch": 0.196,
+      "grad_norm": 8.875,
+      "learning_rate": 9.96735831323493e-06,
+      "loss": 1.5802,
+      "step": 1960
+    },
+    {
+      "epoch": 0.197,
+      "grad_norm": 19.5,
+      "learning_rate": 9.966739627789349e-06,
+      "loss": 1.8218,
+      "step": 1970
+    },
+    {
+      "epoch": 0.198,
+      "grad_norm": 8.5,
+      "learning_rate": 9.966115153589439e-06,
+      "loss": 1.7039,
+      "step": 1980
+    },
+    {
+      "epoch": 0.199,
+      "grad_norm": 12.8125,
+      "learning_rate": 9.965484891363029e-06,
+      "loss": 1.7813,
+      "step": 1990
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 17.25,
+      "learning_rate": 9.96484884184469e-06,
+      "loss": 2.357,
+      "step": 2000
+    },
+    {
+      "epoch": 0.201,
+      "grad_norm": 14.625,
+      "learning_rate": 9.964207005775741e-06,
+      "loss": 2.0646,
+      "step": 2010
+    },
+    {
+      "epoch": 0.202,
+      "grad_norm": 15.1875,
+      "learning_rate": 9.963559383904244e-06,
+      "loss": 1.8416,
+      "step": 2020
+    },
+    {
+      "epoch": 0.203,
+      "grad_norm": 14.8125,
+      "learning_rate": 9.962905976985007e-06,
+      "loss": 1.7906,
+      "step": 2030
+    },
+    {
+      "epoch": 0.204,
+      "grad_norm": 15.125,
+      "learning_rate": 9.962246785779575e-06,
+      "loss": 1.7207,
+      "step": 2040
+    },
+    {
+      "epoch": 0.205,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.96158181105624e-06,
+      "loss": 1.9441,
+      "step": 2050
+    },
+    {
+      "epoch": 0.206,
+      "grad_norm": 8.5625,
+      "learning_rate": 9.960911053590033e-06,
+      "loss": 1.5715,
+      "step": 2060
+    },
+    {
+      "epoch": 0.207,
+      "grad_norm": 13.75,
+      "learning_rate": 9.96023451416272e-06,
+      "loss": 1.5676,
+      "step": 2070
+    },
+    {
+      "epoch": 0.208,
+      "grad_norm": 9.9375,
+      "learning_rate": 9.959552193562816e-06,
+      "loss": 1.7409,
+      "step": 2080
+    },
+    {
+      "epoch": 0.209,
+      "grad_norm": 17.625,
+      "learning_rate": 9.958864092585567e-06,
+      "loss": 1.8311,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 10.1875,
+      "learning_rate": 9.958170212032955e-06,
+      "loss": 1.7118,
+      "step": 2100
+    },
+    {
+      "epoch": 0.211,
+      "grad_norm": 12.0,
+      "learning_rate": 9.957470552713701e-06,
+      "loss": 1.7051,
+      "step": 2110
+    },
+    {
+      "epoch": 0.212,
+      "grad_norm": 13.5625,
+      "learning_rate": 9.956765115443264e-06,
+      "loss": 1.7431,
+      "step": 2120
+    },
+    {
+      "epoch": 0.213,
+      "grad_norm": 13.4375,
+      "learning_rate": 9.956053901043831e-06,
+      "loss": 1.8522,
+      "step": 2130
+    },
+    {
+      "epoch": 0.214,
+      "grad_norm": 11.25,
+      "learning_rate": 9.955336910344326e-06,
+      "loss": 1.8196,
+      "step": 2140
+    },
+    {
+      "epoch": 0.215,
+      "grad_norm": 14.5625,
+      "learning_rate": 9.954614144180403e-06,
+      "loss": 1.6825,
+      "step": 2150
+    },
+    {
+      "epoch": 0.216,
+      "grad_norm": 12.1875,
+      "learning_rate": 9.953885603394449e-06,
+      "loss": 1.9492,
+      "step": 2160
+    },
+    {
+      "epoch": 0.217,
+      "grad_norm": 23.625,
+      "learning_rate": 9.953151288835583e-06,
+      "loss": 2.023,
+      "step": 2170
+    },
+    {
+      "epoch": 0.218,
+      "grad_norm": 14.0,
+      "learning_rate": 9.95241120135965e-06,
+      "loss": 1.6492,
+      "step": 2180
+    },
+    {
+      "epoch": 0.219,
+      "grad_norm": 15.125,
+      "learning_rate": 9.951665341829227e-06,
+      "loss": 1.6448,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 10.375,
+      "learning_rate": 9.950913711113613e-06,
+      "loss": 1.7076,
+      "step": 2200
+    },
+    {
+      "epoch": 0.221,
+      "grad_norm": 10.8125,
+      "learning_rate": 9.950156310088836e-06,
+      "loss": 1.9676,
+      "step": 2210
+    },
+    {
+      "epoch": 0.222,
+      "grad_norm": 10.25,
+      "learning_rate": 9.949393139637655e-06,
+      "loss": 1.9527,
+      "step": 2220
+    },
+    {
+      "epoch": 0.223,
+      "grad_norm": 15.625,
+      "learning_rate": 9.948624200649544e-06,
+      "loss": 1.821,
+      "step": 2230
+    },
+    {
+      "epoch": 0.224,
+      "grad_norm": 14.0,
+      "learning_rate": 9.947849494020705e-06,
+      "loss": 1.8325,
+      "step": 2240
+    },
+    {
+      "epoch": 0.225,
+      "grad_norm": 17.625,
+      "learning_rate": 9.947069020654062e-06,
+      "loss": 2.0558,
+      "step": 2250
+    },
+    {
+      "epoch": 0.226,
+      "grad_norm": 16.75,
+      "learning_rate": 9.946282781459259e-06,
+      "loss": 1.8999,
+      "step": 2260
+    },
+    {
+      "epoch": 0.227,
+      "grad_norm": 14.0,
+      "learning_rate": 9.945490777352662e-06,
+      "loss": 1.7714,
+      "step": 2270
+    },
+    {
+      "epoch": 0.228,
+      "grad_norm": 10.5625,
+      "learning_rate": 9.944693009257354e-06,
+      "loss": 1.8318,
+      "step": 2280
+    },
+    {
+      "epoch": 0.229,
+      "grad_norm": 11.3125,
+      "learning_rate": 9.943889478103137e-06,
+      "loss": 1.7152,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 13.25,
+      "learning_rate": 9.94308018482653e-06,
+      "loss": 1.6911,
+      "step": 2300
+    },
+    {
+      "epoch": 0.231,
+      "grad_norm": 16.875,
+      "learning_rate": 9.942265130370767e-06,
+      "loss": 2.1123,
+      "step": 2310
+    },
+    {
+      "epoch": 0.232,
+      "grad_norm": 14.5,
+      "learning_rate": 9.941444315685798e-06,
+      "loss": 1.6969,
+      "step": 2320
+    },
+    {
+      "epoch": 0.233,
+      "grad_norm": 11.9375,
+      "learning_rate": 9.940617741728284e-06,
+      "loss": 2.0962,
+      "step": 2330
+    },
+    {
+      "epoch": 0.234,
+      "grad_norm": 17.125,
+      "learning_rate": 9.939785409461603e-06,
+      "loss": 1.773,
+      "step": 2340
+    },
+    {
+      "epoch": 0.235,
+      "grad_norm": 20.375,
+      "learning_rate": 9.938947319855839e-06,
+      "loss": 2.4351,
+      "step": 2350
+    },
+    {
+      "epoch": 0.236,
+      "grad_norm": 17.0,
+      "learning_rate": 9.93810347388779e-06,
+      "loss": 2.0726,
+      "step": 2360
+    },
+    {
+      "epoch": 0.237,
+      "grad_norm": 13.9375,
+      "learning_rate": 9.93725387254096e-06,
+      "loss": 1.9348,
+      "step": 2370
+    },
+    {
+      "epoch": 0.238,
+      "grad_norm": 12.25,
+      "learning_rate": 9.936398516805565e-06,
+      "loss": 1.8543,
+      "step": 2380
+    },
+    {
+      "epoch": 0.239,
+      "grad_norm": 15.9375,
+      "learning_rate": 9.935537407678522e-06,
+      "loss": 1.5156,
+      "step": 2390
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 14.4375,
+      "learning_rate": 9.934670546163463e-06,
+      "loss": 1.8627,
+      "step": 2400
+    },
+    {
+      "epoch": 0.241,
+      "grad_norm": 16.625,
+      "learning_rate": 9.933797933270711e-06,
+      "loss": 1.695,
+      "step": 2410
+    },
+    {
+      "epoch": 0.242,
+      "grad_norm": 15.4375,
+      "learning_rate": 9.932919570017305e-06,
+      "loss": 1.7248,
+      "step": 2420
+    },
+    {
+      "epoch": 0.243,
+      "grad_norm": 10.125,
+      "learning_rate": 9.93203545742698e-06,
+      "loss": 1.9953,
+      "step": 2430
+    },
+    {
+      "epoch": 0.244,
+      "grad_norm": 13.125,
+      "learning_rate": 9.931145596530172e-06,
+      "loss": 1.6992,
+      "step": 2440
+    },
+    {
+      "epoch": 0.245,
+      "grad_norm": 12.4375,
+      "learning_rate": 9.930249988364016e-06,
+      "loss": 2.1773,
+      "step": 2450
+    },
+    {
+      "epoch": 0.246,
+      "grad_norm": 13.4375,
+      "learning_rate": 9.92934863397235e-06,
+      "loss": 1.9932,
+      "step": 2460
+    },
+    {
+      "epoch": 0.247,
+      "grad_norm": 17.375,
+      "learning_rate": 9.928441534405702e-06,
+      "loss": 1.9955,
+      "step": 2470
+    },
+    {
+      "epoch": 0.248,
+      "grad_norm": 15.0,
+      "learning_rate": 9.927528690721304e-06,
+      "loss": 2.0199,
+      "step": 2480
+    },
+    {
+      "epoch": 0.249,
+      "grad_norm": 19.625,
+      "learning_rate": 9.926610103983076e-06,
+      "loss": 1.6187,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 14.625,
+      "learning_rate": 9.925685775261636e-06,
+      "loss": 2.1515,
+      "step": 2500
+    },
+    {
+      "epoch": 0.251,
+      "grad_norm": 9.4375,
+      "learning_rate": 9.924755705634292e-06,
+      "loss": 1.9842,
+      "step": 2510
+    },
+    {
+      "epoch": 0.252,
+      "grad_norm": 17.125,
+      "learning_rate": 9.923819896185043e-06,
+      "loss": 1.8656,
+      "step": 2520
+    },
+    {
+      "epoch": 0.253,
+      "grad_norm": 12.4375,
+      "learning_rate": 9.92287834800458e-06,
+      "loss": 1.7042,
+      "step": 2530
+    },
+    {
+      "epoch": 0.254,
+      "grad_norm": 12.6875,
+      "learning_rate": 9.921931062190281e-06,
+      "loss": 1.6456,
+      "step": 2540
+    },
+    {
+      "epoch": 0.255,
+      "grad_norm": 15.25,
+      "learning_rate": 9.92097803984621e-06,
+      "loss": 1.7829,
+      "step": 2550
+    },
+    {
+      "epoch": 0.256,
+      "grad_norm": 9.9375,
+      "learning_rate": 9.92001928208312e-06,
+      "loss": 1.4616,
+      "step": 2560
+    },
+    {
+      "epoch": 0.257,
+      "grad_norm": 11.1875,
+      "learning_rate": 9.919054790018445e-06,
+      "loss": 1.8178,
+      "step": 2570
+    },
+    {
+      "epoch": 0.258,
+      "grad_norm": 15.5625,
+      "learning_rate": 9.918084564776308e-06,
+      "loss": 1.6573,
+      "step": 2580
+    },
+    {
+      "epoch": 0.259,
+      "grad_norm": 10.3125,
+      "learning_rate": 9.917108607487504e-06,
+      "loss": 2.0954,
+      "step": 2590
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 13.875,
+      "learning_rate": 9.91612691928952e-06,
+      "loss": 1.8061,
+      "step": 2600
+    },
+    {
+      "epoch": 0.261,
+      "grad_norm": 20.875,
+      "learning_rate": 9.915139501326517e-06,
+      "loss": 1.7086,
+      "step": 2610
+    },
+    {
+      "epoch": 0.262,
+      "grad_norm": 30.375,
+      "learning_rate": 9.914146354749333e-06,
+      "loss": 2.1538,
+      "step": 2620
+    },
+    {
+      "epoch": 0.263,
+      "grad_norm": 12.625,
+      "learning_rate": 9.913147480715482e-06,
+      "loss": 2.1424,
+      "step": 2630
+    },
+    {
+      "epoch": 0.264,
+      "grad_norm": 17.75,
+      "learning_rate": 9.91214288038916e-06,
+      "loss": 2.0394,
+      "step": 2640
+    },
+    {
+      "epoch": 0.265,
+      "grad_norm": 14.0,
+      "learning_rate": 9.911132554941231e-06,
+      "loss": 2.1303,
+      "step": 2650
+    },
+    {
+      "epoch": 0.266,
+      "grad_norm": 14.4375,
+      "learning_rate": 9.910116505549233e-06,
+      "loss": 1.6925,
+      "step": 2660
+    },
+    {
+      "epoch": 0.267,
+      "grad_norm": 13.125,
+      "learning_rate": 9.909094733397375e-06,
+      "loss": 1.7226,
+      "step": 2670
+    },
+    {
+      "epoch": 0.268,
+      "grad_norm": 18.375,
+      "learning_rate": 9.908067239676538e-06,
+      "loss": 1.7001,
+      "step": 2680
+    },
+    {
+      "epoch": 0.269,
+      "grad_norm": 11.125,
+      "learning_rate": 9.907034025584267e-06,
+      "loss": 1.771,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 10.4375,
+      "learning_rate": 9.905995092324781e-06,
+      "loss": 1.8953,
+      "step": 2700
+    },
+    {
+      "epoch": 0.271,
+      "grad_norm": 14.875,
+      "learning_rate": 9.904950441108957e-06,
+      "loss": 1.5565,
+      "step": 2710
+    },
+    {
+      "epoch": 0.272,
+      "grad_norm": 16.625,
+      "learning_rate": 9.903900073154343e-06,
+      "loss": 2.0861,
+      "step": 2720
+    },
+    {
+      "epoch": 0.273,
+      "grad_norm": 14.5,
+      "learning_rate": 9.902843989685144e-06,
+      "loss": 1.744,
+      "step": 2730
+    },
+    {
+      "epoch": 0.274,
+      "grad_norm": 15.5,
+      "learning_rate": 9.901782191932233e-06,
+      "loss": 2.0352,
+      "step": 2740
+    },
+    {
+      "epoch": 0.275,
+      "grad_norm": 19.375,
+      "learning_rate": 9.900714681133135e-06,
+      "loss": 2.1149,
+      "step": 2750
+    },
+    {
+      "epoch": 0.276,
+      "grad_norm": 11.75,
+      "learning_rate": 9.899641458532041e-06,
+      "loss": 1.9053,
+      "step": 2760
+    },
+    {
+      "epoch": 0.277,
+      "grad_norm": 25.875,
+      "learning_rate": 9.898562525379797e-06,
+      "loss": 1.9185,
+      "step": 2770
+    },
+    {
+      "epoch": 0.278,
+      "grad_norm": 11.5,
+      "learning_rate": 9.897477882933902e-06,
+      "loss": 1.6972,
+      "step": 2780
+    },
+    {
+      "epoch": 0.279,
+      "grad_norm": 10.1875,
+      "learning_rate": 9.896387532458509e-06,
+      "loss": 1.6656,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 13.25,
+      "learning_rate": 9.89529147522443e-06,
+      "loss": 1.8109,
+      "step": 2800
+    },
+    {
+      "epoch": 0.281,
+      "grad_norm": 15.125,
+      "learning_rate": 9.894189712509122e-06,
+      "loss": 1.7342,
+      "step": 2810
+    },
+    {
+      "epoch": 0.282,
+      "grad_norm": 14.9375,
+      "learning_rate": 9.893082245596692e-06,
+      "loss": 1.7223,
+      "step": 2820
+    },
+    {
+      "epoch": 0.283,
+      "grad_norm": 13.125,
+      "learning_rate": 9.891969075777902e-06,
+      "loss": 1.7143,
+      "step": 2830
+    },
+    {
+      "epoch": 0.284,
+      "grad_norm": 11.0625,
+      "learning_rate": 9.890850204350148e-06,
+      "loss": 1.6327,
+      "step": 2840
+    },
+    {
+      "epoch": 0.285,
+      "grad_norm": 15.0625,
+      "learning_rate": 9.889725632617486e-06,
+      "loss": 1.5067,
+      "step": 2850
+    },
+    {
+      "epoch": 0.286,
+      "grad_norm": 20.625,
+      "learning_rate": 9.888595361890605e-06,
+      "loss": 1.8323,
+      "step": 2860
+    },
+    {
+      "epoch": 0.287,
+      "grad_norm": 17.625,
+      "learning_rate": 9.887459393486843e-06,
+      "loss": 2.0715,
+      "step": 2870
+    },
+    {
+      "epoch": 0.288,
+      "grad_norm": 12.0625,
+      "learning_rate": 9.886317728730172e-06,
+      "loss": 1.8645,
+      "step": 2880
+    },
+    {
+      "epoch": 0.289,
+      "grad_norm": 26.875,
+      "learning_rate": 9.885170368951209e-06,
+      "loss": 1.8938,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 21.75,
+      "learning_rate": 9.884017315487205e-06,
+      "loss": 1.623,
+      "step": 2900
+    },
+    {
+      "epoch": 0.291,
+      "grad_norm": 15.5,
+      "learning_rate": 9.882858569682048e-06,
+      "loss": 1.877,
+      "step": 2910
+    },
+    {
+      "epoch": 0.292,
+      "grad_norm": 7.6875,
+      "learning_rate": 9.881694132886263e-06,
+      "loss": 1.8081,
+      "step": 2920
+    },
+    {
+      "epoch": 0.293,
+      "grad_norm": 14.5625,
+      "learning_rate": 9.880524006457004e-06,
+      "loss": 1.7907,
+      "step": 2930
+    },
+    {
+      "epoch": 0.294,
+      "grad_norm": 11.1875,
+      "learning_rate": 9.879348191758058e-06,
+      "loss": 1.5902,
+      "step": 2940
+    },
+    {
+      "epoch": 0.295,
+      "grad_norm": 12.9375,
+      "learning_rate": 9.878166690159843e-06,
+      "loss": 1.8425,
+      "step": 2950
+    },
+    {
+      "epoch": 0.296,
+      "grad_norm": 10.8125,
+      "learning_rate": 9.8769795030394e-06,
+      "loss": 1.7599,
+      "step": 2960
+    },
+    {
+      "epoch": 0.297,
+      "grad_norm": 13.9375,
+      "learning_rate": 9.875786631780404e-06,
+      "loss": 1.8008,
+      "step": 2970
+    },
+    {
+      "epoch": 0.298,
+      "grad_norm": 11.3125,
+      "learning_rate": 9.87458807777315e-06,
+      "loss": 1.8831,
+      "step": 2980
+    },
+    {
+      "epoch": 0.299,
+      "grad_norm": 15.9375,
+      "learning_rate": 9.873383842414557e-06,
+      "loss": 1.9775,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 16.375,
+      "learning_rate": 9.872173927108165e-06,
+      "loss": 1.8183,
+      "step": 3000
+    },
+    {
+      "epoch": 0.301,
+      "grad_norm": 17.375,
+      "learning_rate": 9.870958333264137e-06,
+      "loss": 1.8664,
+      "step": 3010
+    },
+    {
+      "epoch": 0.302,
+      "grad_norm": 10.0,
+      "learning_rate": 9.869737062299252e-06,
+      "loss": 1.6826,
+      "step": 3020
+    },
+    {
+      "epoch": 0.303,
+      "grad_norm": 12.625,
+      "learning_rate": 9.868510115636905e-06,
+      "loss": 1.609,
+      "step": 3030
+    },
+    {
+      "epoch": 0.304,
+      "grad_norm": 16.5,
+      "learning_rate": 9.867277494707106e-06,
+      "loss": 2.1546,
+      "step": 3040
+    },
+    {
+      "epoch": 0.305,
+      "grad_norm": 12.4375,
+      "learning_rate": 9.866039200946483e-06,
+      "loss": 1.6152,
+      "step": 3050
+    },
+    {
+      "epoch": 0.306,
+      "grad_norm": 12.25,
+      "learning_rate": 9.864795235798267e-06,
+      "loss": 1.846,
+      "step": 3060
+    },
+    {
+      "epoch": 0.307,
+      "grad_norm": 17.875,
+      "learning_rate": 9.863545600712311e-06,
+      "loss": 1.9616,
+      "step": 3070
+    },
+    {
+      "epoch": 0.308,
+      "grad_norm": 11.6875,
+      "learning_rate": 9.862290297145065e-06,
+      "loss": 2.3813,
+      "step": 3080
+    },
+    {
+      "epoch": 0.309,
+      "grad_norm": 15.375,
+      "learning_rate": 9.86102932655959e-06,
+      "loss": 1.9739,
+      "step": 3090
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 18.75,
+      "learning_rate": 9.859762690425555e-06,
+      "loss": 1.6783,
+      "step": 3100
+    },
+    {
+      "epoch": 0.311,
+      "grad_norm": 14.5,
+      "learning_rate": 9.858490390219227e-06,
+      "loss": 2.2686,
+      "step": 3110
+    },
+    {
+      "epoch": 0.312,
+      "grad_norm": 10.25,
+      "learning_rate": 9.857212427423478e-06,
+      "loss": 1.6573,
+      "step": 3120
+    },
+    {
+      "epoch": 0.313,
+      "grad_norm": 20.25,
+      "learning_rate": 9.855928803527776e-06,
+      "loss": 1.6554,
+      "step": 3130
+    },
+    {
+      "epoch": 0.314,
+      "grad_norm": 22.625,
+      "learning_rate": 9.854639520028196e-06,
+      "loss": 2.0052,
+      "step": 3140
+    },
+    {
+      "epoch": 0.315,
+      "grad_norm": 8.6875,
+      "learning_rate": 9.853344578427393e-06,
+      "loss": 1.7917,
+      "step": 3150
+    },
+    {
+      "epoch": 0.316,
+      "grad_norm": 10.9375,
+      "learning_rate": 9.852043980234635e-06,
+      "loss": 1.7396,
+      "step": 3160
+    },
+    {
+      "epoch": 0.317,
+      "grad_norm": 11.625,
+      "learning_rate": 9.850737726965773e-06,
+      "loss": 1.8837,
+      "step": 3170
+    },
+    {
+      "epoch": 0.318,
+      "grad_norm": 15.1875,
+      "learning_rate": 9.849425820143245e-06,
+      "loss": 1.5989,
+      "step": 3180
+    },
+    {
+      "epoch": 0.319,
+      "grad_norm": 10.5,
+      "learning_rate": 9.84810826129609e-06,
+      "loss": 1.6803,
+      "step": 3190
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 11.875,
+      "learning_rate": 9.846785051959924e-06,
+      "loss": 1.8546,
+      "step": 3200
+    },
+    {
+      "epoch": 0.321,
+      "grad_norm": 12.4375,
+      "learning_rate": 9.845456193676952e-06,
+      "loss": 1.4096,
+      "step": 3210
+    },
+    {
+      "epoch": 0.322,
+      "grad_norm": 11.1875,
+      "learning_rate": 9.844121687995968e-06,
+      "loss": 1.7182,
+      "step": 3220
+    },
+    {
+      "epoch": 0.323,
+      "grad_norm": 12.25,
+      "learning_rate": 9.842781536472339e-06,
+      "loss": 1.7973,
+      "step": 3230
+    },
+    {
+      "epoch": 0.324,
+      "grad_norm": 10.6875,
+      "learning_rate": 9.841435740668018e-06,
+      "loss": 1.7592,
+      "step": 3240
+    },
+    {
+      "epoch": 0.325,
+      "grad_norm": 9.5,
+      "learning_rate": 9.840084302151537e-06,
+      "loss": 1.8247,
+      "step": 3250
+    },
+    {
+      "epoch": 0.326,
+      "grad_norm": 10.75,
+      "learning_rate": 9.838727222498001e-06,
+      "loss": 1.6531,
+      "step": 3260
+    },
+    {
+      "epoch": 0.327,
+      "grad_norm": 16.0,
+      "learning_rate": 9.837364503289093e-06,
+      "loss": 1.5339,
+      "step": 3270
+    },
+    {
+      "epoch": 0.328,
+      "grad_norm": 16.625,
+      "learning_rate": 9.835996146113064e-06,
+      "loss": 1.7767,
+      "step": 3280
+    },
+    {
+      "epoch": 0.329,
+      "grad_norm": 14.5,
+      "learning_rate": 9.834622152564746e-06,
+      "loss": 1.7445,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 15.375,
+      "learning_rate": 9.833242524245528e-06,
+      "loss": 1.6731,
+      "step": 3300
+    },
+    {
+      "epoch": 0.331,
+      "grad_norm": 9.75,
+      "learning_rate": 9.831857262763375e-06,
+      "loss": 1.5566,
+      "step": 3310
+    },
+    {
+      "epoch": 0.332,
+      "grad_norm": 14.3125,
+      "learning_rate": 9.830466369732814e-06,
+      "loss": 1.8057,
+      "step": 3320
+    },
+    {
+      "epoch": 0.333,
+      "grad_norm": 20.25,
+      "learning_rate": 9.829069846774935e-06,
+      "loss": 1.5034,
+      "step": 3330
+    },
+    {
+      "epoch": 0.334,
+      "grad_norm": 16.0,
+      "learning_rate": 9.827667695517393e-06,
+      "loss": 1.6736,
+      "step": 3340
+    },
+    {
+      "epoch": 0.335,
+      "grad_norm": 14.1875,
+      "learning_rate": 9.8262599175944e-06,
+      "loss": 2.0128,
+      "step": 3350
+    },
+    {
+      "epoch": 0.336,
+      "grad_norm": 12.9375,
+      "learning_rate": 9.824846514646726e-06,
+      "loss": 1.9655,
+      "step": 3360
+    },
+    {
+      "epoch": 0.337,
+      "grad_norm": 15.4375,
+      "learning_rate": 9.8234274883217e-06,
+      "loss": 1.8741,
+      "step": 3370
+    },
+    {
+      "epoch": 0.338,
+      "grad_norm": 13.0,
+      "learning_rate": 9.8220028402732e-06,
+      "loss": 1.6363,
+      "step": 3380
+    },
+    {
+      "epoch": 0.339,
+      "grad_norm": 14.625,
+      "learning_rate": 9.820572572161658e-06,
+      "loss": 2.0062,
+      "step": 3390
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 16.375,
+      "learning_rate": 9.819136685654059e-06,
+      "loss": 1.7753,
+      "step": 3400
+    },
+    {
+      "epoch": 0.341,
+      "grad_norm": 18.625,
+      "learning_rate": 9.817695182423935e-06,
+      "loss": 1.7578,
+      "step": 3410
+    },
+    {
+      "epoch": 0.342,
+      "grad_norm": 17.375,
+      "learning_rate": 9.81624806415136e-06,
+      "loss": 1.6827,
+      "step": 3420
+    },
+    {
+      "epoch": 0.343,
+      "grad_norm": 11.625,
+      "learning_rate": 9.814795332522962e-06,
+      "loss": 1.5738,
+      "step": 3430
+    },
+    {
+      "epoch": 0.344,
+      "grad_norm": 13.0625,
+      "learning_rate": 9.813336989231901e-06,
+      "loss": 1.8086,
+      "step": 3440
+    },
+    {
+      "epoch": 0.345,
+      "grad_norm": 15.1875,
+      "learning_rate": 9.811873035977883e-06,
+      "loss": 1.8446,
+      "step": 3450
+    },
+    {
+      "epoch": 0.346,
+      "grad_norm": 12.5,
+      "learning_rate": 9.810403474467149e-06,
+      "loss": 1.5977,
+      "step": 3460
+    },
+    {
+      "epoch": 0.347,
+      "grad_norm": 14.25,
+      "learning_rate": 9.808928306412482e-06,
+      "loss": 1.6242,
+      "step": 3470
+    },
+    {
+      "epoch": 0.348,
+      "grad_norm": 13.75,
+      "learning_rate": 9.807447533533196e-06,
+      "loss": 1.8395,
+      "step": 3480
+    },
+    {
+      "epoch": 0.349,
+      "grad_norm": 14.875,
+      "learning_rate": 9.805961157555134e-06,
+      "loss": 1.8916,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 15.875,
+      "learning_rate": 9.804469180210677e-06,
+      "loss": 2.4153,
+      "step": 3500
+    },
+    {
+      "epoch": 0.351,
+      "grad_norm": 15.25,
+      "learning_rate": 9.802971603238726e-06,
+      "loss": 1.6078,
+      "step": 3510
+    },
+    {
+      "epoch": 0.352,
+      "grad_norm": 15.6875,
+      "learning_rate": 9.801468428384716e-06,
+      "loss": 1.7768,
+      "step": 3520
+    },
+    {
+      "epoch": 0.353,
+      "grad_norm": 20.0,
+      "learning_rate": 9.799959657400604e-06,
+      "loss": 1.9628,
+      "step": 3530
+    },
+    {
+      "epoch": 0.354,
+      "grad_norm": 13.5,
+      "learning_rate": 9.798445292044865e-06,
+      "loss": 1.6986,
+      "step": 3540
+    },
+    {
+      "epoch": 0.355,
+      "grad_norm": 25.875,
+      "learning_rate": 9.796925334082501e-06,
+      "loss": 2.0916,
+      "step": 3550
+    },
+    {
+      "epoch": 0.356,
+      "grad_norm": 12.5625,
+      "learning_rate": 9.795399785285026e-06,
+      "loss": 1.8497,
+      "step": 3560
+    },
+    {
+      "epoch": 0.357,
+      "grad_norm": 11.125,
+      "learning_rate": 9.793868647430476e-06,
+      "loss": 1.9405,
+      "step": 3570
+    },
+    {
+      "epoch": 0.358,
+      "grad_norm": 15.625,
+      "learning_rate": 9.792331922303396e-06,
+      "loss": 1.9214,
+      "step": 3580
+    },
+    {
+      "epoch": 0.359,
+      "grad_norm": 15.3125,
+      "learning_rate": 9.790789611694846e-06,
+      "loss": 1.6476,
+      "step": 3590
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 10.9375,
+      "learning_rate": 9.789241717402396e-06,
+      "loss": 1.5909,
+      "step": 3600
+    },
+    {
+      "epoch": 0.361,
+      "grad_norm": 16.25,
+      "learning_rate": 9.78768824123012e-06,
+      "loss": 1.8808,
+      "step": 3610
+    },
+    {
+      "epoch": 0.362,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.786129184988601e-06,
+      "loss": 1.8502,
+      "step": 3620
+    },
+    {
+      "epoch": 0.363,
+      "grad_norm": 15.75,
+      "learning_rate": 9.78456455049493e-06,
+      "loss": 1.7812,
+      "step": 3630
+    },
+    {
+      "epoch": 0.364,
+      "grad_norm": 13.5625,
+      "learning_rate": 9.782994339572688e-06,
+      "loss": 1.6957,
+      "step": 3640
+    },
+    {
+      "epoch": 0.365,
+      "grad_norm": 16.25,
+      "learning_rate": 9.781418554051964e-06,
+      "loss": 1.9353,
+      "step": 3650
+    },
+    {
+      "epoch": 0.366,
+      "grad_norm": 12.125,
+      "learning_rate": 9.779837195769345e-06,
+      "loss": 1.5091,
+      "step": 3660
+    },
+    {
+      "epoch": 0.367,
+      "grad_norm": 15.4375,
+      "learning_rate": 9.778250266567908e-06,
+      "loss": 1.9664,
+      "step": 3670
+    },
+    {
+      "epoch": 0.368,
+      "grad_norm": 18.25,
+      "learning_rate": 9.776657768297225e-06,
+      "loss": 1.9983,
+      "step": 3680
+    },
+    {
+      "epoch": 0.369,
+      "grad_norm": 10.8125,
+      "learning_rate": 9.775059702813357e-06,
+      "loss": 1.9459,
+      "step": 3690
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 11.3125,
+      "learning_rate": 9.77345607197886e-06,
+      "loss": 1.8148,
+      "step": 3700
+    },
+    {
+      "epoch": 0.371,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.771846877662768e-06,
+      "loss": 1.773,
+      "step": 3710
+    },
+    {
+      "epoch": 0.372,
+      "grad_norm": 14.25,
+      "learning_rate": 9.770232121740606e-06,
+      "loss": 1.9922,
+      "step": 3720
+    },
+    {
+      "epoch": 0.373,
+      "grad_norm": 13.0625,
+      "learning_rate": 9.768611806094376e-06,
+      "loss": 1.6299,
+      "step": 3730
+    },
+    {
+      "epoch": 0.374,
+      "grad_norm": 11.875,
+      "learning_rate": 9.766985932612566e-06,
+      "loss": 1.8018,
+      "step": 3740
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 14.8125,
+      "learning_rate": 9.765354503190133e-06,
+      "loss": 1.7557,
+      "step": 3750
+    },
+    {
+      "epoch": 0.376,
+      "grad_norm": 10.125,
+      "learning_rate": 9.763717519728518e-06,
+      "loss": 1.6195,
+      "step": 3760
+    },
+    {
+      "epoch": 0.377,
+      "grad_norm": 10.8125,
+      "learning_rate": 9.76207498413563e-06,
+      "loss": 1.6373,
+      "step": 3770
+    },
+    {
+      "epoch": 0.378,
+      "grad_norm": 12.9375,
+      "learning_rate": 9.760426898325849e-06,
+      "loss": 1.7058,
+      "step": 3780
+    },
+    {
+      "epoch": 0.379,
+      "grad_norm": 21.5,
+      "learning_rate": 9.758773264220027e-06,
+      "loss": 1.8831,
+      "step": 3790
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 11.9375,
+      "learning_rate": 9.757114083745482e-06,
+      "loss": 1.9027,
+      "step": 3800
+    },
+    {
+      "epoch": 0.381,
+      "grad_norm": 21.125,
+      "learning_rate": 9.755449358835995e-06,
+      "loss": 1.8613,
+      "step": 3810
+    },
+    {
+      "epoch": 0.382,
+      "grad_norm": 12.8125,
+      "learning_rate": 9.75377909143181e-06,
+      "loss": 1.6181,
+      "step": 3820
+    },
+    {
+      "epoch": 0.383,
+      "grad_norm": 11.375,
+      "learning_rate": 9.752103283479626e-06,
+      "loss": 2.3528,
+      "step": 3830
+    },
+    {
+      "epoch": 0.384,
+      "grad_norm": 15.3125,
+      "learning_rate": 9.750421936932608e-06,
+      "loss": 1.807,
+      "step": 3840
+    },
+    {
+      "epoch": 0.385,
+      "grad_norm": 15.4375,
+      "learning_rate": 9.74873505375037e-06,
+      "loss": 1.4498,
+      "step": 3850
+    },
+    {
+      "epoch": 0.386,
+      "grad_norm": 18.625,
+      "learning_rate": 9.747042635898982e-06,
+      "loss": 1.9096,
+      "step": 3860
+    },
+    {
+      "epoch": 0.387,
+      "grad_norm": 13.125,
+      "learning_rate": 9.745344685350963e-06,
+      "loss": 2.0819,
+      "step": 3870
+    },
+    {
+      "epoch": 0.388,
+      "grad_norm": 15.3125,
+      "learning_rate": 9.743641204085282e-06,
+      "loss": 2.1709,
+      "step": 3880
+    },
+    {
+      "epoch": 0.389,
+      "grad_norm": 15.0,
+      "learning_rate": 9.74193219408735e-06,
+      "loss": 1.7934,
+      "step": 3890
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 18.0,
+      "learning_rate": 9.740217657349028e-06,
+      "loss": 2.1212,
+      "step": 3900
+    },
+    {
+      "epoch": 0.391,
+      "grad_norm": 15.9375,
+      "learning_rate": 9.738497595868615e-06,
+      "loss": 1.8096,
+      "step": 3910
+    },
+    {
+      "epoch": 0.392,
+      "grad_norm": 17.875,
+      "learning_rate": 9.736772011650846e-06,
+      "loss": 1.7102,
+      "step": 3920
+    },
+    {
+      "epoch": 0.393,
+      "grad_norm": 16.5,
+      "learning_rate": 9.735040906706898e-06,
+      "loss": 1.8275,
+      "step": 3930
+    },
+    {
+      "epoch": 0.394,
+      "grad_norm": 13.1875,
+      "learning_rate": 9.733304283054384e-06,
+      "loss": 1.5762,
+      "step": 3940
+    },
+    {
+      "epoch": 0.395,
+      "grad_norm": 10.875,
+      "learning_rate": 9.731562142717341e-06,
+      "loss": 1.6168,
+      "step": 3950
+    },
+    {
+      "epoch": 0.396,
+      "grad_norm": 16.125,
+      "learning_rate": 9.729814487726241e-06,
+      "loss": 1.8766,
+      "step": 3960
+    },
+    {
+      "epoch": 0.397,
+      "grad_norm": 13.9375,
+      "learning_rate": 9.728061320117985e-06,
+      "loss": 1.569,
+      "step": 3970
+    },
+    {
+      "epoch": 0.398,
+      "grad_norm": 9.875,
+      "learning_rate": 9.726302641935895e-06,
+      "loss": 1.4744,
+      "step": 3980
+    },
+    {
+      "epoch": 0.399,
+      "grad_norm": 14.125,
+      "learning_rate": 9.724538455229718e-06,
+      "loss": 1.7172,
+      "step": 3990
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 35.5,
+      "learning_rate": 9.722768762055619e-06,
+      "loss": 1.8122,
+      "step": 4000
+    },
+    {
+      "epoch": 0.401,
+      "grad_norm": 11.5,
+      "learning_rate": 9.720993564476183e-06,
+      "loss": 1.809,
+      "step": 4010
+    },
+    {
+      "epoch": 0.402,
+      "grad_norm": 16.125,
+      "learning_rate": 9.719212864560412e-06,
+      "loss": 1.5481,
+      "step": 4020
+    },
+    {
+      "epoch": 0.403,
+      "grad_norm": 12.875,
+      "learning_rate": 9.717426664383713e-06,
+      "loss": 1.5858,
+      "step": 4030
+    },
+    {
+      "epoch": 0.404,
+      "grad_norm": 16.875,
+      "learning_rate": 9.715634966027915e-06,
+      "loss": 1.6932,
+      "step": 4040
+    },
+    {
+      "epoch": 0.405,
+      "grad_norm": 13.4375,
+      "learning_rate": 9.713837771581248e-06,
+      "loss": 1.6665,
+      "step": 4050
+    },
+    {
+      "epoch": 0.406,
+      "grad_norm": 14.625,
+      "learning_rate": 9.712035083138347e-06,
+      "loss": 1.7522,
+      "step": 4060
+    },
+    {
+      "epoch": 0.407,
+      "grad_norm": 15.9375,
+      "learning_rate": 9.710226902800255e-06,
+      "loss": 1.7761,
+      "step": 4070
+    },
+    {
+      "epoch": 0.408,
+      "grad_norm": 10.4375,
+      "learning_rate": 9.708413232674412e-06,
+      "loss": 1.7572,
+      "step": 4080
+    },
+    {
+      "epoch": 0.409,
+      "grad_norm": 11.6875,
+      "learning_rate": 9.706594074874656e-06,
+      "loss": 2.1036,
+      "step": 4090
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 15.4375,
+      "learning_rate": 9.704769431521226e-06,
+      "loss": 1.7506,
+      "step": 4100
+    },
+    {
+      "epoch": 0.411,
+      "grad_norm": 15.6875,
+      "learning_rate": 9.702939304740747e-06,
+      "loss": 1.7222,
+      "step": 4110
+    },
+    {
+      "epoch": 0.412,
+      "grad_norm": 13.25,
+      "learning_rate": 9.701103696666241e-06,
+      "loss": 1.8004,
+      "step": 4120
+    },
+    {
+      "epoch": 0.413,
+      "grad_norm": 12.375,
+      "learning_rate": 9.699262609437117e-06,
+      "loss": 1.2991,
+      "step": 4130
+    },
+    {
+      "epoch": 0.414,
+      "grad_norm": 16.75,
+      "learning_rate": 9.697416045199167e-06,
+      "loss": 1.5562,
+      "step": 4140
+    },
+    {
+      "epoch": 0.415,
+      "grad_norm": 17.125,
+      "learning_rate": 9.695564006104572e-06,
+      "loss": 3.9309,
+      "step": 4150
+    },
+    {
+      "epoch": 0.416,
+      "grad_norm": 12.5,
+      "learning_rate": 9.693706494311887e-06,
+      "loss": 3.0747,
+      "step": 4160
+    },
+    {
+      "epoch": 0.417,
+      "grad_norm": 9.375,
+      "learning_rate": 9.691843511986051e-06,
+      "loss": 1.9317,
+      "step": 4170
+    },
+    {
+      "epoch": 0.418,
+      "grad_norm": 8.5,
+      "learning_rate": 9.68997506129838e-06,
+      "loss": 1.3797,
+      "step": 4180
+    },
+    {
+      "epoch": 0.419,
+      "grad_norm": 10.6875,
+      "learning_rate": 9.688101144426557e-06,
+      "loss": 1.7277,
+      "step": 4190
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 12.9375,
+      "learning_rate": 9.68622176355464e-06,
+      "loss": 1.8881,
+      "step": 4200
+    },
+    {
+      "epoch": 0.421,
+      "grad_norm": 28.75,
+      "learning_rate": 9.684336920873055e-06,
+      "loss": 2.1434,
+      "step": 4210
+    },
+    {
+      "epoch": 0.422,
+      "grad_norm": 13.8125,
+      "learning_rate": 9.682446618578593e-06,
+      "loss": 1.707,
+      "step": 4220
+    },
+    {
+      "epoch": 0.423,
+      "grad_norm": 12.375,
+      "learning_rate": 9.680550858874409e-06,
+      "loss": 1.7337,
+      "step": 4230
+    },
+    {
+      "epoch": 0.424,
+      "grad_norm": 11.5625,
+      "learning_rate": 9.678649643970018e-06,
+      "loss": 2.0312,
+      "step": 4240
+    },
+    {
+      "epoch": 0.425,
+      "grad_norm": 19.25,
+      "learning_rate": 9.676742976081295e-06,
+      "loss": 1.9142,
+      "step": 4250
+    },
+    {
+      "epoch": 0.426,
+      "grad_norm": 15.5625,
+      "learning_rate": 9.674830857430466e-06,
+      "loss": 1.5333,
+      "step": 4260
+    },
+    {
+      "epoch": 0.427,
+      "grad_norm": 14.5,
+      "learning_rate": 9.672913290246115e-06,
+      "loss": 1.904,
+      "step": 4270
+    },
+    {
+      "epoch": 0.428,
+      "grad_norm": 22.5,
+      "learning_rate": 9.670990276763174e-06,
+      "loss": 1.8497,
+      "step": 4280
+    },
+    {
+      "epoch": 0.429,
+      "grad_norm": 18.0,
+      "learning_rate": 9.669061819222923e-06,
+      "loss": 1.8744,
+      "step": 4290
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 9.8125,
+      "learning_rate": 9.667127919872983e-06,
+      "loss": 1.5508,
+      "step": 4300
+    },
+    {
+      "epoch": 0.431,
+      "grad_norm": 9.9375,
+      "learning_rate": 9.665188580967326e-06,
+      "loss": 1.8347,
+      "step": 4310
+    },
+    {
+      "epoch": 0.432,
+      "grad_norm": 11.75,
+      "learning_rate": 9.663243804766256e-06,
+      "loss": 1.6582,
+      "step": 4320
+    },
+    {
+      "epoch": 0.433,
+      "grad_norm": 16.875,
+      "learning_rate": 9.661293593536418e-06,
+      "loss": 1.8067,
+      "step": 4330
+    },
+    {
+      "epoch": 0.434,
+      "grad_norm": 12.1875,
+      "learning_rate": 9.65933794955079e-06,
+      "loss": 1.5748,
+      "step": 4340
+    },
+    {
+      "epoch": 0.435,
+      "grad_norm": 11.5,
+      "learning_rate": 9.657376875088687e-06,
+      "loss": 1.493,
+      "step": 4350
+    },
+    {
+      "epoch": 0.436,
+      "grad_norm": 13.875,
+      "learning_rate": 9.655410372435743e-06,
+      "loss": 1.5745,
+      "step": 4360
+    },
+    {
+      "epoch": 0.437,
+      "grad_norm": 11.5625,
+      "learning_rate": 9.653438443883928e-06,
+      "loss": 1.728,
+      "step": 4370
+    },
+    {
+      "epoch": 0.438,
+      "grad_norm": 13.0,
+      "learning_rate": 9.651461091731531e-06,
+      "loss": 1.6899,
+      "step": 4380
+    },
+    {
+      "epoch": 0.439,
+      "grad_norm": 12.1875,
+      "learning_rate": 9.649478318283163e-06,
+      "loss": 1.8097,
+      "step": 4390
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 13.6875,
+      "learning_rate": 9.647490125849756e-06,
+      "loss": 1.7653,
+      "step": 4400
+    },
+    {
+      "epoch": 0.441,
+      "grad_norm": 14.5625,
+      "learning_rate": 9.645496516748555e-06,
+      "loss": 1.6708,
+      "step": 4410
+    },
+    {
+      "epoch": 0.442,
+      "grad_norm": 16.75,
+      "learning_rate": 9.64349749330312e-06,
+      "loss": 2.0399,
+      "step": 4420
+    },
+    {
+      "epoch": 0.443,
+      "grad_norm": 17.125,
+      "learning_rate": 9.64149305784332e-06,
+      "loss": 1.8246,
+      "step": 4430
+    },
+    {
+      "epoch": 0.444,
+      "grad_norm": 13.3125,
+      "learning_rate": 9.639483212705331e-06,
+      "loss": 1.6502,
+      "step": 4440
+    },
+    {
+      "epoch": 0.445,
+      "grad_norm": 12.3125,
+      "learning_rate": 9.637467960231637e-06,
+      "loss": 1.784,
+      "step": 4450
+    },
+    {
+      "epoch": 0.446,
+      "grad_norm": 12.375,
+      "learning_rate": 9.635447302771022e-06,
+      "loss": 1.7032,
+      "step": 4460
+    },
+    {
+      "epoch": 0.447,
+      "grad_norm": 25.875,
+      "learning_rate": 9.63342124267857e-06,
+      "loss": 1.7408,
+      "step": 4470
+    },
+    {
+      "epoch": 0.448,
+      "grad_norm": 12.25,
+      "learning_rate": 9.631389782315663e-06,
+      "loss": 1.7033,
+      "step": 4480
+    },
+    {
+      "epoch": 0.449,
+      "grad_norm": 19.25,
+      "learning_rate": 9.629352924049975e-06,
+      "loss": 1.7234,
+      "step": 4490
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 14.875,
+      "learning_rate": 9.627310670255471e-06,
+      "loss": 1.749,
+      "step": 4500
+    },
+    {
+      "epoch": 0.451,
+      "grad_norm": 13.1875,
+      "learning_rate": 9.625263023312408e-06,
+      "loss": 1.8551,
+      "step": 4510
+    },
+    {
+      "epoch": 0.452,
+      "grad_norm": 11.375,
+      "learning_rate": 9.623209985607327e-06,
+      "loss": 1.7065,
+      "step": 4520
+    },
+    {
+      "epoch": 0.453,
+      "grad_norm": 13.25,
+      "learning_rate": 9.621151559533048e-06,
+      "loss": 1.7648,
+      "step": 4530
+    },
+    {
+      "epoch": 0.454,
+      "grad_norm": 7.375,
+      "learning_rate": 9.619087747488675e-06,
+      "loss": 1.8575,
+      "step": 4540
+    },
+    {
+      "epoch": 0.455,
+      "grad_norm": 14.0,
+      "learning_rate": 9.617018551879593e-06,
+      "loss": 1.7919,
+      "step": 4550
+    },
+    {
+      "epoch": 0.456,
+      "grad_norm": 14.625,
+      "learning_rate": 9.614943975117454e-06,
+      "loss": 2.0214,
+      "step": 4560
+    },
+    {
+      "epoch": 0.457,
+      "grad_norm": 16.0,
+      "learning_rate": 9.612864019620187e-06,
+      "loss": 1.7546,
+      "step": 4570
+    },
+    {
+      "epoch": 0.458,
+      "grad_norm": 15.5,
+      "learning_rate": 9.610778687811988e-06,
+      "loss": 1.6869,
+      "step": 4580
+    },
+    {
+      "epoch": 0.459,
+      "grad_norm": 11.125,
+      "learning_rate": 9.608687982123318e-06,
+      "loss": 1.5839,
+      "step": 4590
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 19.75,
+      "learning_rate": 9.606591904990903e-06,
+      "loss": 1.887,
+      "step": 4600
+    },
+    {
+      "epoch": 0.461,
+      "grad_norm": 14.875,
+      "learning_rate": 9.60449045885773e-06,
+      "loss": 1.8664,
+      "step": 4610
+    },
+    {
+      "epoch": 0.462,
+      "grad_norm": 13.5,
+      "learning_rate": 9.602383646173041e-06,
+      "loss": 1.9194,
+      "step": 4620
+    },
+    {
+      "epoch": 0.463,
+      "grad_norm": 9.625,
+      "learning_rate": 9.600271469392335e-06,
+      "loss": 1.3541,
+      "step": 4630
+    },
+    {
+      "epoch": 0.464,
+      "grad_norm": 11.25,
+      "learning_rate": 9.598153930977364e-06,
+      "loss": 1.5938,
+      "step": 4640
+    },
+    {
+      "epoch": 0.465,
+      "grad_norm": 19.75,
+      "learning_rate": 9.596031033396126e-06,
+      "loss": 1.8294,
+      "step": 4650
+    },
+    {
+      "epoch": 0.466,
+      "grad_norm": 12.0625,
+      "learning_rate": 9.593902779122865e-06,
+      "loss": 1.5714,
+      "step": 4660
+    },
+    {
+      "epoch": 0.467,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.591769170638072e-06,
+      "loss": 1.7185,
+      "step": 4670
+    },
+    {
+      "epoch": 0.468,
+      "grad_norm": 15.1875,
+      "learning_rate": 9.589630210428476e-06,
+      "loss": 1.5794,
+      "step": 4680
+    },
+    {
+      "epoch": 0.469,
+      "grad_norm": 14.625,
+      "learning_rate": 9.587485900987042e-06,
+      "loss": 2.0519,
+      "step": 4690
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 11.75,
+      "learning_rate": 9.585336244812969e-06,
+      "loss": 1.6668,
+      "step": 4700
+    },
+    {
+      "epoch": 0.471,
+      "grad_norm": 19.75,
+      "learning_rate": 9.583181244411694e-06,
+      "loss": 1.8728,
+      "step": 4710
+    },
+    {
+      "epoch": 0.472,
+      "grad_norm": 24.875,
+      "learning_rate": 9.581020902294876e-06,
+      "loss": 1.9427,
+      "step": 4720
+    },
+    {
+      "epoch": 0.473,
+      "grad_norm": 16.75,
+      "learning_rate": 9.5788552209804e-06,
+      "loss": 1.8761,
+      "step": 4730
+    },
+    {
+      "epoch": 0.474,
+      "grad_norm": 16.25,
+      "learning_rate": 9.576684202992381e-06,
+      "loss": 1.5595,
+      "step": 4740
+    },
+    {
+      "epoch": 0.475,
+      "grad_norm": 10.9375,
+      "learning_rate": 9.574507850861145e-06,
+      "loss": 1.5226,
+      "step": 4750
+    },
+    {
+      "epoch": 0.476,
+      "grad_norm": 9.75,
+      "learning_rate": 9.572326167123242e-06,
+      "loss": 1.3948,
+      "step": 4760
+    },
+    {
+      "epoch": 0.477,
+      "grad_norm": 10.4375,
+      "learning_rate": 9.570139154321429e-06,
+      "loss": 2.0113,
+      "step": 4770
+    },
+    {
+      "epoch": 0.478,
+      "grad_norm": 15.25,
+      "learning_rate": 9.567946815004681e-06,
+      "loss": 1.7644,
+      "step": 4780
+    },
+    {
+      "epoch": 0.479,
+      "grad_norm": 12.0,
+      "learning_rate": 9.565749151728176e-06,
+      "loss": 1.511,
+      "step": 4790
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 17.5,
+      "learning_rate": 9.563546167053302e-06,
+      "loss": 1.5692,
+      "step": 4800
+    },
+    {
+      "epoch": 0.481,
+      "grad_norm": 13.5,
+      "learning_rate": 9.561337863547642e-06,
+      "loss": 1.7868,
+      "step": 4810
+    },
+    {
+      "epoch": 0.482,
+      "grad_norm": 12.625,
+      "learning_rate": 9.559124243784985e-06,
+      "loss": 1.5642,
+      "step": 4820
+    },
+    {
+      "epoch": 0.483,
+      "grad_norm": 19.5,
+      "learning_rate": 9.556905310345313e-06,
+      "loss": 1.6248,
+      "step": 4830
+    },
+    {
+      "epoch": 0.484,
+      "grad_norm": 18.125,
+      "learning_rate": 9.554681065814802e-06,
+      "loss": 1.8501,
+      "step": 4840
+    },
+    {
+      "epoch": 0.485,
+      "grad_norm": 11.8125,
+      "learning_rate": 9.552451512785816e-06,
+      "loss": 2.0053,
+      "step": 4850
+    },
+    {
+      "epoch": 0.486,
+      "grad_norm": 11.375,
+      "learning_rate": 9.55021665385691e-06,
+      "loss": 1.7916,
+      "step": 4860
+    },
+    {
+      "epoch": 0.487,
+      "grad_norm": 11.875,
+      "learning_rate": 9.54797649163282e-06,
+      "loss": 1.8036,
+      "step": 4870
+    },
+    {
+      "epoch": 0.488,
+      "grad_norm": 18.25,
+      "learning_rate": 9.545731028724463e-06,
+      "loss": 1.905,
+      "step": 4880
+    },
+    {
+      "epoch": 0.489,
+      "grad_norm": 15.75,
+      "learning_rate": 9.543480267748935e-06,
+      "loss": 2.1848,
+      "step": 4890
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 7.78125,
+      "learning_rate": 9.541224211329504e-06,
+      "loss": 1.9417,
+      "step": 4900
+    },
+    {
+      "epoch": 0.491,
+      "grad_norm": 17.375,
+      "learning_rate": 9.53896286209562e-06,
+      "loss": 1.5473,
+      "step": 4910
+    },
+    {
+      "epoch": 0.492,
+      "grad_norm": 13.125,
+      "learning_rate": 9.536696222682887e-06,
+      "loss": 2.0592,
+      "step": 4920
+    },
+    {
+      "epoch": 0.493,
+      "grad_norm": 15.0625,
+      "learning_rate": 9.534424295733085e-06,
+      "loss": 1.7816,
+      "step": 4930
+    },
+    {
+      "epoch": 0.494,
+      "grad_norm": 15.0,
+      "learning_rate": 9.532147083894153e-06,
+      "loss": 1.6778,
+      "step": 4940
+    },
+    {
+      "epoch": 0.495,
+      "grad_norm": 13.0,
+      "learning_rate": 9.529864589820192e-06,
+      "loss": 1.6575,
+      "step": 4950
+    },
+    {
+      "epoch": 0.496,
+      "grad_norm": 9.8125,
+      "learning_rate": 9.527576816171454e-06,
+      "loss": 1.7262,
+      "step": 4960
+    },
+    {
+      "epoch": 0.497,
+      "grad_norm": 14.125,
+      "learning_rate": 9.525283765614353e-06,
+      "loss": 1.7195,
+      "step": 4970
+    },
+    {
+      "epoch": 0.498,
+      "grad_norm": 12.9375,
+      "learning_rate": 9.522985440821444e-06,
+      "loss": 1.7209,
+      "step": 4980
+    },
+    {
+      "epoch": 0.499,
+      "grad_norm": 10.5625,
+      "learning_rate": 9.520681844471435e-06,
+      "loss": 1.8001,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 17.375,
+      "learning_rate": 9.518372979249178e-06,
+      "loss": 1.6056,
+      "step": 5000
+    },
+    {
+      "epoch": 0.501,
+      "grad_norm": 12.5625,
+      "learning_rate": 9.516058847845662e-06,
+      "loss": 1.5862,
+      "step": 5010
+    },
+    {
+      "epoch": 0.502,
+      "grad_norm": 26.25,
+      "learning_rate": 9.513739452958018e-06,
+      "loss": 1.8368,
+      "step": 5020
+    },
+    {
+      "epoch": 0.503,
+      "grad_norm": 10.9375,
+      "learning_rate": 9.511414797289506e-06,
+      "loss": 1.9322,
+      "step": 5030
+    },
+    {
+      "epoch": 0.504,
+      "grad_norm": 19.0,
+      "learning_rate": 9.509084883549528e-06,
+      "loss": 1.8419,
+      "step": 5040
+    },
+    {
+      "epoch": 0.505,
+      "grad_norm": 17.75,
+      "learning_rate": 9.506749714453601e-06,
+      "loss": 1.6057,
+      "step": 5050
+    },
+    {
+      "epoch": 0.506,
+      "grad_norm": 10.5,
+      "learning_rate": 9.504409292723378e-06,
+      "loss": 1.6505,
+      "step": 5060
+    },
+    {
+      "epoch": 0.507,
+      "grad_norm": 14.75,
+      "learning_rate": 9.502063621086628e-06,
+      "loss": 1.6784,
+      "step": 5070
+    },
+    {
+      "epoch": 0.508,
+      "grad_norm": 10.4375,
+      "learning_rate": 9.49971270227724e-06,
+      "loss": 1.4831,
+      "step": 5080
+    },
+    {
+      "epoch": 0.509,
+      "grad_norm": 12.6875,
+      "learning_rate": 9.497356539035218e-06,
+      "loss": 1.8091,
+      "step": 5090
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 19.0,
+      "learning_rate": 9.494995134106682e-06,
+      "loss": 1.8691,
+      "step": 5100
+    },
+    {
+      "epoch": 0.511,
+      "grad_norm": 15.125,
+      "learning_rate": 9.492628490243857e-06,
+      "loss": 1.6902,
+      "step": 5110
+    },
+    {
+      "epoch": 0.512,
+      "grad_norm": 12.8125,
+      "learning_rate": 9.490256610205076e-06,
+      "loss": 1.6769,
+      "step": 5120
+    },
+    {
+      "epoch": 0.513,
+      "grad_norm": 10.625,
+      "learning_rate": 9.487879496754775e-06,
+      "loss": 1.5432,
+      "step": 5130
+    },
+    {
+      "epoch": 0.514,
+      "grad_norm": 14.75,
+      "learning_rate": 9.485497152663487e-06,
+      "loss": 1.7945,
+      "step": 5140
+    },
+    {
+      "epoch": 0.515,
+      "grad_norm": 16.375,
+      "learning_rate": 9.483109580707845e-06,
+      "loss": 1.6994,
+      "step": 5150
+    },
+    {
+      "epoch": 0.516,
+      "grad_norm": 24.25,
+      "learning_rate": 9.480716783670574e-06,
+      "loss": 1.9848,
+      "step": 5160
+    },
+    {
+      "epoch": 0.517,
+      "grad_norm": 10.3125,
+      "learning_rate": 9.478318764340486e-06,
+      "loss": 1.5767,
+      "step": 5170
+    },
+    {
+      "epoch": 0.518,
+      "grad_norm": 14.25,
+      "learning_rate": 9.475915525512484e-06,
+      "loss": 1.6994,
+      "step": 5180
+    },
+    {
+      "epoch": 0.519,
+      "grad_norm": 12.5625,
+      "learning_rate": 9.473507069987552e-06,
+      "loss": 1.7581,
+      "step": 5190
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 15.0625,
+      "learning_rate": 9.471093400572752e-06,
+      "loss": 1.8256,
+      "step": 5200
+    },
+    {
+      "epoch": 0.521,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.468674520081227e-06,
+      "loss": 2.0102,
+      "step": 5210
+    },
+    {
+      "epoch": 0.522,
+      "grad_norm": 18.75,
+      "learning_rate": 9.466250431332192e-06,
+      "loss": 1.8705,
+      "step": 5220
+    },
+    {
+      "epoch": 0.523,
+      "grad_norm": 14.75,
+      "learning_rate": 9.463821137150931e-06,
+      "loss": 1.7178,
+      "step": 5230
+    },
+    {
+      "epoch": 0.524,
+      "grad_norm": 15.375,
+      "learning_rate": 9.461386640368794e-06,
+      "loss": 1.6406,
+      "step": 5240
+    },
+    {
+      "epoch": 0.525,
+      "grad_norm": 15.0625,
+      "learning_rate": 9.458946943823202e-06,
+      "loss": 1.68,
+      "step": 5250
+    },
+    {
+      "epoch": 0.526,
+      "grad_norm": 13.5,
+      "learning_rate": 9.456502050357624e-06,
+      "loss": 1.8566,
+      "step": 5260
+    },
+    {
+      "epoch": 0.527,
+      "grad_norm": 11.5,
+      "learning_rate": 9.454051962821598e-06,
+      "loss": 1.754,
+      "step": 5270
+    },
+    {
+      "epoch": 0.528,
+      "grad_norm": 11.9375,
+      "learning_rate": 9.451596684070707e-06,
+      "loss": 1.7704,
+      "step": 5280
+    },
+    {
+      "epoch": 0.529,
+      "grad_norm": Infinity,
+      "learning_rate": 9.44913621696659e-06,
+      "loss": 1.8255,
+      "step": 5290
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 9.5625,
+      "learning_rate": 9.44667056437693e-06,
+      "loss": 1.7151,
+      "step": 5300
+    },
+    {
+      "epoch": 0.531,
+      "grad_norm": 15.25,
+      "learning_rate": 9.444199729175456e-06,
+      "loss": 1.7014,
+      "step": 5310
+    },
+    {
+      "epoch": 0.532,
+      "grad_norm": 9.375,
+      "learning_rate": 9.441723714241935e-06,
+      "loss": 1.6094,
+      "step": 5320
+    },
+    {
+      "epoch": 0.533,
+      "grad_norm": 15.875,
+      "learning_rate": 9.439242522462171e-06,
+      "loss": 1.4946,
+      "step": 5330
+    },
+    {
+      "epoch": 0.534,
+      "grad_norm": 14.4375,
+      "learning_rate": 9.436756156728006e-06,
+      "loss": 1.4674,
+      "step": 5340
+    },
+    {
+      "epoch": 0.535,
+      "grad_norm": 9.6875,
+      "learning_rate": 9.434264619937304e-06,
+      "loss": 1.5922,
+      "step": 5350
+    },
+    {
+      "epoch": 0.536,
+      "grad_norm": 9.75,
+      "learning_rate": 9.431767914993966e-06,
+      "loss": 1.5858,
+      "step": 5360
+    },
+    {
+      "epoch": 0.537,
+      "grad_norm": 19.875,
+      "learning_rate": 9.429266044807907e-06,
+      "loss": 1.8128,
+      "step": 5370
+    },
+    {
+      "epoch": 0.538,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.426759012295068e-06,
+      "loss": 2.2777,
+      "step": 5380
+    },
+    {
+      "epoch": 0.539,
+      "grad_norm": 8.0625,
+      "learning_rate": 9.424246820377408e-06,
+      "loss": 1.5322,
+      "step": 5390
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 9.25,
+      "learning_rate": 9.421729471982892e-06,
+      "loss": 1.7463,
+      "step": 5400
+    },
+    {
+      "epoch": 0.541,
+      "grad_norm": 27.5,
+      "learning_rate": 9.4192069700455e-06,
+      "loss": 1.7338,
+      "step": 5410
+    },
+    {
+      "epoch": 0.542,
+      "grad_norm": 25.75,
+      "learning_rate": 9.416679317505222e-06,
+      "loss": 2.5891,
+      "step": 5420
+    },
+    {
+      "epoch": 0.543,
+      "grad_norm": 13.3125,
+      "learning_rate": 9.41414651730804e-06,
+      "loss": 1.8474,
+      "step": 5430
+    },
+    {
+      "epoch": 0.544,
+      "grad_norm": 10.6875,
+      "learning_rate": 9.41160857240595e-06,
+      "loss": 1.462,
+      "step": 5440
+    },
+    {
+      "epoch": 0.545,
+      "grad_norm": 15.375,
+      "learning_rate": 9.40906548575693e-06,
+      "loss": 1.7087,
+      "step": 5450
+    },
+    {
+      "epoch": 0.546,
+      "grad_norm": 17.625,
+      "learning_rate": 9.406517260324962e-06,
+      "loss": 1.8401,
+      "step": 5460
+    },
+    {
+      "epoch": 0.547,
+      "grad_norm": 9.5625,
+      "learning_rate": 9.40396389908001e-06,
+      "loss": 1.5494,
+      "step": 5470
+    },
+    {
+      "epoch": 0.548,
+      "grad_norm": 17.625,
+      "learning_rate": 9.401405404998028e-06,
+      "loss": 1.737,
+      "step": 5480
+    },
+    {
+      "epoch": 0.549,
+      "grad_norm": 12.1875,
+      "learning_rate": 9.39884178106095e-06,
+      "loss": 1.5974,
+      "step": 5490
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 15.5625,
+      "learning_rate": 9.39627303025669e-06,
+      "loss": 1.7712,
+      "step": 5500
+    },
+    {
+      "epoch": 0.551,
+      "grad_norm": 12.5,
+      "learning_rate": 9.393699155579138e-06,
+      "loss": 1.7298,
+      "step": 5510
+    },
+    {
+      "epoch": 0.552,
+      "grad_norm": 11.3125,
+      "learning_rate": 9.391120160028151e-06,
+      "loss": 1.6243,
+      "step": 5520
+    },
+    {
+      "epoch": 0.553,
+      "grad_norm": 14.1875,
+      "learning_rate": 9.388536046609564e-06,
+      "loss": 1.81,
+      "step": 5530
+    },
+    {
+      "epoch": 0.554,
+      "grad_norm": 18.125,
+      "learning_rate": 9.385946818335165e-06,
+      "loss": 2.0227,
+      "step": 5540
+    },
+    {
+      "epoch": 0.555,
+      "grad_norm": 15.6875,
+      "learning_rate": 9.383352478222716e-06,
+      "loss": 1.7729,
+      "step": 5550
+    },
+    {
+      "epoch": 0.556,
+      "grad_norm": 12.875,
+      "learning_rate": 9.380753029295925e-06,
+      "loss": 1.6293,
+      "step": 5560
+    },
+    {
+      "epoch": 0.557,
+      "grad_norm": 16.25,
+      "learning_rate": 9.378148474584465e-06,
+      "loss": 1.8049,
+      "step": 5570
+    },
+    {
+      "epoch": 0.558,
+      "grad_norm": 17.625,
+      "learning_rate": 9.37553881712395e-06,
+      "loss": 1.8867,
+      "step": 5580
+    },
+    {
+      "epoch": 0.559,
+      "grad_norm": 9.5,
+      "learning_rate": 9.372924059955948e-06,
+      "loss": 1.4487,
+      "step": 5590
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 9.375,
+      "learning_rate": 9.37030420612797e-06,
+      "loss": 1.542,
+      "step": 5600
+    },
+    {
+      "epoch": 0.561,
+      "grad_norm": 14.1875,
+      "learning_rate": 9.367679258693463e-06,
+      "loss": 1.8637,
+      "step": 5610
+    },
+    {
+      "epoch": 0.562,
+      "grad_norm": 14.625,
+      "learning_rate": 9.365049220711816e-06,
+      "loss": 1.8099,
+      "step": 5620
+    },
+    {
+      "epoch": 0.563,
+      "grad_norm": 15.9375,
+      "learning_rate": 9.362414095248344e-06,
+      "loss": 1.5403,
+      "step": 5630
+    },
+    {
+      "epoch": 0.564,
+      "grad_norm": 9.375,
+      "learning_rate": 9.359773885374301e-06,
+      "loss": 1.5533,
+      "step": 5640
+    },
+    {
+      "epoch": 0.565,
+      "grad_norm": 11.3125,
+      "learning_rate": 9.35712859416686e-06,
+      "loss": 1.6681,
+      "step": 5650
+    },
+    {
+      "epoch": 0.566,
+      "grad_norm": 21.0,
+      "learning_rate": 9.35447822470912e-06,
+      "loss": 1.468,
+      "step": 5660
+    },
+    {
+      "epoch": 0.567,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.351822780090091e-06,
+      "loss": 1.5312,
+      "step": 5670
+    },
+    {
+      "epoch": 0.568,
+      "grad_norm": 13.0,
+      "learning_rate": 9.349162263404712e-06,
+      "loss": 1.3524,
+      "step": 5680
+    },
+    {
+      "epoch": 0.569,
+      "grad_norm": 12.8125,
+      "learning_rate": 9.34649667775382e-06,
+      "loss": 1.686,
+      "step": 5690
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 13.875,
+      "learning_rate": 9.343826026244167e-06,
+      "loss": 1.7423,
+      "step": 5700
+    },
+    {
+      "epoch": 0.571,
+      "grad_norm": 10.25,
+      "learning_rate": 9.341150311988407e-06,
+      "loss": 1.6786,
+      "step": 5710
+    },
+    {
+      "epoch": 0.572,
+      "grad_norm": 11.0625,
+      "learning_rate": 9.338469538105099e-06,
+      "loss": 1.4444,
+      "step": 5720
+    },
+    {
+      "epoch": 0.573,
+      "grad_norm": 10.3125,
+      "learning_rate": 9.33578370771869e-06,
+      "loss": 2.936,
+      "step": 5730
+    },
+    {
+      "epoch": 0.574,
+      "grad_norm": 11.875,
+      "learning_rate": 9.33309282395953e-06,
+      "loss": 1.4382,
+      "step": 5740
+    },
+    {
+      "epoch": 0.575,
+      "grad_norm": 11.125,
+      "learning_rate": 9.33039688996385e-06,
+      "loss": 2.0261,
+      "step": 5750
+    },
+    {
+      "epoch": 0.576,
+      "grad_norm": 12.9375,
+      "learning_rate": 9.327695908873774e-06,
+      "loss": 1.631,
+      "step": 5760
+    },
+    {
+      "epoch": 0.577,
+      "grad_norm": 12.125,
+      "learning_rate": 9.32498988383731e-06,
+      "loss": 2.014,
+      "step": 5770
+    },
+    {
+      "epoch": 0.578,
+      "grad_norm": 12.5,
+      "learning_rate": 9.322278818008335e-06,
+      "loss": 1.4757,
+      "step": 5780
+    },
+    {
+      "epoch": 0.579,
+      "grad_norm": 14.5,
+      "learning_rate": 9.319562714546609e-06,
+      "loss": 1.4455,
+      "step": 5790
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 21.5,
+      "learning_rate": 9.31684157661776e-06,
+      "loss": 1.5836,
+      "step": 5800
+    },
+    {
+      "epoch": 0.581,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.314115407393285e-06,
+      "loss": 1.6605,
+      "step": 5810
+    },
+    {
+      "epoch": 0.582,
+      "grad_norm": 13.9375,
+      "learning_rate": 9.311384210050545e-06,
+      "loss": 1.6715,
+      "step": 5820
+    },
+    {
+      "epoch": 0.583,
+      "grad_norm": 10.0,
+      "learning_rate": 9.308647987772765e-06,
+      "loss": 1.7904,
+      "step": 5830
+    },
+    {
+      "epoch": 0.584,
+      "grad_norm": 9.25,
+      "learning_rate": 9.305906743749016e-06,
+      "loss": 1.7058,
+      "step": 5840
+    },
+    {
+      "epoch": 0.585,
+      "grad_norm": 11.125,
+      "learning_rate": 9.303160481174233e-06,
+      "loss": 1.7445,
+      "step": 5850
+    },
+    {
+      "epoch": 0.586,
+      "grad_norm": 11.125,
+      "learning_rate": 9.300409203249194e-06,
+      "loss": 1.6601,
+      "step": 5860
+    },
+    {
+      "epoch": 0.587,
+      "grad_norm": 14.3125,
+      "learning_rate": 9.297652913180526e-06,
+      "loss": 1.9526,
+      "step": 5870
+    },
+    {
+      "epoch": 0.588,
+      "grad_norm": 10.5,
+      "learning_rate": 9.294891614180692e-06,
+      "loss": 1.8144,
+      "step": 5880
+    },
+    {
+      "epoch": 0.589,
+      "grad_norm": 21.125,
+      "learning_rate": 9.292125309468e-06,
+      "loss": 1.634,
+      "step": 5890
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 15.0,
+      "learning_rate": 9.28935400226659e-06,
+      "loss": 1.4131,
+      "step": 5900
+    },
+    {
+      "epoch": 0.591,
+      "grad_norm": 13.6875,
+      "learning_rate": 9.286577695806426e-06,
+      "loss": 1.5622,
+      "step": 5910
+    },
+    {
+      "epoch": 0.592,
+      "grad_norm": 20.125,
+      "learning_rate": 9.28379639332331e-06,
+      "loss": 1.833,
+      "step": 5920
+    },
+    {
+      "epoch": 0.593,
+      "grad_norm": 10.8125,
+      "learning_rate": 9.281010098058858e-06,
+      "loss": 1.8979,
+      "step": 5930
+    },
+    {
+      "epoch": 0.594,
+      "grad_norm": 20.375,
+      "learning_rate": 9.278218813260507e-06,
+      "loss": 1.693,
+      "step": 5940
+    },
+    {
+      "epoch": 0.595,
+      "grad_norm": 35.0,
+      "learning_rate": 9.275422542181512e-06,
+      "loss": 1.8859,
+      "step": 5950
+    },
+    {
+      "epoch": 0.596,
+      "grad_norm": 8.9375,
+      "learning_rate": 9.272621288080936e-06,
+      "loss": 1.6075,
+      "step": 5960
+    },
+    {
+      "epoch": 0.597,
+      "grad_norm": 10.125,
+      "learning_rate": 9.269815054223653e-06,
+      "loss": 1.74,
+      "step": 5970
+    },
+    {
+      "epoch": 0.598,
+      "grad_norm": 25.875,
+      "learning_rate": 9.267003843880341e-06,
+      "loss": 1.8712,
+      "step": 5980
+    },
+    {
+      "epoch": 0.599,
+      "grad_norm": 21.375,
+      "learning_rate": 9.264187660327471e-06,
+      "loss": 1.4676,
+      "step": 5990
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 12.375,
+      "learning_rate": 9.261366506847322e-06,
+      "loss": 1.524,
+      "step": 6000
+    },
+    {
+      "epoch": 0.601,
+      "grad_norm": 9.125,
+      "learning_rate": 9.258540386727956e-06,
+      "loss": 1.6472,
+      "step": 6010
+    },
+    {
+      "epoch": 0.602,
+      "grad_norm": 10.375,
+      "learning_rate": 9.25570930326323e-06,
+      "loss": 1.4288,
+      "step": 6020
+    },
+    {
+      "epoch": 0.603,
+      "grad_norm": 12.9375,
+      "learning_rate": 9.252873259752779e-06,
+      "loss": 2.0402,
+      "step": 6030
+    },
+    {
+      "epoch": 0.604,
+      "grad_norm": 16.125,
+      "learning_rate": 9.250032259502023e-06,
+      "loss": 1.5911,
+      "step": 6040
+    },
+    {
+      "epoch": 0.605,
+      "grad_norm": 14.4375,
+      "learning_rate": 9.247186305822162e-06,
+      "loss": 2.2132,
+      "step": 6050
+    },
+    {
+      "epoch": 0.606,
+      "grad_norm": 15.25,
+      "learning_rate": 9.244335402030165e-06,
+      "loss": 1.9692,
+      "step": 6060
+    },
+    {
+      "epoch": 0.607,
+      "grad_norm": 10.1875,
+      "learning_rate": 9.24147955144877e-06,
+      "loss": 1.5724,
+      "step": 6070
+    },
+    {
+      "epoch": 0.608,
+      "grad_norm": 21.5,
+      "learning_rate": 9.238618757406483e-06,
+      "loss": 1.989,
+      "step": 6080
+    },
+    {
+      "epoch": 0.609,
+      "grad_norm": 14.3125,
+      "learning_rate": 9.23575302323757e-06,
+      "loss": 2.0297,
+      "step": 6090
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 12.125,
+      "learning_rate": 9.232882352282055e-06,
+      "loss": 1.6989,
+      "step": 6100
+    },
+    {
+      "epoch": 0.611,
+      "grad_norm": 17.375,
+      "learning_rate": 9.230006747885716e-06,
+      "loss": 1.7045,
+      "step": 6110
+    },
+    {
+      "epoch": 0.612,
+      "grad_norm": 13.0,
+      "learning_rate": 9.227126213400083e-06,
+      "loss": 1.6774,
+      "step": 6120
+    },
+    {
+      "epoch": 0.613,
+      "grad_norm": 16.75,
+      "learning_rate": 9.224240752182428e-06,
+      "loss": 1.7543,
+      "step": 6130
+    },
+    {
+      "epoch": 0.614,
+      "grad_norm": 16.375,
+      "learning_rate": 9.22135036759577e-06,
+      "loss": 1.9639,
+      "step": 6140
+    },
+    {
+      "epoch": 0.615,
+      "grad_norm": 17.125,
+      "learning_rate": 9.21845506300886e-06,
+      "loss": 1.7489,
+      "step": 6150
+    },
+    {
+      "epoch": 0.616,
+      "grad_norm": 11.4375,
+      "learning_rate": 9.21555484179619e-06,
+      "loss": 1.6082,
+      "step": 6160
+    },
+    {
+      "epoch": 0.617,
+      "grad_norm": 13.3125,
+      "learning_rate": 9.212649707337974e-06,
+      "loss": 2.1134,
+      "step": 6170
+    },
+    {
+      "epoch": 0.618,
+      "grad_norm": 9.8125,
+      "learning_rate": 9.209739663020165e-06,
+      "loss": 1.5271,
+      "step": 6180
+    },
+    {
+      "epoch": 0.619,
+      "grad_norm": 13.375,
+      "learning_rate": 9.206824712234425e-06,
+      "loss": 1.6395,
+      "step": 6190
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 19.875,
+      "learning_rate": 9.203904858378143e-06,
+      "loss": 1.8458,
+      "step": 6200
+    },
+    {
+      "epoch": 0.621,
+      "grad_norm": 10.375,
+      "learning_rate": 9.20098010485442e-06,
+      "loss": 1.6746,
+      "step": 6210
+    },
+    {
+      "epoch": 0.622,
+      "grad_norm": 10.3125,
+      "learning_rate": 9.198050455072065e-06,
+      "loss": 1.893,
+      "step": 6220
+    },
+    {
+      "epoch": 0.623,
+      "grad_norm": 20.75,
+      "learning_rate": 9.195115912445599e-06,
+      "loss": 1.6477,
+      "step": 6230
+    },
+    {
+      "epoch": 0.624,
+      "grad_norm": 15.875,
+      "learning_rate": 9.192176480395243e-06,
+      "loss": 1.7614,
+      "step": 6240
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 10.25,
+      "learning_rate": 9.189232162346915e-06,
+      "loss": 2.0202,
+      "step": 6250
+    },
+    {
+      "epoch": 0.626,
+      "grad_norm": 13.5625,
+      "learning_rate": 9.18628296173223e-06,
+      "loss": 1.8276,
+      "step": 6260
+    },
+    {
+      "epoch": 0.627,
+      "grad_norm": 17.25,
+      "learning_rate": 9.183328881988492e-06,
+      "loss": 1.5644,
+      "step": 6270
+    },
+    {
+      "epoch": 0.628,
+      "grad_norm": 13.8125,
+      "learning_rate": 9.180369926558692e-06,
+      "loss": 1.4385,
+      "step": 6280
+    },
+    {
+      "epoch": 0.629,
+      "grad_norm": 11.8125,
+      "learning_rate": 9.177406098891504e-06,
+      "loss": 1.7474,
+      "step": 6290
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 10.25,
+      "learning_rate": 9.174437402441284e-06,
+      "loss": 1.7097,
+      "step": 6300
+    },
+    {
+      "epoch": 0.631,
+      "grad_norm": 15.0,
+      "learning_rate": 9.171463840668056e-06,
+      "loss": 1.6333,
+      "step": 6310
+    },
+    {
+      "epoch": 0.632,
+      "grad_norm": 9.125,
+      "learning_rate": 9.168485417037518e-06,
+      "loss": 1.6146,
+      "step": 6320
+    },
+    {
+      "epoch": 0.633,
+      "grad_norm": 11.3125,
+      "learning_rate": 9.165502135021036e-06,
+      "loss": 1.463,
+      "step": 6330
+    },
+    {
+      "epoch": 0.634,
+      "grad_norm": 17.75,
+      "learning_rate": 9.162513998095634e-06,
+      "loss": 1.7589,
+      "step": 6340
+    },
+    {
+      "epoch": 0.635,
+      "grad_norm": 11.5625,
+      "learning_rate": 9.159521009744e-06,
+      "loss": 1.6267,
+      "step": 6350
+    },
+    {
+      "epoch": 0.636,
+      "grad_norm": 12.1875,
+      "learning_rate": 9.156523173454471e-06,
+      "loss": 1.43,
+      "step": 6360
+    },
+    {
+      "epoch": 0.637,
+      "grad_norm": 10.125,
+      "learning_rate": 9.15352049272104e-06,
+      "loss": 1.8519,
+      "step": 6370
+    },
+    {
+      "epoch": 0.638,
+      "grad_norm": 9.3125,
+      "learning_rate": 9.150512971043339e-06,
+      "loss": 1.4846,
+      "step": 6380
+    },
+    {
+      "epoch": 0.639,
+      "grad_norm": 14.75,
+      "learning_rate": 9.147500611926652e-06,
+      "loss": 1.603,
+      "step": 6390
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 12.875,
+      "learning_rate": 9.14448341888189e-06,
+      "loss": 1.5779,
+      "step": 6400
+    },
+    {
+      "epoch": 0.641,
+      "grad_norm": 11.875,
+      "learning_rate": 9.141461395425604e-06,
+      "loss": 1.8154,
+      "step": 6410
+    },
+    {
+      "epoch": 0.642,
+      "grad_norm": 10.4375,
+      "learning_rate": 9.138434545079974e-06,
+      "loss": 1.4858,
+      "step": 6420
+    },
+    {
+      "epoch": 0.643,
+      "grad_norm": 9.1875,
+      "learning_rate": 9.13540287137281e-06,
+      "loss": 1.6305,
+      "step": 6430
+    },
+    {
+      "epoch": 0.644,
+      "grad_norm": 12.6875,
+      "learning_rate": 9.132366377837535e-06,
+      "loss": 1.5257,
+      "step": 6440
+    },
+    {
+      "epoch": 0.645,
+      "grad_norm": 12.25,
+      "learning_rate": 9.129325068013196e-06,
+      "loss": 1.7423,
+      "step": 6450
+    },
+    {
+      "epoch": 0.646,
+      "grad_norm": 14.6875,
+      "learning_rate": 9.12627894544445e-06,
+      "loss": 1.5406,
+      "step": 6460
+    },
+    {
+      "epoch": 0.647,
+      "grad_norm": 16.875,
+      "learning_rate": 9.123228013681565e-06,
+      "loss": 1.7134,
+      "step": 6470
+    },
+    {
+      "epoch": 0.648,
+      "grad_norm": 10.3125,
+      "learning_rate": 9.120172276280414e-06,
+      "loss": 1.9003,
+      "step": 6480
+    },
+    {
+      "epoch": 0.649,
+      "grad_norm": 9.1875,
+      "learning_rate": 9.117111736802473e-06,
+      "loss": 1.6211,
+      "step": 6490
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 10.25,
+      "learning_rate": 9.114046398814808e-06,
+      "loss": 1.8892,
+      "step": 6500
+    },
+    {
+      "epoch": 0.651,
+      "grad_norm": 12.0625,
+      "learning_rate": 9.110976265890089e-06,
+      "loss": 1.5762,
+      "step": 6510
+    },
+    {
+      "epoch": 0.652,
+      "grad_norm": 9.0625,
+      "learning_rate": 9.10790134160656e-06,
+      "loss": 1.6289,
+      "step": 6520
+    },
+    {
+      "epoch": 0.653,
+      "grad_norm": 11.6875,
+      "learning_rate": 9.104821629548062e-06,
+      "loss": 1.5915,
+      "step": 6530
+    },
+    {
+      "epoch": 0.654,
+      "grad_norm": 9.25,
+      "learning_rate": 9.101737133304012e-06,
+      "loss": 1.663,
+      "step": 6540
+    },
+    {
+      "epoch": 0.655,
+      "grad_norm": 12.625,
+      "learning_rate": 9.098647856469399e-06,
+      "loss": 1.8115,
+      "step": 6550
+    },
+    {
+      "epoch": 0.656,
+      "grad_norm": 9.0,
+      "learning_rate": 9.09555380264479e-06,
+      "loss": 1.5693,
+      "step": 6560
+    },
+    {
+      "epoch": 0.657,
+      "grad_norm": 9.875,
+      "learning_rate": 9.092454975436316e-06,
+      "loss": 1.6829,
+      "step": 6570
+    },
+    {
+      "epoch": 0.658,
+      "grad_norm": 11.1875,
+      "learning_rate": 9.08935137845567e-06,
+      "loss": 1.5629,
+      "step": 6580
+    },
+    {
+      "epoch": 0.659,
+      "grad_norm": 14.5,
+      "learning_rate": 9.086243015320111e-06,
+      "loss": 1.5559,
+      "step": 6590
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 10.1875,
+      "learning_rate": 9.083129889652444e-06,
+      "loss": 1.5256,
+      "step": 6600
+    },
+    {
+      "epoch": 0.661,
+      "grad_norm": 8.6875,
+      "learning_rate": 9.080012005081031e-06,
+      "loss": 1.4865,
+      "step": 6610
+    },
+    {
+      "epoch": 0.662,
+      "grad_norm": 20.625,
+      "learning_rate": 9.076889365239778e-06,
+      "loss": 1.8058,
+      "step": 6620
+    },
+    {
+      "epoch": 0.663,
+      "grad_norm": 17.75,
+      "learning_rate": 9.073761973768134e-06,
+      "loss": 1.7733,
+      "step": 6630
+    },
+    {
+      "epoch": 0.664,
+      "grad_norm": 12.0625,
+      "learning_rate": 9.070629834311087e-06,
+      "loss": 1.5687,
+      "step": 6640
+    },
+    {
+      "epoch": 0.665,
+      "grad_norm": 14.6875,
+      "learning_rate": 9.067492950519154e-06,
+      "loss": 1.867,
+      "step": 6650
+    },
+    {
+      "epoch": 0.666,
+      "grad_norm": 12.5625,
+      "learning_rate": 9.064351326048389e-06,
+      "loss": 1.5874,
+      "step": 6660
+    },
+    {
+      "epoch": 0.667,
+      "grad_norm": 16.125,
+      "learning_rate": 9.061204964560368e-06,
+      "loss": 1.6928,
+      "step": 6670
+    },
+    {
+      "epoch": 0.668,
+      "grad_norm": 16.625,
+      "learning_rate": 9.058053869722184e-06,
+      "loss": 1.6633,
+      "step": 6680
+    },
+    {
+      "epoch": 0.669,
+      "grad_norm": 12.5,
+      "learning_rate": 9.054898045206453e-06,
+      "loss": 1.4977,
+      "step": 6690
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 18.75,
+      "learning_rate": 9.0517374946913e-06,
+      "loss": 1.6687,
+      "step": 6700
+    },
+    {
+      "epoch": 0.671,
+      "grad_norm": 18.75,
+      "learning_rate": 9.048572221860358e-06,
+      "loss": 1.6292,
+      "step": 6710
+    },
+    {
+      "epoch": 0.672,
+      "grad_norm": 15.5625,
+      "learning_rate": 9.045402230402767e-06,
+      "loss": 1.8846,
+      "step": 6720
+    },
+    {
+      "epoch": 0.673,
+      "grad_norm": 15.875,
+      "learning_rate": 9.042227524013164e-06,
+      "loss": 1.8538,
+      "step": 6730
+    },
+    {
+      "epoch": 0.674,
+      "grad_norm": 12.0,
+      "learning_rate": 9.039048106391677e-06,
+      "loss": 1.4716,
+      "step": 6740
+    },
+    {
+      "epoch": 0.675,
+      "grad_norm": 15.75,
+      "learning_rate": 9.035863981243935e-06,
+      "loss": 1.5906,
+      "step": 6750
+    },
+    {
+      "epoch": 0.676,
+      "grad_norm": 13.0625,
+      "learning_rate": 9.032675152281048e-06,
+      "loss": 1.7296,
+      "step": 6760
+    },
+    {
+      "epoch": 0.677,
+      "grad_norm": 13.125,
+      "learning_rate": 9.029481623219604e-06,
+      "loss": 1.6592,
+      "step": 6770
+    },
+    {
+      "epoch": 0.678,
+      "grad_norm": 14.8125,
+      "learning_rate": 9.026283397781679e-06,
+      "loss": 1.6597,
+      "step": 6780
+    },
+    {
+      "epoch": 0.679,
+      "grad_norm": 17.0,
+      "learning_rate": 9.023080479694813e-06,
+      "loss": 4.117,
+      "step": 6790
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 13.5625,
+      "learning_rate": 9.019872872692021e-06,
+      "loss": 1.6793,
+      "step": 6800
+    },
+    {
+      "epoch": 0.681,
+      "grad_norm": 17.25,
+      "learning_rate": 9.016660580511783e-06,
+      "loss": 1.9001,
+      "step": 6810
+    },
+    {
+      "epoch": 0.682,
+      "grad_norm": 10.75,
+      "learning_rate": 9.013443606898036e-06,
+      "loss": 1.5253,
+      "step": 6820
+    },
+    {
+      "epoch": 0.683,
+      "grad_norm": 11.25,
+      "learning_rate": 9.010221955600175e-06,
+      "loss": 1.6314,
+      "step": 6830
+    },
+    {
+      "epoch": 0.684,
+      "grad_norm": 13.0,
+      "learning_rate": 9.00699563037305e-06,
+      "loss": 2.0591,
+      "step": 6840
+    },
+    {
+      "epoch": 0.685,
+      "grad_norm": 14.875,
+      "learning_rate": 9.003764634976951e-06,
+      "loss": 1.7015,
+      "step": 6850
+    },
+    {
+      "epoch": 0.686,
+      "grad_norm": 12.5625,
+      "learning_rate": 9.00052897317762e-06,
+      "loss": 1.7212,
+      "step": 6860
+    },
+    {
+      "epoch": 0.687,
+      "grad_norm": 12.8125,
+      "learning_rate": 8.997288648746229e-06,
+      "loss": 1.7654,
+      "step": 6870
+    },
+    {
+      "epoch": 0.688,
+      "grad_norm": 13.375,
+      "learning_rate": 8.994043665459393e-06,
+      "loss": 1.6362,
+      "step": 6880
+    },
+    {
+      "epoch": 0.689,
+      "grad_norm": 8.9375,
+      "learning_rate": 8.990794027099153e-06,
+      "loss": 2.0192,
+      "step": 6890
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 30.0,
+      "learning_rate": 8.98753973745297e-06,
+      "loss": 1.4845,
+      "step": 6900
+    },
+    {
+      "epoch": 0.691,
+      "grad_norm": 13.1875,
+      "learning_rate": 8.984280800313735e-06,
+      "loss": 1.4502,
+      "step": 6910
+    },
+    {
+      "epoch": 0.692,
+      "grad_norm": 15.1875,
+      "learning_rate": 8.981017219479753e-06,
+      "loss": 1.73,
+      "step": 6920
+    },
+    {
+      "epoch": 0.693,
+      "grad_norm": 12.5625,
+      "learning_rate": 8.977748998754736e-06,
+      "loss": 1.3641,
+      "step": 6930
+    },
+    {
+      "epoch": 0.694,
+      "grad_norm": 12.25,
+      "learning_rate": 8.97447614194781e-06,
+      "loss": 1.7327,
+      "step": 6940
+    },
+    {
+      "epoch": 0.695,
+      "grad_norm": 16.375,
+      "learning_rate": 8.971198652873505e-06,
+      "loss": 1.7731,
+      "step": 6950
+    },
+    {
+      "epoch": 0.696,
+      "grad_norm": 9.8125,
+      "learning_rate": 8.967916535351745e-06,
+      "loss": 1.5143,
+      "step": 6960
+    },
+    {
+      "epoch": 0.697,
+      "grad_norm": 12.0625,
+      "learning_rate": 8.964629793207847e-06,
+      "loss": 1.797,
+      "step": 6970
+    },
+    {
+      "epoch": 0.698,
+      "grad_norm": 14.9375,
+      "learning_rate": 8.961338430272528e-06,
+      "loss": 3.1581,
+      "step": 6980
+    },
+    {
+      "epoch": 0.699,
+      "grad_norm": 17.25,
+      "learning_rate": 8.958042450381881e-06,
+      "loss": 1.7383,
+      "step": 6990
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 10.25,
+      "learning_rate": 8.954741857377383e-06,
+      "loss": 1.5392,
+      "step": 7000
+    },
+    {
+      "epoch": 0.701,
+      "grad_norm": 11.0,
+      "learning_rate": 8.95143665510589e-06,
+      "loss": 1.5256,
+      "step": 7010
+    },
+    {
+      "epoch": 0.702,
+      "grad_norm": 11.3125,
+      "learning_rate": 8.948126847419626e-06,
+      "loss": 1.7499,
+      "step": 7020
+    },
+    {
+      "epoch": 0.703,
+      "grad_norm": 17.0,
+      "learning_rate": 8.944812438176185e-06,
+      "loss": 1.9737,
+      "step": 7030
+    },
+    {
+      "epoch": 0.704,
+      "grad_norm": 23.5,
+      "learning_rate": 8.941493431238527e-06,
+      "loss": 1.7724,
+      "step": 7040
+    },
+    {
+      "epoch": 0.705,
+      "grad_norm": 11.875,
+      "learning_rate": 8.938169830474963e-06,
+      "loss": 1.7497,
+      "step": 7050
+    },
+    {
+      "epoch": 0.706,
+      "grad_norm": 9.75,
+      "learning_rate": 8.934841639759167e-06,
+      "loss": 1.6256,
+      "step": 7060
+    },
+    {
+      "epoch": 0.707,
+      "grad_norm": 10.9375,
+      "learning_rate": 8.931508862970153e-06,
+      "loss": 1.8323,
+      "step": 7070
+    },
+    {
+      "epoch": 0.708,
+      "grad_norm": 12.9375,
+      "learning_rate": 8.92817150399229e-06,
+      "loss": 1.6482,
+      "step": 7080
+    },
+    {
+      "epoch": 0.709,
+      "grad_norm": 14.8125,
+      "learning_rate": 8.92482956671528e-06,
+      "loss": 1.6025,
+      "step": 7090
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 12.5625,
+      "learning_rate": 8.921483055034166e-06,
+      "loss": 1.8051,
+      "step": 7100
+    },
+    {
+      "epoch": 0.711,
+      "grad_norm": 12.25,
+      "learning_rate": 8.918131972849317e-06,
+      "loss": 1.5073,
+      "step": 7110
+    },
+    {
+      "epoch": 0.712,
+      "grad_norm": 13.1875,
+      "learning_rate": 8.914776324066436e-06,
+      "loss": 1.6819,
+      "step": 7120
+    },
+    {
+      "epoch": 0.713,
+      "grad_norm": 15.875,
+      "learning_rate": 8.911416112596543e-06,
+      "loss": 1.6594,
+      "step": 7130
+    },
+    {
+      "epoch": 0.714,
+      "grad_norm": 12.625,
+      "learning_rate": 8.908051342355979e-06,
+      "loss": 1.7567,
+      "step": 7140
+    },
+    {
+      "epoch": 0.715,
+      "grad_norm": 12.875,
+      "learning_rate": 8.904682017266394e-06,
+      "loss": 1.6558,
+      "step": 7150
+    },
+    {
+      "epoch": 0.716,
+      "grad_norm": 13.3125,
+      "learning_rate": 8.90130814125475e-06,
+      "loss": 1.4967,
+      "step": 7160
+    },
+    {
+      "epoch": 0.717,
+      "grad_norm": 9.125,
+      "learning_rate": 8.897929718253315e-06,
+      "loss": 1.5906,
+      "step": 7170
+    },
+    {
+      "epoch": 0.718,
+      "grad_norm": 13.8125,
+      "learning_rate": 8.894546752199654e-06,
+      "loss": 1.508,
+      "step": 7180
+    },
+    {
+      "epoch": 0.719,
+      "grad_norm": 13.0,
+      "learning_rate": 8.891159247036624e-06,
+      "loss": 1.6976,
+      "step": 7190
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 17.375,
+      "learning_rate": 8.887767206712378e-06,
+      "loss": 1.5396,
+      "step": 7200
+    },
+    {
+      "epoch": 0.721,
+      "grad_norm": 11.875,
+      "learning_rate": 8.88437063518035e-06,
+      "loss": 1.7258,
+      "step": 7210
+    },
+    {
+      "epoch": 0.722,
+      "grad_norm": 11.875,
+      "learning_rate": 8.88096953639926e-06,
+      "loss": 1.5536,
+      "step": 7220
+    },
+    {
+      "epoch": 0.723,
+      "grad_norm": 21.125,
+      "learning_rate": 8.877563914333101e-06,
+      "loss": 1.5561,
+      "step": 7230
+    },
+    {
+      "epoch": 0.724,
+      "grad_norm": 11.5625,
+      "learning_rate": 8.874153772951138e-06,
+      "loss": 1.6298,
+      "step": 7240
+    },
+    {
+      "epoch": 0.725,
+      "grad_norm": 19.625,
+      "learning_rate": 8.870739116227902e-06,
+      "loss": 1.82,
+      "step": 7250
+    },
+    {
+      "epoch": 0.726,
+      "grad_norm": 19.875,
+      "learning_rate": 8.867319948143193e-06,
+      "loss": 1.7358,
+      "step": 7260
+    },
+    {
+      "epoch": 0.727,
+      "grad_norm": 13.9375,
+      "learning_rate": 8.863896272682061e-06,
+      "loss": 1.6065,
+      "step": 7270
+    },
+    {
+      "epoch": 0.728,
+      "grad_norm": 13.125,
+      "learning_rate": 8.860468093834818e-06,
+      "loss": 1.7938,
+      "step": 7280
+    },
+    {
+      "epoch": 0.729,
+      "grad_norm": 19.375,
+      "learning_rate": 8.857035415597015e-06,
+      "loss": 1.7685,
+      "step": 7290
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 11.125,
+      "learning_rate": 8.853598241969451e-06,
+      "loss": 1.5424,
+      "step": 7300
+    },
+    {
+      "epoch": 0.731,
+      "grad_norm": 13.5,
+      "learning_rate": 8.850156576958171e-06,
+      "loss": 1.6812,
+      "step": 7310
+    },
+    {
+      "epoch": 0.732,
+      "grad_norm": 9.5625,
+      "learning_rate": 8.846710424574443e-06,
+      "loss": 1.3544,
+      "step": 7320
+    },
+    {
+      "epoch": 0.733,
+      "grad_norm": 11.625,
+      "learning_rate": 8.843259788834775e-06,
+      "loss": 1.6644,
+      "step": 7330
+    },
+    {
+      "epoch": 0.734,
+      "grad_norm": 16.625,
+      "learning_rate": 8.839804673760895e-06,
+      "loss": 1.8758,
+      "step": 7340
+    },
+    {
+      "epoch": 0.735,
+      "grad_norm": 13.1875,
+      "learning_rate": 8.836345083379752e-06,
+      "loss": 1.6584,
+      "step": 7350
+    },
+    {
+      "epoch": 0.736,
+      "grad_norm": 10.6875,
+      "learning_rate": 8.832881021723511e-06,
+      "loss": 1.9174,
+      "step": 7360
+    },
+    {
+      "epoch": 0.737,
+      "grad_norm": 16.25,
+      "learning_rate": 8.829412492829554e-06,
+      "loss": 1.7212,
+      "step": 7370
+    },
+    {
+      "epoch": 0.738,
+      "grad_norm": 16.875,
+      "learning_rate": 8.825939500740462e-06,
+      "loss": 1.7066,
+      "step": 7380
+    },
+    {
+      "epoch": 0.739,
+      "grad_norm": 14.5625,
+      "learning_rate": 8.822462049504019e-06,
+      "loss": 1.6863,
+      "step": 7390
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 14.125,
+      "learning_rate": 8.818980143173212e-06,
+      "loss": 1.6795,
+      "step": 7400
+    },
+    {
+      "epoch": 0.741,
+      "grad_norm": 11.3125,
+      "learning_rate": 8.815493785806212e-06,
+      "loss": 1.5905,
+      "step": 7410
+    },
+    {
+      "epoch": 0.742,
+      "grad_norm": 10.875,
+      "learning_rate": 8.812002981466385e-06,
+      "loss": 1.7352,
+      "step": 7420
+    },
+    {
+      "epoch": 0.743,
+      "grad_norm": 16.25,
+      "learning_rate": 8.808507734222278e-06,
+      "loss": 1.6931,
+      "step": 7430
+    },
+    {
+      "epoch": 0.744,
+      "grad_norm": 14.875,
+      "learning_rate": 8.80500804814761e-06,
+      "loss": 1.6367,
+      "step": 7440
+    },
+    {
+      "epoch": 0.745,
+      "grad_norm": 8.0625,
+      "learning_rate": 8.801503927321285e-06,
+      "loss": 1.457,
+      "step": 7450
+    },
+    {
+      "epoch": 0.746,
+      "grad_norm": 16.25,
+      "learning_rate": 8.797995375827366e-06,
+      "loss": 1.7783,
+      "step": 7460
+    },
+    {
+      "epoch": 0.747,
+      "grad_norm": 11.1875,
+      "learning_rate": 8.794482397755082e-06,
+      "loss": 1.3363,
+      "step": 7470
+    },
+    {
+      "epoch": 0.748,
+      "grad_norm": 14.0,
+      "learning_rate": 8.790964997198827e-06,
+      "loss": 1.7542,
+      "step": 7480
+    },
+    {
+      "epoch": 0.749,
+      "grad_norm": 13.125,
+      "learning_rate": 8.787443178258142e-06,
+      "loss": 1.8713,
+      "step": 7490
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 13.1875,
+      "learning_rate": 8.78391694503772e-06,
+      "loss": 1.5536,
+      "step": 7500
+    },
+    {
+      "epoch": 0.751,
+      "grad_norm": 10.5625,
+      "learning_rate": 8.7803863016474e-06,
+      "loss": 1.6686,
+      "step": 7510
+    },
+    {
+      "epoch": 0.752,
+      "grad_norm": 13.25,
+      "learning_rate": 8.776851252202166e-06,
+      "loss": 1.6434,
+      "step": 7520
+    },
+    {
+      "epoch": 0.753,
+      "grad_norm": 14.5625,
+      "learning_rate": 8.773311800822123e-06,
+      "loss": 1.6628,
+      "step": 7530
+    },
+    {
+      "epoch": 0.754,
+      "grad_norm": 12.8125,
+      "learning_rate": 8.76976795163252e-06,
+      "loss": 1.4194,
+      "step": 7540
+    },
+    {
+      "epoch": 0.755,
+      "grad_norm": 19.625,
+      "learning_rate": 8.766219708763727e-06,
+      "loss": 1.624,
+      "step": 7550
+    },
+    {
+      "epoch": 0.756,
+      "grad_norm": 12.1875,
+      "learning_rate": 8.762667076351236e-06,
+      "loss": 1.6434,
+      "step": 7560
+    },
+    {
+      "epoch": 0.757,
+      "grad_norm": 12.6875,
+      "learning_rate": 8.75911005853565e-06,
+      "loss": 1.6848,
+      "step": 7570
+    },
+    {
+      "epoch": 0.758,
+      "grad_norm": 11.1875,
+      "learning_rate": 8.75554865946269e-06,
+      "loss": 1.8256,
+      "step": 7580
+    },
+    {
+      "epoch": 0.759,
+      "grad_norm": 18.0,
+      "learning_rate": 8.751982883283182e-06,
+      "loss": 1.7122,
+      "step": 7590
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 12.125,
+      "learning_rate": 8.748412734153046e-06,
+      "loss": 1.5689,
+      "step": 7600
+    },
+    {
+      "epoch": 0.761,
+      "grad_norm": 17.25,
+      "learning_rate": 8.74483821623331e-06,
+      "loss": 1.3936,
+      "step": 7610
+    },
+    {
+      "epoch": 0.762,
+      "grad_norm": 13.125,
+      "learning_rate": 8.741259333690088e-06,
+      "loss": 1.4154,
+      "step": 7620
+    },
+    {
+      "epoch": 0.763,
+      "grad_norm": 9.5,
+      "learning_rate": 8.73767609069458e-06,
+      "loss": 1.9622,
+      "step": 7630
+    },
+    {
+      "epoch": 0.764,
+      "grad_norm": 12.5,
+      "learning_rate": 8.73408849142307e-06,
+      "loss": 1.5571,
+      "step": 7640
+    },
+    {
+      "epoch": 0.765,
+      "grad_norm": 13.5625,
+      "learning_rate": 8.730496540056917e-06,
+      "loss": 1.4703,
+      "step": 7650
+    },
+    {
+      "epoch": 0.766,
+      "grad_norm": 14.5,
+      "learning_rate": 8.726900240782559e-06,
+      "loss": 1.5589,
+      "step": 7660
+    },
+    {
+      "epoch": 0.767,
+      "grad_norm": 13.625,
+      "learning_rate": 8.723299597791492e-06,
+      "loss": 1.6046,
+      "step": 7670
+    },
+    {
+      "epoch": 0.768,
+      "grad_norm": 16.75,
+      "learning_rate": 8.719694615280283e-06,
+      "loss": 1.8941,
+      "step": 7680
+    },
+    {
+      "epoch": 0.769,
+      "grad_norm": 14.1875,
+      "learning_rate": 8.716085297450551e-06,
+      "loss": 1.572,
+      "step": 7690
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 10.9375,
+      "learning_rate": 8.71247164850897e-06,
+      "loss": 1.6006,
+      "step": 7700
+    },
+    {
+      "epoch": 0.771,
+      "grad_norm": 13.6875,
+      "learning_rate": 8.708853672667263e-06,
+      "loss": 1.5284,
+      "step": 7710
+    },
+    {
+      "epoch": 0.772,
+      "grad_norm": 11.4375,
+      "learning_rate": 8.705231374142194e-06,
+      "loss": 1.7202,
+      "step": 7720
+    },
+    {
+      "epoch": 0.773,
+      "grad_norm": 13.8125,
+      "learning_rate": 8.701604757155566e-06,
+      "loss": 2.6671,
+      "step": 7730
+    },
+    {
+      "epoch": 0.774,
+      "grad_norm": 12.125,
+      "learning_rate": 8.697973825934215e-06,
+      "loss": 1.5748,
+      "step": 7740
+    },
+    {
+      "epoch": 0.775,
+      "grad_norm": 11.375,
+      "learning_rate": 8.694338584710008e-06,
+      "loss": 1.7724,
+      "step": 7750
+    },
+    {
+      "epoch": 0.776,
+      "grad_norm": 14.1875,
+      "learning_rate": 8.69069903771983e-06,
+      "loss": 1.3886,
+      "step": 7760
+    },
+    {
+      "epoch": 0.777,
+      "grad_norm": 11.9375,
+      "learning_rate": 8.687055189205587e-06,
+      "loss": 1.4272,
+      "step": 7770
+    },
+    {
+      "epoch": 0.778,
+      "grad_norm": 19.0,
+      "learning_rate": 8.683407043414197e-06,
+      "loss": 1.6348,
+      "step": 7780
+    },
+    {
+      "epoch": 0.779,
+      "grad_norm": 11.25,
+      "learning_rate": 8.679754604597595e-06,
+      "loss": 1.7418,
+      "step": 7790
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 12.125,
+      "learning_rate": 8.676097877012706e-06,
+      "loss": 3.8698,
+      "step": 7800
+    },
+    {
+      "epoch": 0.781,
+      "grad_norm": 13.125,
+      "learning_rate": 8.672436864921463e-06,
+      "loss": 1.6474,
+      "step": 7810
+    },
+    {
+      "epoch": 0.782,
+      "grad_norm": 14.625,
+      "learning_rate": 8.668771572590789e-06,
+      "loss": 1.7423,
+      "step": 7820
+    },
+    {
+      "epoch": 0.783,
+      "grad_norm": 13.6875,
+      "learning_rate": 8.665102004292595e-06,
+      "loss": 1.5277,
+      "step": 7830
+    },
+    {
+      "epoch": 0.784,
+      "grad_norm": 13.875,
+      "learning_rate": 8.66142816430378e-06,
+      "loss": 1.6163,
+      "step": 7840
+    },
+    {
+      "epoch": 0.785,
+      "grad_norm": 10.1875,
+      "learning_rate": 8.657750056906218e-06,
+      "loss": 1.6906,
+      "step": 7850
+    },
+    {
+      "epoch": 0.786,
+      "grad_norm": 11.6875,
+      "learning_rate": 8.654067686386758e-06,
+      "loss": 1.695,
+      "step": 7860
+    },
+    {
+      "epoch": 0.787,
+      "grad_norm": 21.75,
+      "learning_rate": 8.650381057037214e-06,
+      "loss": 2.2176,
+      "step": 7870
+    },
+    {
+      "epoch": 0.788,
+      "grad_norm": 16.875,
+      "learning_rate": 8.646690173154369e-06,
+      "loss": 1.8026,
+      "step": 7880
+    },
+    {
+      "epoch": 0.789,
+      "grad_norm": 13.875,
+      "learning_rate": 8.642995039039963e-06,
+      "loss": 1.5637,
+      "step": 7890
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 16.0,
+      "learning_rate": 8.639295659000688e-06,
+      "loss": 1.7421,
+      "step": 7900
+    },
+    {
+      "epoch": 0.791,
+      "grad_norm": 17.5,
+      "learning_rate": 8.635592037348185e-06,
+      "loss": 1.7596,
+      "step": 7910
+    },
+    {
+      "epoch": 0.792,
+      "grad_norm": 10.5625,
+      "learning_rate": 8.631884178399044e-06,
+      "loss": 1.6035,
+      "step": 7920
+    },
+    {
+      "epoch": 0.793,
+      "grad_norm": 12.5,
+      "learning_rate": 8.628172086474784e-06,
+      "loss": 1.6658,
+      "step": 7930
+    },
+    {
+      "epoch": 0.794,
+      "grad_norm": 11.0625,
+      "learning_rate": 8.624455765901865e-06,
+      "loss": 1.8439,
+      "step": 7940
+    },
+    {
+      "epoch": 0.795,
+      "grad_norm": 9.5625,
+      "learning_rate": 8.620735221011673e-06,
+      "loss": 1.6354,
+      "step": 7950
+    },
+    {
+      "epoch": 0.796,
+      "grad_norm": 12.1875,
+      "learning_rate": 8.617010456140515e-06,
+      "loss": 1.4467,
+      "step": 7960
+    },
+    {
+      "epoch": 0.797,
+      "grad_norm": 14.25,
+      "learning_rate": 8.613281475629622e-06,
+      "loss": 1.5,
+      "step": 7970
+    },
+    {
+      "epoch": 0.798,
+      "grad_norm": 17.75,
+      "learning_rate": 8.609548283825134e-06,
+      "loss": 1.6453,
+      "step": 7980
+    },
+    {
+      "epoch": 0.799,
+      "grad_norm": 18.75,
+      "learning_rate": 8.6058108850781e-06,
+      "loss": 1.7577,
+      "step": 7990
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 11.9375,
+      "learning_rate": 8.602069283744476e-06,
+      "loss": 1.5583,
+      "step": 8000
+    },
+    {
+      "epoch": 0.801,
+      "grad_norm": 15.1875,
+      "learning_rate": 8.598323484185108e-06,
+      "loss": 1.542,
+      "step": 8010
+    },
+    {
+      "epoch": 0.802,
+      "grad_norm": 11.3125,
+      "learning_rate": 8.594573490765743e-06,
+      "loss": 1.633,
+      "step": 8020
+    },
+    {
+      "epoch": 0.803,
+      "grad_norm": 12.25,
+      "learning_rate": 8.590819307857013e-06,
+      "loss": 1.6265,
+      "step": 8030
+    },
+    {
+      "epoch": 0.804,
+      "grad_norm": 15.5,
+      "learning_rate": 8.587060939834431e-06,
+      "loss": 1.6642,
+      "step": 8040
+    },
+    {
+      "epoch": 0.805,
+      "grad_norm": 10.9375,
+      "learning_rate": 8.583298391078389e-06,
+      "loss": 1.8216,
+      "step": 8050
+    },
+    {
+      "epoch": 0.806,
+      "grad_norm": 11.875,
+      "learning_rate": 8.579531665974155e-06,
+      "loss": 1.5399,
+      "step": 8060
+    },
+    {
+      "epoch": 0.807,
+      "grad_norm": 13.125,
+      "learning_rate": 8.575760768911862e-06,
+      "loss": 1.5519,
+      "step": 8070
+    },
+    {
+      "epoch": 0.808,
+      "grad_norm": 12.3125,
+      "learning_rate": 8.5719857042865e-06,
+      "loss": 1.7087,
+      "step": 8080
+    },
+    {
+      "epoch": 0.809,
+      "grad_norm": 26.5,
+      "learning_rate": 8.568206476497929e-06,
+      "loss": 1.6511,
+      "step": 8090
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 17.375,
+      "learning_rate": 8.564423089950847e-06,
+      "loss": 1.9342,
+      "step": 8100
+    },
+    {
+      "epoch": 0.811,
+      "grad_norm": 12.5,
+      "learning_rate": 8.56063554905481e-06,
+      "loss": 1.3681,
+      "step": 8110
+    },
+    {
+      "epoch": 0.812,
+      "grad_norm": 11.25,
+      "learning_rate": 8.55684385822421e-06,
+      "loss": 1.6698,
+      "step": 8120
+    },
+    {
+      "epoch": 0.813,
+      "grad_norm": 8.5625,
+      "learning_rate": 8.553048021878277e-06,
+      "loss": 1.4201,
+      "step": 8130
+    },
+    {
+      "epoch": 0.814,
+      "grad_norm": 11.5625,
+      "learning_rate": 8.549248044441072e-06,
+      "loss": 1.8975,
+      "step": 8140
+    },
+    {
+      "epoch": 0.815,
+      "grad_norm": 13.75,
+      "learning_rate": 8.545443930341486e-06,
+      "loss": 1.5355,
+      "step": 8150
+    },
+    {
+      "epoch": 0.816,
+      "grad_norm": 17.0,
+      "learning_rate": 8.541635684013227e-06,
+      "loss": 1.7242,
+      "step": 8160
+    },
+    {
+      "epoch": 0.817,
+      "grad_norm": 12.4375,
+      "learning_rate": 8.537823309894823e-06,
+      "loss": 1.7754,
+      "step": 8170
+    },
+    {
+      "epoch": 0.818,
+      "grad_norm": 11.6875,
+      "learning_rate": 8.53400681242961e-06,
+      "loss": 1.425,
+      "step": 8180
+    },
+    {
+      "epoch": 0.819,
+      "grad_norm": 11.875,
+      "learning_rate": 8.530186196065728e-06,
+      "loss": 1.657,
+      "step": 8190
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 13.0625,
+      "learning_rate": 8.526361465256123e-06,
+      "loss": 1.5268,
+      "step": 8200
+    },
+    {
+      "epoch": 0.821,
+      "grad_norm": 14.5,
+      "learning_rate": 8.522532624458533e-06,
+      "loss": 1.7827,
+      "step": 8210
+    },
+    {
+      "epoch": 0.822,
+      "grad_norm": 14.5625,
+      "learning_rate": 8.518699678135486e-06,
+      "loss": 1.6268,
+      "step": 8220
+    },
+    {
+      "epoch": 0.823,
+      "grad_norm": 11.8125,
+      "learning_rate": 8.514862630754297e-06,
+      "loss": 1.5976,
+      "step": 8230
+    },
+    {
+      "epoch": 0.824,
+      "grad_norm": 15.625,
+      "learning_rate": 8.51102148678706e-06,
+      "loss": 1.5027,
+      "step": 8240
+    },
+    {
+      "epoch": 0.825,
+      "grad_norm": 14.4375,
+      "learning_rate": 8.50717625071064e-06,
+      "loss": 1.9334,
+      "step": 8250
+    },
+    {
+      "epoch": 0.826,
+      "grad_norm": 12.375,
+      "learning_rate": 8.503326927006681e-06,
+      "loss": 1.7559,
+      "step": 8260
+    },
+    {
+      "epoch": 0.827,
+      "grad_norm": 14.5,
+      "learning_rate": 8.499473520161578e-06,
+      "loss": 1.7564,
+      "step": 8270
+    },
+    {
+      "epoch": 0.828,
+      "grad_norm": 14.0,
+      "learning_rate": 8.495616034666495e-06,
+      "loss": 1.6141,
+      "step": 8280
+    },
+    {
+      "epoch": 0.829,
+      "grad_norm": 13.8125,
+      "learning_rate": 8.491754475017344e-06,
+      "loss": 1.515,
+      "step": 8290
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 14.75,
+      "learning_rate": 8.48788884571479e-06,
+      "loss": 1.6165,
+      "step": 8300
+    },
+    {
+      "epoch": 0.831,
+      "grad_norm": 15.8125,
+      "learning_rate": 8.484019151264239e-06,
+      "loss": 1.4339,
+      "step": 8310
+    },
+    {
+      "epoch": 0.832,
+      "grad_norm": 19.0,
+      "learning_rate": 8.480145396175834e-06,
+      "loss": 1.7732,
+      "step": 8320
+    },
+    {
+      "epoch": 0.833,
+      "grad_norm": 9.9375,
+      "learning_rate": 8.476267584964454e-06,
+      "loss": 1.7103,
+      "step": 8330
+    },
+    {
+      "epoch": 0.834,
+      "grad_norm": 18.0,
+      "learning_rate": 8.472385722149697e-06,
+      "loss": 1.5345,
+      "step": 8340
+    },
+    {
+      "epoch": 0.835,
+      "grad_norm": 16.5,
+      "learning_rate": 8.468499812255896e-06,
+      "loss": 1.9681,
+      "step": 8350
+    },
+    {
+      "epoch": 0.836,
+      "grad_norm": 28.0,
+      "learning_rate": 8.464609859812089e-06,
+      "loss": 1.85,
+      "step": 8360
+    },
+    {
+      "epoch": 0.837,
+      "grad_norm": 24.125,
+      "learning_rate": 8.460715869352035e-06,
+      "loss": 1.9731,
+      "step": 8370
+    },
+    {
+      "epoch": 0.838,
+      "grad_norm": 24.625,
+      "learning_rate": 8.456817845414191e-06,
+      "loss": 1.7324,
+      "step": 8380
+    },
+    {
+      "epoch": 0.839,
+      "grad_norm": 8.0,
+      "learning_rate": 8.452915792541723e-06,
+      "loss": 1.6852,
+      "step": 8390
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 13.9375,
+      "learning_rate": 8.449009715282486e-06,
+      "loss": 1.4861,
+      "step": 8400
+    },
+    {
+      "epoch": 0.841,
+      "grad_norm": 11.875,
+      "learning_rate": 8.445099618189028e-06,
+      "loss": 1.5005,
+      "step": 8410
+    },
+    {
+      "epoch": 0.842,
+      "grad_norm": 16.0,
+      "learning_rate": 8.441185505818583e-06,
+      "loss": 1.5594,
+      "step": 8420
+    },
+    {
+      "epoch": 0.843,
+      "grad_norm": 13.375,
+      "learning_rate": 8.437267382733065e-06,
+      "loss": 1.7025,
+      "step": 8430
+    },
+    {
+      "epoch": 0.844,
+      "grad_norm": 15.5,
+      "learning_rate": 8.433345253499062e-06,
+      "loss": 1.506,
+      "step": 8440
+    },
+    {
+      "epoch": 0.845,
+      "grad_norm": 14.3125,
+      "learning_rate": 8.42941912268783e-06,
+      "loss": 1.4843,
+      "step": 8450
+    },
+    {
+      "epoch": 0.846,
+      "grad_norm": 14.1875,
+      "learning_rate": 8.425488994875289e-06,
+      "loss": 1.5652,
+      "step": 8460
+    },
+    {
+      "epoch": 0.847,
+      "grad_norm": 11.125,
+      "learning_rate": 8.421554874642018e-06,
+      "loss": 1.4776,
+      "step": 8470
+    },
+    {
+      "epoch": 0.848,
+      "grad_norm": 16.25,
+      "learning_rate": 8.417616766573251e-06,
+      "loss": 1.6741,
+      "step": 8480
+    },
+    {
+      "epoch": 0.849,
+      "grad_norm": 18.0,
+      "learning_rate": 8.413674675258868e-06,
+      "loss": 2.4516,
+      "step": 8490
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 12.375,
+      "learning_rate": 8.40972860529339e-06,
+      "loss": 1.6855,
+      "step": 8500
+    },
+    {
+      "epoch": 0.851,
+      "grad_norm": 12.6875,
+      "learning_rate": 8.405778561275979e-06,
+      "loss": 1.8114,
+      "step": 8510
+    },
+    {
+      "epoch": 0.852,
+      "grad_norm": 16.125,
+      "learning_rate": 8.401824547810426e-06,
+      "loss": 1.7166,
+      "step": 8520
+    },
+    {
+      "epoch": 0.853,
+      "grad_norm": 15.0,
+      "learning_rate": 8.39786656950515e-06,
+      "loss": 1.635,
+      "step": 8530
+    },
+    {
+      "epoch": 0.854,
+      "grad_norm": 12.3125,
+      "learning_rate": 8.393904630973188e-06,
+      "loss": 1.5757,
+      "step": 8540
+    },
+    {
+      "epoch": 0.855,
+      "grad_norm": 14.4375,
+      "learning_rate": 8.389938736832195e-06,
+      "loss": 1.9013,
+      "step": 8550
+    },
+    {
+      "epoch": 0.856,
+      "grad_norm": 13.625,
+      "learning_rate": 8.385968891704436e-06,
+      "loss": 1.8897,
+      "step": 8560
+    },
+    {
+      "epoch": 0.857,
+      "grad_norm": 13.0625,
+      "learning_rate": 8.381995100216785e-06,
+      "loss": 1.5768,
+      "step": 8570
+    },
+    {
+      "epoch": 0.858,
+      "grad_norm": 15.5,
+      "learning_rate": 8.378017367000706e-06,
+      "loss": 1.755,
+      "step": 8580
+    },
+    {
+      "epoch": 0.859,
+      "grad_norm": 9.25,
+      "learning_rate": 8.374035696692264e-06,
+      "loss": 1.6164,
+      "step": 8590
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 14.25,
+      "learning_rate": 8.370050093932111e-06,
+      "loss": 1.9127,
+      "step": 8600
+    },
+    {
+      "epoch": 0.861,
+      "grad_norm": 12.75,
+      "learning_rate": 8.366060563365486e-06,
+      "loss": 1.7116,
+      "step": 8610
+    },
+    {
+      "epoch": 0.862,
+      "grad_norm": 16.25,
+      "learning_rate": 8.362067109642198e-06,
+      "loss": 1.4693,
+      "step": 8620
+    },
+    {
+      "epoch": 0.863,
+      "grad_norm": 14.875,
+      "learning_rate": 8.358069737416635e-06,
+      "loss": 1.6014,
+      "step": 8630
+    },
+    {
+      "epoch": 0.864,
+      "grad_norm": 9.4375,
+      "learning_rate": 8.35406845134775e-06,
+      "loss": 1.5829,
+      "step": 8640
+    },
+    {
+      "epoch": 0.865,
+      "grad_norm": 12.25,
+      "learning_rate": 8.350063256099055e-06,
+      "loss": 1.8925,
+      "step": 8650
+    },
+    {
+      "epoch": 0.866,
+      "grad_norm": 26.25,
+      "learning_rate": 8.346054156338623e-06,
+      "loss": 2.1381,
+      "step": 8660
+    },
+    {
+      "epoch": 0.867,
+      "grad_norm": 12.5,
+      "learning_rate": 8.342041156739076e-06,
+      "loss": 1.6108,
+      "step": 8670
+    },
+    {
+      "epoch": 0.868,
+      "grad_norm": 12.125,
+      "learning_rate": 8.338024261977578e-06,
+      "loss": 1.7501,
+      "step": 8680
+    },
+    {
+      "epoch": 0.869,
+      "grad_norm": 15.75,
+      "learning_rate": 8.334003476735838e-06,
+      "loss": 1.6533,
+      "step": 8690
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 10.8125,
+      "learning_rate": 8.329978805700096e-06,
+      "loss": 1.622,
+      "step": 8700
+    },
+    {
+      "epoch": 0.871,
+      "grad_norm": 21.375,
+      "learning_rate": 8.32595025356112e-06,
+      "loss": 1.667,
+      "step": 8710
+    },
+    {
+      "epoch": 0.872,
+      "grad_norm": 15.3125,
+      "learning_rate": 8.321917825014204e-06,
+      "loss": 1.7687,
+      "step": 8720
+    },
+    {
+      "epoch": 0.873,
+      "grad_norm": 10.4375,
+      "learning_rate": 8.31788152475916e-06,
+      "loss": 1.4987,
+      "step": 8730
+    },
+    {
+      "epoch": 0.874,
+      "grad_norm": 15.625,
+      "learning_rate": 8.313841357500313e-06,
+      "loss": 1.6467,
+      "step": 8740
+    },
+    {
+      "epoch": 0.875,
+      "grad_norm": 15.0625,
+      "learning_rate": 8.309797327946491e-06,
+      "loss": 1.8845,
+      "step": 8750
+    },
+    {
+      "epoch": 0.876,
+      "grad_norm": 17.75,
+      "learning_rate": 8.305749440811026e-06,
+      "loss": 1.5916,
+      "step": 8760
+    },
+    {
+      "epoch": 0.877,
+      "grad_norm": 15.375,
+      "learning_rate": 8.301697700811749e-06,
+      "loss": 2.095,
+      "step": 8770
+    },
+    {
+      "epoch": 0.878,
+      "grad_norm": 18.875,
+      "learning_rate": 8.297642112670976e-06,
+      "loss": 1.5417,
+      "step": 8780
+    },
+    {
+      "epoch": 0.879,
+      "grad_norm": 20.0,
+      "learning_rate": 8.293582681115512e-06,
+      "loss": 1.7099,
+      "step": 8790
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 17.0,
+      "learning_rate": 8.289519410876641e-06,
+      "loss": 1.9181,
+      "step": 8800
+    },
+    {
+      "epoch": 0.881,
+      "grad_norm": 10.8125,
+      "learning_rate": 8.285452306690118e-06,
+      "loss": 1.916,
+      "step": 8810
+    },
+    {
+      "epoch": 0.882,
+      "grad_norm": 16.875,
+      "learning_rate": 8.28138137329617e-06,
+      "loss": 1.7846,
+      "step": 8820
+    },
+    {
+      "epoch": 0.883,
+      "grad_norm": 12.625,
+      "learning_rate": 8.277306615439485e-06,
+      "loss": 1.5234,
+      "step": 8830
+    },
+    {
+      "epoch": 0.884,
+      "grad_norm": 11.9375,
+      "learning_rate": 8.27322803786921e-06,
+      "loss": 1.4728,
+      "step": 8840
+    },
+    {
+      "epoch": 0.885,
+      "grad_norm": 11.6875,
+      "learning_rate": 8.269145645338942e-06,
+      "loss": 1.5011,
+      "step": 8850
+    },
+    {
+      "epoch": 0.886,
+      "grad_norm": 13.875,
+      "learning_rate": 8.265059442606729e-06,
+      "loss": 1.5748,
+      "step": 8860
+    },
+    {
+      "epoch": 0.887,
+      "grad_norm": 16.875,
+      "learning_rate": 8.260969434435051e-06,
+      "loss": 1.505,
+      "step": 8870
+    },
+    {
+      "epoch": 0.888,
+      "grad_norm": 10.8125,
+      "learning_rate": 8.25687562559083e-06,
+      "loss": 1.5206,
+      "step": 8880
+    },
+    {
+      "epoch": 0.889,
+      "grad_norm": 11.9375,
+      "learning_rate": 8.252778020845418e-06,
+      "loss": 1.6573,
+      "step": 8890
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 14.1875,
+      "learning_rate": 8.248676624974587e-06,
+      "loss": 1.5645,
+      "step": 8900
+    },
+    {
+      "epoch": 0.891,
+      "grad_norm": 12.1875,
+      "learning_rate": 8.244571442758533e-06,
+      "loss": 1.4981,
+      "step": 8910
+    },
+    {
+      "epoch": 0.892,
+      "grad_norm": 11.4375,
+      "learning_rate": 8.24046247898186e-06,
+      "loss": 1.6817,
+      "step": 8920
+    },
+    {
+      "epoch": 0.893,
+      "grad_norm": 9.5,
+      "learning_rate": 8.236349738433578e-06,
+      "loss": 1.6215,
+      "step": 8930
+    },
+    {
+      "epoch": 0.894,
+      "grad_norm": 15.5,
+      "learning_rate": 8.232233225907105e-06,
+      "loss": 2.0036,
+      "step": 8940
+    },
+    {
+      "epoch": 0.895,
+      "grad_norm": 10.8125,
+      "learning_rate": 8.228112946200252e-06,
+      "loss": 1.5015,
+      "step": 8950
+    },
+    {
+      "epoch": 0.896,
+      "grad_norm": 14.5625,
+      "learning_rate": 8.223988904115223e-06,
+      "loss": 1.5849,
+      "step": 8960
+    },
+    {
+      "epoch": 0.897,
+      "grad_norm": 16.75,
+      "learning_rate": 8.219861104458602e-06,
+      "loss": 1.6913,
+      "step": 8970
+    },
+    {
+      "epoch": 0.898,
+      "grad_norm": 15.9375,
+      "learning_rate": 8.215729552041358e-06,
+      "loss": 1.8015,
+      "step": 8980
+    },
+    {
+      "epoch": 0.899,
+      "grad_norm": 11.6875,
+      "learning_rate": 8.211594251678832e-06,
+      "loss": 1.9554,
+      "step": 8990
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 19.875,
+      "learning_rate": 8.207455208190728e-06,
+      "loss": 1.5851,
+      "step": 9000
+    },
+    {
+      "epoch": 0.901,
+      "grad_norm": 13.625,
+      "learning_rate": 8.203312426401122e-06,
+      "loss": 1.8281,
+      "step": 9010
+    },
+    {
+      "epoch": 0.902,
+      "grad_norm": 13.25,
+      "learning_rate": 8.199165911138439e-06,
+      "loss": 1.658,
+      "step": 9020
+    },
+    {
+      "epoch": 0.903,
+      "grad_norm": 14.875,
+      "learning_rate": 8.19501566723546e-06,
+      "loss": 1.6515,
+      "step": 9030
+    },
+    {
+      "epoch": 0.904,
+      "grad_norm": 12.6875,
+      "learning_rate": 8.190861699529312e-06,
+      "loss": 1.6133,
+      "step": 9040
+    },
+    {
+      "epoch": 0.905,
+      "grad_norm": 17.25,
+      "learning_rate": 8.186704012861458e-06,
+      "loss": 1.6303,
+      "step": 9050
+    },
+    {
+      "epoch": 0.906,
+      "grad_norm": 13.1875,
+      "learning_rate": 8.1825426120777e-06,
+      "loss": 1.712,
+      "step": 9060
+    },
+    {
+      "epoch": 0.907,
+      "grad_norm": 18.625,
+      "learning_rate": 8.178377502028166e-06,
+      "loss": 1.562,
+      "step": 9070
+    },
+    {
+      "epoch": 0.908,
+      "grad_norm": 10.1875,
+      "learning_rate": 8.174208687567305e-06,
+      "loss": 1.4945,
+      "step": 9080
+    },
+    {
+      "epoch": 0.909,
+      "grad_norm": 10.1875,
+      "learning_rate": 8.170036173553891e-06,
+      "loss": 1.9697,
+      "step": 9090
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 12.4375,
+      "learning_rate": 8.165859964851002e-06,
+      "loss": 1.5412,
+      "step": 9100
+    },
+    {
+      "epoch": 0.911,
+      "grad_norm": 8.125,
+      "learning_rate": 8.16168006632603e-06,
+      "loss": 1.3788,
+      "step": 9110
+    },
+    {
+      "epoch": 0.912,
+      "grad_norm": 16.0,
+      "learning_rate": 8.157496482850657e-06,
+      "loss": 1.544,
+      "step": 9120
+    },
+    {
+      "epoch": 0.913,
+      "grad_norm": 20.0,
+      "learning_rate": 8.153309219300869e-06,
+      "loss": 1.8253,
+      "step": 9130
+    },
+    {
+      "epoch": 0.914,
+      "grad_norm": 10.6875,
+      "learning_rate": 8.149118280556938e-06,
+      "loss": 1.7203,
+      "step": 9140
+    },
+    {
+      "epoch": 0.915,
+      "grad_norm": 7.96875,
+      "learning_rate": 8.14492367150342e-06,
+      "loss": 1.4856,
+      "step": 9150
+    },
+    {
+      "epoch": 0.916,
+      "grad_norm": 11.75,
+      "learning_rate": 8.140725397029145e-06,
+      "loss": 1.6438,
+      "step": 9160
+    },
+    {
+      "epoch": 0.917,
+      "grad_norm": 13.0625,
+      "learning_rate": 8.136523462027222e-06,
+      "loss": 1.5484,
+      "step": 9170
+    },
+    {
+      "epoch": 0.918,
+      "grad_norm": 12.125,
+      "learning_rate": 8.13231787139502e-06,
+      "loss": 1.8676,
+      "step": 9180
+    },
+    {
+      "epoch": 0.919,
+      "grad_norm": 19.5,
+      "learning_rate": 8.128108630034173e-06,
+      "loss": 1.7753,
+      "step": 9190
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 16.375,
+      "learning_rate": 8.123895742850566e-06,
+      "loss": 1.6419,
+      "step": 9200
+    },
+    {
+      "epoch": 0.921,
+      "grad_norm": 20.25,
+      "learning_rate": 8.119679214754337e-06,
+      "loss": 1.5739,
+      "step": 9210
+    },
+    {
+      "epoch": 0.922,
+      "grad_norm": 15.4375,
+      "learning_rate": 8.115459050659867e-06,
+      "loss": 1.4622,
+      "step": 9220
+    },
+    {
+      "epoch": 0.923,
+      "grad_norm": 11.25,
+      "learning_rate": 8.11123525548577e-06,
+      "loss": 1.4869,
+      "step": 9230
+    },
+    {
+      "epoch": 0.924,
+      "grad_norm": 12.5,
+      "learning_rate": 8.107007834154898e-06,
+      "loss": 1.6214,
+      "step": 9240
+    },
+    {
+      "epoch": 0.925,
+      "grad_norm": 12.875,
+      "learning_rate": 8.102776791594327e-06,
+      "loss": 1.7234,
+      "step": 9250
+    },
+    {
+      "epoch": 0.926,
+      "grad_norm": 10.8125,
+      "learning_rate": 8.098542132735354e-06,
+      "loss": 1.4459,
+      "step": 9260
+    },
+    {
+      "epoch": 0.927,
+      "grad_norm": 13.5625,
+      "learning_rate": 8.094303862513489e-06,
+      "loss": 1.3766,
+      "step": 9270
+    },
+    {
+      "epoch": 0.928,
+      "grad_norm": 13.4375,
+      "learning_rate": 8.090061985868453e-06,
+      "loss": 1.4917,
+      "step": 9280
+    },
+    {
+      "epoch": 0.929,
+      "grad_norm": 7.375,
+      "learning_rate": 8.085816507744171e-06,
+      "loss": 1.7351,
+      "step": 9290
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 14.125,
+      "learning_rate": 8.081567433088763e-06,
+      "loss": 1.7199,
+      "step": 9300
+    },
+    {
+      "epoch": 0.931,
+      "grad_norm": 12.875,
+      "learning_rate": 8.077314766854541e-06,
+      "loss": 1.6191,
+      "step": 9310
+    },
+    {
+      "epoch": 0.932,
+      "grad_norm": 11.0,
+      "learning_rate": 8.073058513998004e-06,
+      "loss": 1.5704,
+      "step": 9320
+    },
+    {
+      "epoch": 0.933,
+      "grad_norm": 15.4375,
+      "learning_rate": 8.068798679479835e-06,
+      "loss": 1.7123,
+      "step": 9330
+    },
+    {
+      "epoch": 0.934,
+      "grad_norm": 17.375,
+      "learning_rate": 8.064535268264883e-06,
+      "loss": 1.9369,
+      "step": 9340
+    },
+    {
+      "epoch": 0.935,
+      "grad_norm": 13.875,
+      "learning_rate": 8.060268285322173e-06,
+      "loss": 1.5713,
+      "step": 9350
+    },
+    {
+      "epoch": 0.936,
+      "grad_norm": 12.125,
+      "learning_rate": 8.05599773562489e-06,
+      "loss": 1.4607,
+      "step": 9360
+    },
+    {
+      "epoch": 0.937,
+      "grad_norm": 12.6875,
+      "learning_rate": 8.051723624150377e-06,
+      "loss": 1.7941,
+      "step": 9370
+    },
+    {
+      "epoch": 0.938,
+      "grad_norm": 15.125,
+      "learning_rate": 8.047445955880123e-06,
+      "loss": 1.5987,
+      "step": 9380
+    },
+    {
+      "epoch": 0.939,
+      "grad_norm": 10.875,
+      "learning_rate": 8.043164735799772e-06,
+      "loss": 1.7803,
+      "step": 9390
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 18.75,
+      "learning_rate": 8.0388799688991e-06,
+      "loss": 1.7874,
+      "step": 9400
+    },
+    {
+      "epoch": 0.941,
+      "grad_norm": 12.125,
+      "learning_rate": 8.03459166017202e-06,
+      "loss": 1.7642,
+      "step": 9410
+    },
+    {
+      "epoch": 0.942,
+      "grad_norm": 22.625,
+      "learning_rate": 8.030299814616575e-06,
+      "loss": 1.5965,
+      "step": 9420
+    },
+    {
+      "epoch": 0.943,
+      "grad_norm": 11.75,
+      "learning_rate": 8.026004437234924e-06,
+      "loss": 1.8644,
+      "step": 9430
+    },
+    {
+      "epoch": 0.944,
+      "grad_norm": 14.8125,
+      "learning_rate": 8.021705533033348e-06,
+      "loss": 1.6477,
+      "step": 9440
+    },
+    {
+      "epoch": 0.945,
+      "grad_norm": 12.25,
+      "learning_rate": 8.017403107022237e-06,
+      "loss": 1.6619,
+      "step": 9450
+    },
+    {
+      "epoch": 0.946,
+      "grad_norm": 16.25,
+      "learning_rate": 8.013097164216083e-06,
+      "loss": 1.9495,
+      "step": 9460
+    },
+    {
+      "epoch": 0.947,
+      "grad_norm": 11.9375,
+      "learning_rate": 8.008787709633481e-06,
+      "loss": 1.9536,
+      "step": 9470
+    },
+    {
+      "epoch": 0.948,
+      "grad_norm": 15.25,
+      "learning_rate": 8.004474748297117e-06,
+      "loss": 1.571,
+      "step": 9480
+    },
+    {
+      "epoch": 0.949,
+      "grad_norm": 13.625,
+      "learning_rate": 8.000158285233763e-06,
+      "loss": 1.5525,
+      "step": 9490
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 9.125,
+      "learning_rate": 7.995838325474273e-06,
+      "loss": 1.5748,
+      "step": 9500
+    },
+    {
+      "epoch": 0.951,
+      "grad_norm": 11.4375,
+      "learning_rate": 7.991514874053579e-06,
+      "loss": 1.8771,
+      "step": 9510
+    },
+    {
+      "epoch": 0.952,
+      "grad_norm": 24.0,
+      "learning_rate": 7.98718793601068e-06,
+      "loss": 1.9602,
+      "step": 9520
+    },
+    {
+      "epoch": 0.953,
+      "grad_norm": 11.625,
+      "learning_rate": 7.982857516388637e-06,
+      "loss": 1.722,
+      "step": 9530
+    },
+    {
+      "epoch": 0.954,
+      "grad_norm": 20.375,
+      "learning_rate": 7.978523620234574e-06,
+      "loss": 1.8724,
+      "step": 9540
+    },
+    {
+      "epoch": 0.955,
+      "grad_norm": 16.125,
+      "learning_rate": 7.974186252599662e-06,
+      "loss": 1.5147,
+      "step": 9550
+    },
+    {
+      "epoch": 0.956,
+      "grad_norm": 24.25,
+      "learning_rate": 7.969845418539118e-06,
+      "loss": 1.7523,
+      "step": 9560
+    },
+    {
+      "epoch": 0.957,
+      "grad_norm": 14.1875,
+      "learning_rate": 7.965501123112205e-06,
+      "loss": 1.5979,
+      "step": 9570
+    },
+    {
+      "epoch": 0.958,
+      "grad_norm": 10.8125,
+      "learning_rate": 7.961153371382213e-06,
+      "loss": 1.7139,
+      "step": 9580
+    },
+    {
+      "epoch": 0.959,
+      "grad_norm": 11.875,
+      "learning_rate": 7.956802168416466e-06,
+      "loss": 1.7225,
+      "step": 9590
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 9.5,
+      "learning_rate": 7.952447519286305e-06,
+      "loss": 1.4406,
+      "step": 9600
+    },
+    {
+      "epoch": 0.961,
+      "grad_norm": 21.75,
+      "learning_rate": 7.948089429067093e-06,
+      "loss": 1.5884,
+      "step": 9610
+    },
+    {
+      "epoch": 0.962,
+      "grad_norm": 12.5,
+      "learning_rate": 7.9437279028382e-06,
+      "loss": 1.5404,
+      "step": 9620
+    },
+    {
+      "epoch": 0.963,
+      "grad_norm": 15.5625,
+      "learning_rate": 7.939362945683005e-06,
+      "loss": 1.5367,
+      "step": 9630
+    },
+    {
+      "epoch": 0.964,
+      "grad_norm": 14.375,
+      "learning_rate": 7.934994562688876e-06,
+      "loss": 1.6465,
+      "step": 9640
+    },
+    {
+      "epoch": 0.965,
+      "grad_norm": 16.375,
+      "learning_rate": 7.930622758947184e-06,
+      "loss": 1.758,
+      "step": 9650
+    },
+    {
+      "epoch": 0.966,
+      "grad_norm": 9.3125,
+      "learning_rate": 7.926247539553285e-06,
+      "loss": 1.7495,
+      "step": 9660
+    },
+    {
+      "epoch": 0.967,
+      "grad_norm": 13.4375,
+      "learning_rate": 7.921868909606509e-06,
+      "loss": 1.6867,
+      "step": 9670
+    },
+    {
+      "epoch": 0.968,
+      "grad_norm": 21.0,
+      "learning_rate": 7.917486874210172e-06,
+      "loss": 1.6605,
+      "step": 9680
+    },
+    {
+      "epoch": 0.969,
+      "grad_norm": 9.875,
+      "learning_rate": 7.91310143847155e-06,
+      "loss": 1.9101,
+      "step": 9690
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 10.1875,
+      "learning_rate": 7.908712607501886e-06,
+      "loss": 1.3646,
+      "step": 9700
+    },
+    {
+      "epoch": 0.971,
+      "grad_norm": 11.5,
+      "learning_rate": 7.90432038641638e-06,
+      "loss": 1.5109,
+      "step": 9710
+    },
+    {
+      "epoch": 0.972,
+      "grad_norm": 21.375,
+      "learning_rate": 7.899924780334183e-06,
+      "loss": 2.0526,
+      "step": 9720
+    },
+    {
+      "epoch": 0.973,
+      "grad_norm": 21.875,
+      "learning_rate": 7.895525794378388e-06,
+      "loss": 1.9549,
+      "step": 9730
+    },
+    {
+      "epoch": 0.974,
+      "grad_norm": 16.125,
+      "learning_rate": 7.891123433676035e-06,
+      "loss": 1.568,
+      "step": 9740
+    },
+    {
+      "epoch": 0.975,
+      "grad_norm": 17.5,
+      "learning_rate": 7.88671770335809e-06,
+      "loss": 2.0297,
+      "step": 9750
+    },
+    {
+      "epoch": 0.976,
+      "grad_norm": 10.875,
+      "learning_rate": 7.88230860855945e-06,
+      "loss": 1.6811,
+      "step": 9760
+    },
+    {
+      "epoch": 0.977,
+      "grad_norm": 19.125,
+      "learning_rate": 7.87789615441893e-06,
+      "loss": 1.8511,
+      "step": 9770
+    },
+    {
+      "epoch": 0.978,
+      "grad_norm": 17.625,
+      "learning_rate": 7.87348034607926e-06,
+      "loss": 1.9458,
+      "step": 9780
+    },
+    {
+      "epoch": 0.979,
+      "grad_norm": 11.5625,
+      "learning_rate": 7.869061188687091e-06,
+      "loss": 1.7619,
+      "step": 9790
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 17.0,
+      "learning_rate": 7.864638687392958e-06,
+      "loss": 1.5866,
+      "step": 9800
+    },
+    {
+      "epoch": 0.981,
+      "grad_norm": 13.75,
+      "learning_rate": 7.860212847351309e-06,
+      "loss": 1.4831,
+      "step": 9810
+    },
+    {
+      "epoch": 0.982,
+      "grad_norm": 16.75,
+      "learning_rate": 7.855783673720478e-06,
+      "loss": 1.5121,
+      "step": 9820
+    },
+    {
+      "epoch": 0.983,
+      "grad_norm": 12.5,
+      "learning_rate": 7.85135117166268e-06,
+      "loss": 1.4691,
+      "step": 9830
+    },
+    {
+      "epoch": 0.984,
+      "grad_norm": 13.0625,
+      "learning_rate": 7.846915346344019e-06,
+      "loss": 2.0694,
+      "step": 9840
+    },
+    {
+      "epoch": 0.985,
+      "grad_norm": 12.6875,
+      "learning_rate": 7.842476202934462e-06,
+      "loss": 1.79,
+      "step": 9850
+    },
+    {
+      "epoch": 0.986,
+      "grad_norm": 17.375,
+      "learning_rate": 7.838033746607849e-06,
+      "loss": 1.807,
+      "step": 9860
+    },
+    {
+      "epoch": 0.987,
+      "grad_norm": 21.25,
+      "learning_rate": 7.83358798254188e-06,
+      "loss": 4.0416,
+      "step": 9870
+    },
+    {
+      "epoch": 0.988,
+      "grad_norm": 13.8125,
+      "learning_rate": 7.829138915918113e-06,
+      "loss": 1.4448,
+      "step": 9880
+    },
+    {
+      "epoch": 0.989,
+      "grad_norm": 14.625,
+      "learning_rate": 7.82468655192195e-06,
+      "loss": 1.8409,
+      "step": 9890
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 13.0,
+      "learning_rate": 7.820230895742638e-06,
+      "loss": 1.6878,
+      "step": 9900
+    },
+    {
+      "epoch": 0.991,
+      "grad_norm": 13.8125,
+      "learning_rate": 7.81577195257326e-06,
+      "loss": 1.5651,
+      "step": 9910
+    },
+    {
+      "epoch": 0.992,
+      "grad_norm": 12.5625,
+      "learning_rate": 7.811309727610738e-06,
+      "loss": 1.6544,
+      "step": 9920
+    },
+    {
+      "epoch": 0.993,
+      "grad_norm": 9.0,
+      "learning_rate": 7.806844226055807e-06,
+      "loss": 1.5209,
+      "step": 9930
+    },
+    {
+      "epoch": 0.994,
+      "grad_norm": 12.0,
+      "learning_rate": 7.802375453113028e-06,
+      "loss": 1.3959,
+      "step": 9940
+    },
+    {
+      "epoch": 0.995,
+      "grad_norm": 15.0,
+      "learning_rate": 7.797903413990773e-06,
+      "loss": 1.6071,
+      "step": 9950
+    },
+    {
+      "epoch": 0.996,
+      "grad_norm": 9.9375,
+      "learning_rate": 7.793428113901222e-06,
+      "loss": 1.9001,
+      "step": 9960
+    },
+    {
+      "epoch": 0.997,
+      "grad_norm": 10.8125,
+      "learning_rate": 7.788949558060354e-06,
+      "loss": 1.5101,
+      "step": 9970
+    },
+    {
+      "epoch": 0.998,
+      "grad_norm": 11.75,
+      "learning_rate": 7.784467751687942e-06,
+      "loss": 1.6358,
+      "step": 9980
+    },
+    {
+      "epoch": 0.999,
+      "grad_norm": 9.75,
+      "learning_rate": 7.77998270000755e-06,
+      "loss": 1.5159,
+      "step": 9990
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 12.625,
+      "learning_rate": 7.775494408246522e-06,
+      "loss": 1.7177,
+      "step": 10000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}