diff --git "a/checkpoint-2000/trainer_state.json" "b/checkpoint-2000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-2000/trainer_state.json"
@@ -0,0 +1,14033 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.02,
+  "eval_steps": 500,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 1e-05,
+      "grad_norm": 1.085229352812366,
+      "learning_rate": 3e-06,
+      "loss": 10.849,
+      "step": 1
+    },
+    {
+      "epoch": 2e-05,
+      "grad_norm": 1.0764689186661929,
+      "learning_rate": 6e-06,
+      "loss": 10.8489,
+      "step": 2
+    },
+    {
+      "epoch": 3e-05,
+      "grad_norm": 1.0926036068515363,
+      "learning_rate": 9e-06,
+      "loss": 10.8486,
+      "step": 3
+    },
+    {
+      "epoch": 4e-05,
+      "grad_norm": 1.0859011783792423,
+      "learning_rate": 1.2e-05,
+      "loss": 10.848,
+      "step": 4
+    },
+    {
+      "epoch": 5e-05,
+      "grad_norm": 1.0906873388641662,
+      "learning_rate": 1.5e-05,
+      "loss": 10.8453,
+      "step": 5
+    },
+    {
+      "epoch": 6e-05,
+      "grad_norm": 1.0895888734627917,
+      "learning_rate": 1.8e-05,
+      "loss": 10.8447,
+      "step": 6
+    },
+    {
+      "epoch": 7e-05,
+      "grad_norm": 1.0913329404483254,
+      "learning_rate": 2.1000000000000002e-05,
+      "loss": 10.8355,
+      "step": 7
+    },
+    {
+      "epoch": 8e-05,
+      "grad_norm": 1.0766237663279077,
+      "learning_rate": 2.4e-05,
+      "loss": 10.8141,
+      "step": 8
+    },
+    {
+      "epoch": 9e-05,
+      "grad_norm": 1.0617425338278697,
+      "learning_rate": 2.7e-05,
+      "loss": 10.8099,
+      "step": 9
+    },
+    {
+      "epoch": 0.0001,
+      "grad_norm": 1.080271312928173,
+      "learning_rate": 3e-05,
+      "loss": 10.7983,
+      "step": 10
+    },
+    {
+      "epoch": 0.00011,
+      "grad_norm": 1.0663119503184246,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 10.7844,
+      "step": 11
+    },
+    {
+      "epoch": 0.00012,
+      "grad_norm": 1.0694505502575085,
+      "learning_rate": 3.6e-05,
+      "loss": 10.7739,
+      "step": 12
+    },
+    {
+      "epoch": 0.00013,
+      "grad_norm": 1.0509206118753271,
+      "learning_rate": 3.9e-05,
+      "loss": 10.7554,
+      "step": 13
+    },
+    {
+      "epoch": 0.00014,
+      "grad_norm": 1.0427932624919014,
+      "learning_rate": 4.2000000000000004e-05,
+      "loss": 10.7452,
+      "step": 14
+    },
+    {
+      "epoch": 0.00015,
+      "grad_norm": 1.0323600599358198,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 10.736,
+      "step": 15
+    },
+    {
+      "epoch": 0.00016,
+      "grad_norm": 1.0039727604034705,
+      "learning_rate": 4.8e-05,
+      "loss": 10.7191,
+      "step": 16
+    },
+    {
+      "epoch": 0.00017,
+      "grad_norm": 0.9736033767894897,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 10.7049,
+      "step": 17
+    },
+    {
+      "epoch": 0.00018,
+      "grad_norm": 0.9665971933165531,
+      "learning_rate": 5.4e-05,
+      "loss": 10.69,
+      "step": 18
+    },
+    {
+      "epoch": 0.00019,
+      "grad_norm": 0.9428089164290921,
+      "learning_rate": 5.7e-05,
+      "loss": 10.6767,
+      "step": 19
+    },
+    {
+      "epoch": 0.0002,
+      "grad_norm": 0.943192779381936,
+      "learning_rate": 6e-05,
+      "loss": 10.666,
+      "step": 20
+    },
+    {
+      "epoch": 0.00021,
+      "grad_norm": 0.9246956411226879,
+      "learning_rate": 6.3e-05,
+      "loss": 10.6525,
+      "step": 21
+    },
+    {
+      "epoch": 0.00022,
+      "grad_norm": 0.9206776841693647,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 10.6378,
+      "step": 22
+    },
+    {
+      "epoch": 0.00023,
+      "grad_norm": 0.9152142474292139,
+      "learning_rate": 6.9e-05,
+      "loss": 10.626,
+      "step": 23
+    },
+    {
+      "epoch": 0.00024,
+      "grad_norm": 0.9101286615991204,
+      "learning_rate": 7.2e-05,
+      "loss": 10.613,
+      "step": 24
+    },
+    {
+      "epoch": 0.00025,
+      "grad_norm": 0.9100685942512898,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 10.5998,
+      "step": 25
+    },
+    {
+      "epoch": 0.00026,
+      "grad_norm": 0.9093157110032968,
+      "learning_rate": 7.8e-05,
+      "loss": 10.5859,
+      "step": 26
+    },
+    {
+      "epoch": 0.00027,
+      "grad_norm": 0.9036082412984342,
+      "learning_rate": 8.1e-05,
+      "loss": 10.5742,
+      "step": 27
+    },
+    {
+      "epoch": 0.00028,
+      "grad_norm": 0.9033464449252919,
+      "learning_rate": 8.400000000000001e-05,
+      "loss": 10.5612,
+      "step": 28
+    },
+    {
+      "epoch": 0.00029,
+      "grad_norm": 0.9067163029768021,
+      "learning_rate": 8.7e-05,
+      "loss": 10.5465,
+      "step": 29
+    },
+    {
+      "epoch": 0.0003,
+      "grad_norm": 0.9152916501496612,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 10.5321,
+      "step": 30
+    },
+    {
+      "epoch": 0.00031,
+      "grad_norm": 0.9122391999445313,
+      "learning_rate": 9.3e-05,
+      "loss": 10.5167,
+      "step": 31
+    },
+    {
+      "epoch": 0.00032,
+      "grad_norm": 0.9090668181794543,
+      "learning_rate": 9.6e-05,
+      "loss": 10.5034,
+      "step": 32
+    },
+    {
+      "epoch": 0.00033,
+      "grad_norm": 0.9095629132760054,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 10.4865,
+      "step": 33
+    },
+    {
+      "epoch": 0.00034,
+      "grad_norm": 0.9107692784496799,
+      "learning_rate": 0.00010200000000000001,
+      "loss": 10.4706,
+      "step": 34
+    },
+    {
+      "epoch": 0.00035,
+      "grad_norm": 0.9162740070354684,
+      "learning_rate": 0.00010500000000000002,
+      "loss": 10.4535,
+      "step": 35
+    },
+    {
+      "epoch": 0.00036,
+      "grad_norm": 0.9114324016484531,
+      "learning_rate": 0.000108,
+      "loss": 10.4367,
+      "step": 36
+    },
+    {
+      "epoch": 0.00037,
+      "grad_norm": 0.901331547883087,
+      "learning_rate": 0.000111,
+      "loss": 10.4196,
+      "step": 37
+    },
+    {
+      "epoch": 0.00038,
+      "grad_norm": 0.9056352937834914,
+      "learning_rate": 0.000114,
+      "loss": 10.4003,
+      "step": 38
+    },
+    {
+      "epoch": 0.00039,
+      "grad_norm": 0.9071824304759276,
+      "learning_rate": 0.000117,
+      "loss": 10.3815,
+      "step": 39
+    },
+    {
+      "epoch": 0.0004,
+      "grad_norm": 0.911150856035614,
+      "learning_rate": 0.00012,
+      "loss": 10.3599,
+      "step": 40
+    },
+    {
+      "epoch": 0.00041,
+      "grad_norm": 0.9001357892993758,
+      "learning_rate": 0.000123,
+      "loss": 10.3411,
+      "step": 41
+    },
+    {
+      "epoch": 0.00042,
+      "grad_norm": 0.9057215347961051,
+      "learning_rate": 0.000126,
+      "loss": 10.3208,
+      "step": 42
+    },
+    {
+      "epoch": 0.00043,
+      "grad_norm": 0.9101872556211666,
+      "learning_rate": 0.000129,
+      "loss": 10.2978,
+      "step": 43
+    },
+    {
+      "epoch": 0.00044,
+      "grad_norm": 0.9097345366629672,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 10.2768,
+      "step": 44
+    },
+    {
+      "epoch": 0.00045,
+      "grad_norm": 0.9111670268341607,
+      "learning_rate": 0.000135,
+      "loss": 10.2539,
+      "step": 45
+    },
+    {
+      "epoch": 0.00046,
+      "grad_norm": 0.9030473796809102,
+      "learning_rate": 0.000138,
+      "loss": 10.2322,
+      "step": 46
+    },
+    {
+      "epoch": 0.00047,
+      "grad_norm": 0.9086055117133346,
+      "learning_rate": 0.000141,
+      "loss": 10.2064,
+      "step": 47
+    },
+    {
+      "epoch": 0.00048,
+      "grad_norm": 0.9060414917101882,
+      "learning_rate": 0.000144,
+      "loss": 10.1837,
+      "step": 48
+    },
+    {
+      "epoch": 0.00049,
+      "grad_norm": 0.9059366436676172,
+      "learning_rate": 0.000147,
+      "loss": 10.1599,
+      "step": 49
+    },
+    {
+      "epoch": 0.0005,
+      "grad_norm": 0.9106165244124662,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 10.1343,
+      "step": 50
+    },
+    {
+      "epoch": 0.00051,
+      "grad_norm": 0.9110992341155927,
+      "learning_rate": 0.000153,
+      "loss": 10.1078,
+      "step": 51
+    },
+    {
+      "epoch": 0.00052,
+      "grad_norm": 0.9077294523989683,
+      "learning_rate": 0.000156,
+      "loss": 10.0815,
+      "step": 52
+    },
+    {
+      "epoch": 0.00053,
+      "grad_norm": 0.8995624264210066,
+      "learning_rate": 0.000159,
+      "loss": 10.0581,
+      "step": 53
+    },
+    {
+      "epoch": 0.00054,
+      "grad_norm": 0.9187536344258231,
+      "learning_rate": 0.000162,
+      "loss": 10.026,
+      "step": 54
+    },
+    {
+      "epoch": 0.00055,
+      "grad_norm": 0.9102867236672618,
+      "learning_rate": 0.000165,
+      "loss": 10.0019,
+      "step": 55
+    },
+    {
+      "epoch": 0.00056,
+      "grad_norm": 0.9031170239719724,
+      "learning_rate": 0.00016800000000000002,
+      "loss": 9.9743,
+      "step": 56
+    },
+    {
+      "epoch": 0.00057,
+      "grad_norm": 0.9090239107255728,
+      "learning_rate": 0.000171,
+      "loss": 9.9467,
+      "step": 57
+    },
+    {
+      "epoch": 0.00058,
+      "grad_norm": 0.9070896749665766,
+      "learning_rate": 0.000174,
+      "loss": 9.9223,
+      "step": 58
+    },
+    {
+      "epoch": 0.00059,
+      "grad_norm": 0.9116582783399498,
+      "learning_rate": 0.000177,
+      "loss": 9.8905,
+      "step": 59
+    },
+    {
+      "epoch": 0.0006,
+      "grad_norm": 0.899071209460366,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 9.8642,
+      "step": 60
+    },
+    {
+      "epoch": 0.00061,
+      "grad_norm": 0.8984032767613607,
+      "learning_rate": 0.000183,
+      "loss": 9.8367,
+      "step": 61
+    },
+    {
+      "epoch": 0.00062,
+      "grad_norm": 0.9043779152804675,
+      "learning_rate": 0.000186,
+      "loss": 9.8058,
+      "step": 62
+    },
+    {
+      "epoch": 0.00063,
+      "grad_norm": 0.8969230453763369,
+      "learning_rate": 0.000189,
+      "loss": 9.7784,
+      "step": 63
+    },
+    {
+      "epoch": 0.00064,
+      "grad_norm": 0.8929896633083918,
+      "learning_rate": 0.000192,
+      "loss": 9.7485,
+      "step": 64
+    },
+    {
+      "epoch": 0.00065,
+      "grad_norm": 0.9016512171852502,
+      "learning_rate": 0.00019500000000000002,
+      "loss": 9.7165,
+      "step": 65
+    },
+    {
+      "epoch": 0.00066,
+      "grad_norm": 0.8946267084059,
+      "learning_rate": 0.00019800000000000002,
+      "loss": 9.6927,
+      "step": 66
+    },
+    {
+      "epoch": 0.00067,
+      "grad_norm": 0.8963872265737496,
+      "learning_rate": 0.000201,
+      "loss": 9.6633,
+      "step": 67
+    },
+    {
+      "epoch": 0.00068,
+      "grad_norm": 0.9035635569767697,
+      "learning_rate": 0.00020400000000000003,
+      "loss": 9.6313,
+      "step": 68
+    },
+    {
+      "epoch": 0.00069,
+      "grad_norm": 0.8884430485081615,
+      "learning_rate": 0.00020700000000000002,
+      "loss": 9.604,
+      "step": 69
+    },
+    {
+      "epoch": 0.0007,
+      "grad_norm": 0.901710386427562,
+      "learning_rate": 0.00021000000000000004,
+      "loss": 9.5675,
+      "step": 70
+    },
+    {
+      "epoch": 0.00071,
+      "grad_norm": 0.8913815692585527,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 9.5403,
+      "step": 71
+    },
+    {
+      "epoch": 0.00072,
+      "grad_norm": 0.8947982003050186,
+      "learning_rate": 0.000216,
+      "loss": 9.5138,
+      "step": 72
+    },
+    {
+      "epoch": 0.00073,
+      "grad_norm": 0.8936046055705469,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 9.4802,
+      "step": 73
+    },
+    {
+      "epoch": 0.00074,
+      "grad_norm": 0.894403551075387,
+      "learning_rate": 0.000222,
+      "loss": 9.4539,
+      "step": 74
+    },
+    {
+      "epoch": 0.00075,
+      "grad_norm": 0.8897139679621429,
+      "learning_rate": 0.000225,
+      "loss": 9.4187,
+      "step": 75
+    },
+    {
+      "epoch": 0.00076,
+      "grad_norm": 0.89755782108852,
+      "learning_rate": 0.000228,
+      "loss": 9.3929,
+      "step": 76
+    },
+    {
+      "epoch": 0.00077,
+      "grad_norm": 0.8917412779828411,
+      "learning_rate": 0.000231,
+      "loss": 9.3649,
+      "step": 77
+    },
+    {
+      "epoch": 0.00078,
+      "grad_norm": 0.886930008239094,
+      "learning_rate": 0.000234,
+      "loss": 9.3351,
+      "step": 78
+    },
+    {
+      "epoch": 0.00079,
+      "grad_norm": 0.8893115023720741,
+      "learning_rate": 0.00023700000000000001,
+      "loss": 9.2992,
+      "step": 79
+    },
+    {
+      "epoch": 0.0008,
+      "grad_norm": 0.8917772276535281,
+      "learning_rate": 0.00024,
+      "loss": 9.2632,
+      "step": 80
+    },
+    {
+      "epoch": 0.00081,
+      "grad_norm": 0.8971231461853929,
+      "learning_rate": 0.00024300000000000002,
+      "loss": 9.2316,
+      "step": 81
+    },
+    {
+      "epoch": 0.00082,
+      "grad_norm": 0.9071155901294882,
+      "learning_rate": 0.000246,
+      "loss": 9.206,
+      "step": 82
+    },
+    {
+      "epoch": 0.00083,
+      "grad_norm": 0.8963243651316662,
+      "learning_rate": 0.00024900000000000004,
+      "loss": 9.1708,
+      "step": 83
+    },
+    {
+      "epoch": 0.00084,
+      "grad_norm": 0.9002519637093493,
+      "learning_rate": 0.000252,
+      "loss": 9.1395,
+      "step": 84
+    },
+    {
+      "epoch": 0.00085,
+      "grad_norm": 0.8962870771351267,
+      "learning_rate": 0.000255,
+      "loss": 9.1159,
+      "step": 85
+    },
+    {
+      "epoch": 0.00086,
+      "grad_norm": 0.8973951423301171,
+      "learning_rate": 0.000258,
+      "loss": 9.0799,
+      "step": 86
+    },
+    {
+      "epoch": 0.00087,
+      "grad_norm": 0.8894399774297843,
+      "learning_rate": 0.000261,
+      "loss": 9.0571,
+      "step": 87
+    },
+    {
+      "epoch": 0.00088,
+      "grad_norm": 0.8954974565265822,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 9.0211,
+      "step": 88
+    },
+    {
+      "epoch": 0.00089,
+      "grad_norm": 0.8903887799592864,
+      "learning_rate": 0.000267,
+      "loss": 8.9943,
+      "step": 89
+    },
+    {
+      "epoch": 0.0009,
+      "grad_norm": 0.8896831425922581,
+      "learning_rate": 0.00027,
+      "loss": 8.9666,
+      "step": 90
+    },
+    {
+      "epoch": 0.00091,
+      "grad_norm": 0.8841935030693385,
+      "learning_rate": 0.000273,
+      "loss": 8.9365,
+      "step": 91
+    },
+    {
+      "epoch": 0.00092,
+      "grad_norm": 0.8862593694142118,
+      "learning_rate": 0.000276,
+      "loss": 8.9052,
+      "step": 92
+    },
+    {
+      "epoch": 0.00093,
+      "grad_norm": 0.8881041222250594,
+      "learning_rate": 0.000279,
+      "loss": 8.8752,
+      "step": 93
+    },
+    {
+      "epoch": 0.00094,
+      "grad_norm": 0.8868931156100198,
+      "learning_rate": 0.000282,
+      "loss": 8.8494,
+      "step": 94
+    },
+    {
+      "epoch": 0.00095,
+      "grad_norm": 0.8826333164427848,
+      "learning_rate": 0.000285,
+      "loss": 8.8203,
+      "step": 95
+    },
+    {
+      "epoch": 0.00096,
+      "grad_norm": 0.8910955494857569,
+      "learning_rate": 0.000288,
+      "loss": 8.7905,
+      "step": 96
+    },
+    {
+      "epoch": 0.00097,
+      "grad_norm": 0.8809279978723125,
+      "learning_rate": 0.000291,
+      "loss": 8.7668,
+      "step": 97
+    },
+    {
+      "epoch": 0.00098,
+      "grad_norm": 0.8829202734417477,
+      "learning_rate": 0.000294,
+      "loss": 8.73,
+      "step": 98
+    },
+    {
+      "epoch": 0.00099,
+      "grad_norm": 0.8858154061322314,
+      "learning_rate": 0.000297,
+      "loss": 8.7072,
+      "step": 99
+    },
+    {
+      "epoch": 0.001,
+      "grad_norm": 0.8861735275161908,
+      "learning_rate": 0.00030000000000000003,
+      "loss": 8.6806,
+      "step": 100
+    },
+    {
+      "epoch": 0.00101,
+      "grad_norm": 0.8886468466177152,
+      "learning_rate": 0.00030300000000000005,
+      "loss": 8.6455,
+      "step": 101
+    },
+    {
+      "epoch": 0.00102,
+      "grad_norm": 0.8755547614914917,
+      "learning_rate": 0.000306,
+      "loss": 8.625,
+      "step": 102
+    },
+    {
+      "epoch": 0.00103,
+      "grad_norm": 0.879237266457317,
+      "learning_rate": 0.000309,
+      "loss": 8.6003,
+      "step": 103
+    },
+    {
+      "epoch": 0.00104,
+      "grad_norm": 0.8697854479956653,
+      "learning_rate": 0.000312,
+      "loss": 8.5786,
+      "step": 104
+    },
+    {
+      "epoch": 0.00105,
+      "grad_norm": 0.8733949043334459,
+      "learning_rate": 0.000315,
+      "loss": 8.552,
+      "step": 105
+    },
+    {
+      "epoch": 0.00106,
+      "grad_norm": 0.8706148334292045,
+      "learning_rate": 0.000318,
+      "loss": 8.5224,
+      "step": 106
+    },
+    {
+      "epoch": 0.00107,
+      "grad_norm": 0.8653683244965015,
+      "learning_rate": 0.000321,
+      "loss": 8.4981,
+      "step": 107
+    },
+    {
+      "epoch": 0.00108,
+      "grad_norm": 0.8737239863866451,
+      "learning_rate": 0.000324,
+      "loss": 8.4698,
+      "step": 108
+    },
+    {
+      "epoch": 0.00109,
+      "grad_norm": 0.8706512729216435,
+      "learning_rate": 0.000327,
+      "loss": 8.4501,
+      "step": 109
+    },
+    {
+      "epoch": 0.0011,
+      "grad_norm": 0.8842127152624679,
+      "learning_rate": 0.00033,
+      "loss": 8.4274,
+      "step": 110
+    },
+    {
+      "epoch": 0.00111,
+      "grad_norm": 0.9007754832304464,
+      "learning_rate": 0.000333,
+      "loss": 8.3985,
+      "step": 111
+    },
+    {
+      "epoch": 0.00112,
+      "grad_norm": 0.9159855921530741,
+      "learning_rate": 0.00033600000000000004,
+      "loss": 8.3784,
+      "step": 112
+    },
+    {
+      "epoch": 0.00113,
+      "grad_norm": 0.9063278036144603,
+      "learning_rate": 0.000339,
+      "loss": 8.3391,
+      "step": 113
+    },
+    {
+      "epoch": 0.00114,
+      "grad_norm": 0.8437820836704115,
+      "learning_rate": 0.000342,
+      "loss": 8.3286,
+      "step": 114
+    },
+    {
+      "epoch": 0.00115,
+      "grad_norm": 0.8612821674982505,
+      "learning_rate": 0.00034500000000000004,
+      "loss": 8.3022,
+      "step": 115
+    },
+    {
+      "epoch": 0.00116,
+      "grad_norm": 0.8581012057508914,
+      "learning_rate": 0.000348,
+      "loss": 8.2786,
+      "step": 116
+    },
+    {
+      "epoch": 0.00117,
+      "grad_norm": 0.834681162463853,
+      "learning_rate": 0.000351,
+      "loss": 8.2472,
+      "step": 117
+    },
+    {
+      "epoch": 0.00118,
+      "grad_norm": 0.8526508674143746,
+      "learning_rate": 0.000354,
+      "loss": 8.2312,
+      "step": 118
+    },
+    {
+      "epoch": 0.00119,
+      "grad_norm": 0.8471842679679056,
+      "learning_rate": 0.000357,
+      "loss": 8.2186,
+      "step": 119
+    },
+    {
+      "epoch": 0.0012,
+      "grad_norm": 0.8238029079166322,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 8.1917,
+      "step": 120
+    },
+    {
+      "epoch": 0.00121,
+      "grad_norm": 0.832770075662114,
+      "learning_rate": 0.000363,
+      "loss": 8.1641,
+      "step": 121
+    },
+    {
+      "epoch": 0.00122,
+      "grad_norm": 0.8176689288160716,
+      "learning_rate": 0.000366,
+      "loss": 8.1471,
+      "step": 122
+    },
+    {
+      "epoch": 0.00123,
+      "grad_norm": 0.8121975389696077,
+      "learning_rate": 0.000369,
+      "loss": 8.1243,
+      "step": 123
+    },
+    {
+      "epoch": 0.00124,
+      "grad_norm": 0.8287555396618358,
+      "learning_rate": 0.000372,
+      "loss": 8.1037,
+      "step": 124
+    },
+    {
+      "epoch": 0.00125,
+      "grad_norm": 0.8387006564379252,
+      "learning_rate": 0.000375,
+      "loss": 8.0802,
+      "step": 125
+    },
+    {
+      "epoch": 0.00126,
+      "grad_norm": 0.7985855201767323,
+      "learning_rate": 0.000378,
+      "loss": 8.0672,
+      "step": 126
+    },
+    {
+      "epoch": 0.00127,
+      "grad_norm": 0.7830434817433392,
+      "learning_rate": 0.000381,
+      "loss": 8.0437,
+      "step": 127
+    },
+    {
+      "epoch": 0.00128,
+      "grad_norm": 0.8466276209824322,
+      "learning_rate": 0.000384,
+      "loss": 8.0264,
+      "step": 128
+    },
+    {
+      "epoch": 0.00129,
+      "grad_norm": 1.1259201196462498,
+      "learning_rate": 0.00038700000000000003,
+      "loss": 8.0199,
+      "step": 129
+    },
+    {
+      "epoch": 0.0013,
+      "grad_norm": 1.356061683657538,
+      "learning_rate": 0.00039000000000000005,
+      "loss": 8.0054,
+      "step": 130
+    },
+    {
+      "epoch": 0.00131,
+      "grad_norm": 0.781068809712636,
+      "learning_rate": 0.000393,
+      "loss": 7.9576,
+      "step": 131
+    },
+    {
+      "epoch": 0.00132,
+      "grad_norm": 1.0367188815921278,
+      "learning_rate": 0.00039600000000000003,
+      "loss": 7.9591,
+      "step": 132
+    },
+    {
+      "epoch": 0.00133,
+      "grad_norm": 0.7841067594214959,
+      "learning_rate": 0.00039900000000000005,
+      "loss": 7.9293,
+      "step": 133
+    },
+    {
+      "epoch": 0.00134,
+      "grad_norm": 0.8594113418007541,
+      "learning_rate": 0.000402,
+      "loss": 7.9217,
+      "step": 134
+    },
+    {
+      "epoch": 0.00135,
+      "grad_norm": 0.7468719464596628,
+      "learning_rate": 0.00040500000000000003,
+      "loss": 7.8883,
+      "step": 135
+    },
+    {
+      "epoch": 0.00136,
+      "grad_norm": 0.8415432392576198,
+      "learning_rate": 0.00040800000000000005,
+      "loss": 7.8771,
+      "step": 136
+    },
+    {
+      "epoch": 0.00137,
+      "grad_norm": 0.7386877288273068,
+      "learning_rate": 0.000411,
+      "loss": 7.8542,
+      "step": 137
+    },
+    {
+      "epoch": 0.00138,
+      "grad_norm": 0.7450176106214967,
+      "learning_rate": 0.00041400000000000003,
+      "loss": 7.8322,
+      "step": 138
+    },
+    {
+      "epoch": 0.00139,
+      "grad_norm": 0.7424698225823185,
+      "learning_rate": 0.00041700000000000005,
+      "loss": 7.8197,
+      "step": 139
+    },
+    {
+      "epoch": 0.0014,
+      "grad_norm": 0.7268306219948636,
+      "learning_rate": 0.00042000000000000007,
+      "loss": 7.8018,
+      "step": 140
+    },
+    {
+      "epoch": 0.00141,
+      "grad_norm": 0.7115037341291065,
+      "learning_rate": 0.000423,
+      "loss": 7.7905,
+      "step": 141
+    },
+    {
+      "epoch": 0.00142,
+      "grad_norm": 0.6725409058271569,
+      "learning_rate": 0.00042599999999999995,
+      "loss": 7.7772,
+      "step": 142
+    },
+    {
+      "epoch": 0.00143,
+      "grad_norm": 0.6830809074405504,
+      "learning_rate": 0.00042899999999999997,
+      "loss": 7.7496,
+      "step": 143
+    },
+    {
+      "epoch": 0.00144,
+      "grad_norm": 0.6791849969278475,
+      "learning_rate": 0.000432,
+      "loss": 7.7318,
+      "step": 144
+    },
+    {
+      "epoch": 0.00145,
+      "grad_norm": 0.6462817248800249,
+      "learning_rate": 0.000435,
+      "loss": 7.7251,
+      "step": 145
+    },
+    {
+      "epoch": 0.00146,
+      "grad_norm": 0.6695930969912641,
+      "learning_rate": 0.00043799999999999997,
+      "loss": 7.6893,
+      "step": 146
+    },
+    {
+      "epoch": 0.00147,
+      "grad_norm": 0.7012896651032599,
+      "learning_rate": 0.000441,
+      "loss": 7.6817,
+      "step": 147
+    },
+    {
+      "epoch": 0.00148,
+      "grad_norm": 0.7755580877429182,
+      "learning_rate": 0.000444,
+      "loss": 7.6698,
+      "step": 148
+    },
+    {
+      "epoch": 0.00149,
+      "grad_norm": 1.088947674236225,
+      "learning_rate": 0.00044699999999999997,
+      "loss": 7.6713,
+      "step": 149
+    },
+    {
+      "epoch": 0.0015,
+      "grad_norm": 0.9396541473912592,
+      "learning_rate": 0.00045,
+      "loss": 7.6436,
+      "step": 150
+    },
+    {
+      "epoch": 0.00151,
+      "grad_norm": 0.640217037835256,
+      "learning_rate": 0.000453,
+      "loss": 7.6133,
+      "step": 151
+    },
+    {
+      "epoch": 0.00152,
+      "grad_norm": 0.9098496631236208,
+      "learning_rate": 0.000456,
+      "loss": 7.6149,
+      "step": 152
+    },
+    {
+      "epoch": 0.00153,
+      "grad_norm": 0.6175759444520236,
+      "learning_rate": 0.000459,
+      "loss": 7.5962,
+      "step": 153
+    },
+    {
+      "epoch": 0.00154,
+      "grad_norm": 0.6884616968083866,
+      "learning_rate": 0.000462,
+      "loss": 7.5772,
+      "step": 154
+    },
+    {
+      "epoch": 0.00155,
+      "grad_norm": 0.6360938051960316,
+      "learning_rate": 0.000465,
+      "loss": 7.5598,
+      "step": 155
+    },
+    {
+      "epoch": 0.00156,
+      "grad_norm": 0.6937352973699618,
+      "learning_rate": 0.000468,
+      "loss": 7.5366,
+      "step": 156
+    },
+    {
+      "epoch": 0.00157,
+      "grad_norm": 0.6334556877551312,
+      "learning_rate": 0.000471,
+      "loss": 7.53,
+      "step": 157
+    },
+    {
+      "epoch": 0.00158,
+      "grad_norm": 0.9707170671679272,
+      "learning_rate": 0.00047400000000000003,
+      "loss": 7.5153,
+      "step": 158
+    },
+    {
+      "epoch": 0.00159,
+      "grad_norm": 1.0240750882994218,
+      "learning_rate": 0.000477,
+      "loss": 7.5076,
+      "step": 159
+    },
+    {
+      "epoch": 0.0016,
+      "grad_norm": 0.9322894026235434,
+      "learning_rate": 0.00048,
+      "loss": 7.4836,
+      "step": 160
+    },
+    {
+      "epoch": 0.00161,
+      "grad_norm": 0.5279037898518898,
+      "learning_rate": 0.00048300000000000003,
+      "loss": 7.4527,
+      "step": 161
+    },
+    {
+      "epoch": 0.00162,
+      "grad_norm": 0.687356662308957,
+      "learning_rate": 0.00048600000000000005,
+      "loss": 7.4601,
+      "step": 162
+    },
+    {
+      "epoch": 0.00163,
+      "grad_norm": 0.5623951705594973,
+      "learning_rate": 0.0004890000000000001,
+      "loss": 7.4388,
+      "step": 163
+    },
+    {
+      "epoch": 0.00164,
+      "grad_norm": 0.5581337114560441,
+      "learning_rate": 0.000492,
+      "loss": 7.4399,
+      "step": 164
+    },
+    {
+      "epoch": 0.00165,
+      "grad_norm": 0.5516159301488641,
+      "learning_rate": 0.000495,
+      "loss": 7.4126,
+      "step": 165
+    },
+    {
+      "epoch": 0.00166,
+      "grad_norm": 0.5242244578051735,
+      "learning_rate": 0.0004980000000000001,
+      "loss": 7.3876,
+      "step": 166
+    },
+    {
+      "epoch": 0.00167,
+      "grad_norm": 0.45997959649003123,
+      "learning_rate": 0.000501,
+      "loss": 7.3779,
+      "step": 167
+    },
+    {
+      "epoch": 0.00168,
+      "grad_norm": 0.5436289820614866,
+      "learning_rate": 0.000504,
+      "loss": 7.3569,
+      "step": 168
+    },
+    {
+      "epoch": 0.00169,
+      "grad_norm": 0.4983067598465849,
+      "learning_rate": 0.0005070000000000001,
+      "loss": 7.3495,
+      "step": 169
+    },
+    {
+      "epoch": 0.0017,
+      "grad_norm": 0.4402852485923817,
+      "learning_rate": 0.00051,
+      "loss": 7.3316,
+      "step": 170
+    },
+    {
+      "epoch": 0.00171,
+      "grad_norm": 0.5221521396945228,
+      "learning_rate": 0.000513,
+      "loss": 7.3138,
+      "step": 171
+    },
+    {
+      "epoch": 0.00172,
+      "grad_norm": 0.45544347662440743,
+      "learning_rate": 0.000516,
+      "loss": 7.3129,
+      "step": 172
+    },
+    {
+      "epoch": 0.00173,
+      "grad_norm": 0.4745602833877857,
+      "learning_rate": 0.0005189999999999999,
+      "loss": 7.2961,
+      "step": 173
+    },
+    {
+      "epoch": 0.00174,
+      "grad_norm": 0.5121580158942174,
+      "learning_rate": 0.000522,
+      "loss": 7.311,
+      "step": 174
+    },
+    {
+      "epoch": 0.00175,
+      "grad_norm": 0.680505499537256,
+      "learning_rate": 0.000525,
+      "loss": 7.2769,
+      "step": 175
+    },
+    {
+      "epoch": 0.00176,
+      "grad_norm": 0.7210959926983863,
+      "learning_rate": 0.0005279999999999999,
+      "loss": 7.2549,
+      "step": 176
+    },
+    {
+      "epoch": 0.00177,
+      "grad_norm": 0.7510526045152774,
+      "learning_rate": 0.000531,
+      "loss": 7.2572,
+      "step": 177
+    },
+    {
+      "epoch": 0.00178,
+      "grad_norm": 0.7343359791017195,
+      "learning_rate": 0.000534,
+      "loss": 7.2506,
+      "step": 178
+    },
+    {
+      "epoch": 0.00179,
+      "grad_norm": 0.6185909230661502,
+      "learning_rate": 0.000537,
+      "loss": 7.2371,
+      "step": 179
+    },
+    {
+      "epoch": 0.0018,
+      "grad_norm": 0.42887581745789505,
+      "learning_rate": 0.00054,
+      "loss": 7.2041,
+      "step": 180
+    },
+    {
+      "epoch": 0.00181,
+      "grad_norm": 0.5359194273155663,
+      "learning_rate": 0.000543,
+      "loss": 7.1961,
+      "step": 181
+    },
+    {
+      "epoch": 0.00182,
+      "grad_norm": 0.4369454993609359,
+      "learning_rate": 0.000546,
+      "loss": 7.188,
+      "step": 182
+    },
+    {
+      "epoch": 0.00183,
+      "grad_norm": 0.41001210068633426,
+      "learning_rate": 0.000549,
+      "loss": 7.1769,
+      "step": 183
+    },
+    {
+      "epoch": 0.00184,
+      "grad_norm": 0.4396906853721559,
+      "learning_rate": 0.000552,
+      "loss": 7.1646,
+      "step": 184
+    },
+    {
+      "epoch": 0.00185,
+      "grad_norm": 0.4311329351383538,
+      "learning_rate": 0.000555,
+      "loss": 7.1403,
+      "step": 185
+    },
+    {
+      "epoch": 0.00186,
+      "grad_norm": 0.44013221474943204,
+      "learning_rate": 0.000558,
+      "loss": 7.1495,
+      "step": 186
+    },
+    {
+      "epoch": 0.00187,
+      "grad_norm": 0.42719641993731927,
+      "learning_rate": 0.000561,
+      "loss": 7.1244,
+      "step": 187
+    },
+    {
+      "epoch": 0.00188,
+      "grad_norm": 0.39192004433941763,
+      "learning_rate": 0.000564,
+      "loss": 7.1022,
+      "step": 188
+    },
+    {
+      "epoch": 0.00189,
+      "grad_norm": 0.5474017259570552,
+      "learning_rate": 0.000567,
+      "loss": 7.1002,
+      "step": 189
+    },
+    {
+      "epoch": 0.0019,
+      "grad_norm": 0.5320208949420774,
+      "learning_rate": 0.00057,
+      "loss": 7.0824,
+      "step": 190
+    },
+    {
+      "epoch": 0.00191,
+      "grad_norm": 0.7661733453917681,
+      "learning_rate": 0.000573,
+      "loss": 7.0764,
+      "step": 191
+    },
+    {
+      "epoch": 0.00192,
+      "grad_norm": 0.9858385620454592,
+      "learning_rate": 0.000576,
+      "loss": 7.0775,
+      "step": 192
+    },
+    {
+      "epoch": 0.00193,
+      "grad_norm": 0.9591489873604585,
+      "learning_rate": 0.000579,
+      "loss": 7.0725,
+      "step": 193
+    },
+    {
+      "epoch": 0.00194,
+      "grad_norm": 0.8774582305545432,
+      "learning_rate": 0.000582,
+      "loss": 7.0715,
+      "step": 194
+    },
+    {
+      "epoch": 0.00195,
+      "grad_norm": 0.9015156284206135,
+      "learning_rate": 0.000585,
+      "loss": 7.0487,
+      "step": 195
+    },
+    {
+      "epoch": 0.00196,
+      "grad_norm": 0.8000421121121074,
+      "learning_rate": 0.000588,
+      "loss": 7.0339,
+      "step": 196
+    },
+    {
+      "epoch": 0.00197,
+      "grad_norm": 0.8468645119231468,
+      "learning_rate": 0.000591,
+      "loss": 7.0329,
+      "step": 197
+    },
+    {
+      "epoch": 0.00198,
+      "grad_norm": 0.6924161058762034,
+      "learning_rate": 0.000594,
+      "loss": 7.0197,
+      "step": 198
+    },
+    {
+      "epoch": 0.00199,
+      "grad_norm": 0.5671884633245193,
+      "learning_rate": 0.0005970000000000001,
+      "loss": 6.9992,
+      "step": 199
+    },
+    {
+      "epoch": 0.002,
+      "grad_norm": 0.5557515599131739,
+      "learning_rate": 0.0006000000000000001,
+      "loss": 6.9982,
+      "step": 200
+    },
+    {
+      "epoch": 0.00201,
+      "grad_norm": 0.4619645912916116,
+      "learning_rate": 0.000603,
+      "loss": 6.9694,
+      "step": 201
+    },
+    {
+      "epoch": 0.00202,
+      "grad_norm": 0.4929222182059183,
+      "learning_rate": 0.0006060000000000001,
+      "loss": 6.9601,
+      "step": 202
+    },
+    {
+      "epoch": 0.00203,
+      "grad_norm": 0.37702087689736097,
+      "learning_rate": 0.0006090000000000001,
+      "loss": 6.9599,
+      "step": 203
+    },
+    {
+      "epoch": 0.00204,
+      "grad_norm": 0.45518892473332057,
+      "learning_rate": 0.000612,
+      "loss": 6.9476,
+      "step": 204
+    },
+    {
+      "epoch": 0.00205,
+      "grad_norm": 0.384809834745988,
+      "learning_rate": 0.000615,
+      "loss": 6.9322,
+      "step": 205
+    },
+    {
+      "epoch": 0.00206,
+      "grad_norm": 0.38976835520829006,
+      "learning_rate": 0.000618,
+      "loss": 6.9255,
+      "step": 206
+    },
+    {
+      "epoch": 0.00207,
+      "grad_norm": 0.4375749283114229,
+      "learning_rate": 0.000621,
+      "loss": 6.9166,
+      "step": 207
+    },
+    {
+      "epoch": 0.00208,
+      "grad_norm": 0.6317927736115376,
+      "learning_rate": 0.000624,
+      "loss": 6.9157,
+      "step": 208
+    },
+    {
+      "epoch": 0.00209,
+      "grad_norm": 0.7772617938427908,
+      "learning_rate": 0.000627,
+      "loss": 6.9007,
+      "step": 209
+    },
+    {
+      "epoch": 0.0021,
+      "grad_norm": 1.2298132092555871,
+      "learning_rate": 0.00063,
+      "loss": 6.9015,
+      "step": 210
+    },
+    {
+      "epoch": 0.00211,
+      "grad_norm": 0.7435502579532725,
+      "learning_rate": 0.000633,
+      "loss": 6.8883,
+      "step": 211
+    },
+    {
+      "epoch": 0.00212,
+      "grad_norm": 0.3794792840193541,
+      "learning_rate": 0.000636,
+      "loss": 6.8693,
+      "step": 212
+    },
+    {
+      "epoch": 0.00213,
+      "grad_norm": 0.688114698891817,
+      "learning_rate": 0.000639,
+      "loss": 6.8589,
+      "step": 213
+    },
+    {
+      "epoch": 0.00214,
+      "grad_norm": 0.49868168831557785,
+      "learning_rate": 0.000642,
+      "loss": 6.8504,
+      "step": 214
+    },
+    {
+      "epoch": 0.00215,
+      "grad_norm": 0.6292678113208914,
+      "learning_rate": 0.000645,
+      "loss": 6.8506,
+      "step": 215
+    },
+    {
+      "epoch": 0.00216,
+      "grad_norm": 0.3673676300147008,
+      "learning_rate": 0.000648,
+      "loss": 6.8383,
+      "step": 216
+    },
+    {
+      "epoch": 0.00217,
+      "grad_norm": 0.4511278951279821,
+      "learning_rate": 0.000651,
+      "loss": 6.8131,
+      "step": 217
+    },
+    {
+      "epoch": 0.00218,
+      "grad_norm": 0.3783399316849399,
+      "learning_rate": 0.000654,
+      "loss": 6.8146,
+      "step": 218
+    },
+    {
+      "epoch": 0.00219,
+      "grad_norm": 0.3651267362185199,
+      "learning_rate": 0.000657,
+      "loss": 6.7867,
+      "step": 219
+    },
+    {
+      "epoch": 0.0022,
+      "grad_norm": 0.4708567379601776,
+      "learning_rate": 0.00066,
+      "loss": 6.8074,
+      "step": 220
+    },
+    {
+      "epoch": 0.00221,
+      "grad_norm": 0.5218809374007617,
+      "learning_rate": 0.0006630000000000001,
+      "loss": 6.7837,
+      "step": 221
+    },
+    {
+      "epoch": 0.00222,
+      "grad_norm": 0.7518789100021657,
+      "learning_rate": 0.000666,
+      "loss": 6.7774,
+      "step": 222
+    },
+    {
+      "epoch": 0.00223,
+      "grad_norm": 0.9964198530393009,
+      "learning_rate": 0.000669,
+      "loss": 6.7824,
+      "step": 223
+    },
+    {
+      "epoch": 0.00224,
+      "grad_norm": 0.807522133159797,
+      "learning_rate": 0.0006720000000000001,
+      "loss": 6.7748,
+      "step": 224
+    },
+    {
+      "epoch": 0.00225,
+      "grad_norm": 0.6803785939854445,
+      "learning_rate": 0.000675,
+      "loss": 6.7563,
+      "step": 225
+    },
+    {
+      "epoch": 0.00226,
+      "grad_norm": 1.0762485305388094,
+      "learning_rate": 0.000678,
+      "loss": 6.7681,
+      "step": 226
+    },
+    {
+      "epoch": 0.00227,
+      "grad_norm": 0.9129777590140102,
+      "learning_rate": 0.0006810000000000001,
+      "loss": 6.734,
+      "step": 227
+    },
+    {
+      "epoch": 0.00228,
+      "grad_norm": 1.299087365220233,
+      "learning_rate": 0.000684,
+      "loss": 6.7517,
+      "step": 228
+    },
+    {
+      "epoch": 0.00229,
+      "grad_norm": 0.7392729796935101,
+      "learning_rate": 0.000687,
+      "loss": 6.7331,
+      "step": 229
+    },
+    {
+      "epoch": 0.0023,
+      "grad_norm": 0.6158971034672628,
+      "learning_rate": 0.0006900000000000001,
+      "loss": 6.7089,
+      "step": 230
+    },
+    {
+      "epoch": 0.00231,
+      "grad_norm": 0.7016695647497186,
+      "learning_rate": 0.000693,
+      "loss": 6.6956,
+      "step": 231
+    },
+    {
+      "epoch": 0.00232,
+      "grad_norm": 0.6434185638703606,
+      "learning_rate": 0.000696,
+      "loss": 6.7082,
+      "step": 232
+    },
+    {
+      "epoch": 0.00233,
+      "grad_norm": 0.6044879418578446,
+      "learning_rate": 0.0006990000000000001,
+      "loss": 6.6921,
+      "step": 233
+    },
+    {
+      "epoch": 0.00234,
+      "grad_norm": 0.4480061456613071,
+      "learning_rate": 0.000702,
+      "loss": 6.679,
+      "step": 234
+    },
+    {
+      "epoch": 0.00235,
+      "grad_norm": 0.4492075259819563,
+      "learning_rate": 0.000705,
+      "loss": 6.6661,
+      "step": 235
+    },
+    {
+      "epoch": 0.00236,
+      "grad_norm": 0.3889271878969786,
+      "learning_rate": 0.000708,
+      "loss": 6.6458,
+      "step": 236
+    },
+    {
+      "epoch": 0.00237,
+      "grad_norm": 0.41607485698419117,
+      "learning_rate": 0.0007109999999999999,
+      "loss": 6.6575,
+      "step": 237
+    },
+    {
+      "epoch": 0.00238,
+      "grad_norm": 0.3515958046168432,
+      "learning_rate": 0.000714,
+      "loss": 6.6432,
+      "step": 238
+    },
+    {
+      "epoch": 0.00239,
+      "grad_norm": 0.4049983983930021,
+      "learning_rate": 0.000717,
+      "loss": 6.6274,
+      "step": 239
+    },
+    {
+      "epoch": 0.0024,
+      "grad_norm": 0.3247748587680522,
+      "learning_rate": 0.0007199999999999999,
+      "loss": 6.6414,
+      "step": 240
+    },
+    {
+      "epoch": 0.00241,
+      "grad_norm": 0.36262136655648425,
+      "learning_rate": 0.000723,
+      "loss": 6.6267,
+      "step": 241
+    },
+    {
+      "epoch": 0.00242,
+      "grad_norm": 0.3238608381517686,
+      "learning_rate": 0.000726,
+      "loss": 6.598,
+      "step": 242
+    },
+    {
+      "epoch": 0.00243,
+      "grad_norm": 0.31681258130419926,
+      "learning_rate": 0.000729,
+      "loss": 6.5877,
+      "step": 243
+    },
+    {
+      "epoch": 0.00244,
+      "grad_norm": 0.3616370322868285,
+      "learning_rate": 0.000732,
+      "loss": 6.5911,
+      "step": 244
+    },
+    {
+      "epoch": 0.00245,
+      "grad_norm": 0.4634076592576489,
+      "learning_rate": 0.000735,
+      "loss": 6.5815,
+      "step": 245
+    },
+    {
+      "epoch": 0.00246,
+      "grad_norm": 0.7486687333799987,
+      "learning_rate": 0.000738,
+      "loss": 6.5693,
+      "step": 246
+    },
+    {
+      "epoch": 0.00247,
+      "grad_norm": 1.662590620725261,
+      "learning_rate": 0.000741,
+      "loss": 6.5942,
+      "step": 247
+    },
+    {
+      "epoch": 0.00248,
+      "grad_norm": 0.7962452736060305,
+      "learning_rate": 0.000744,
+      "loss": 6.5799,
+      "step": 248
+    },
+    {
+      "epoch": 0.00249,
+      "grad_norm": 1.091558045553116,
+      "learning_rate": 0.000747,
+      "loss": 6.5681,
+      "step": 249
+    },
+    {
+      "epoch": 0.0025,
+      "grad_norm": 1.1566118279124307,
+      "learning_rate": 0.00075,
+      "loss": 6.5821,
+      "step": 250
+    },
+    {
+      "epoch": 0.00251,
+      "grad_norm": 0.7784764288643516,
+      "learning_rate": 0.000753,
+      "loss": 6.5311,
+      "step": 251
+    },
+    {
+      "epoch": 0.00252,
+      "grad_norm": 0.8327004278614876,
+      "learning_rate": 0.000756,
+      "loss": 6.5476,
+      "step": 252
+    },
+    {
+      "epoch": 0.00253,
+      "grad_norm": 0.8184368732684698,
+      "learning_rate": 0.000759,
+      "loss": 6.5327,
+      "step": 253
+    },
+    {
+      "epoch": 0.00254,
+      "grad_norm": 1.1486609061260633,
+      "learning_rate": 0.000762,
+      "loss": 6.541,
+      "step": 254
+    },
+    {
+      "epoch": 0.00255,
+      "grad_norm": 0.8085196503687528,
+      "learning_rate": 0.0007650000000000001,
+      "loss": 6.533,
+      "step": 255
+    },
+    {
+      "epoch": 0.00256,
+      "grad_norm": 0.6227788989369496,
+      "learning_rate": 0.000768,
+      "loss": 6.5039,
+      "step": 256
+    },
+    {
+      "epoch": 0.00257,
+      "grad_norm": 0.7880922816831363,
+      "learning_rate": 0.000771,
+      "loss": 6.5163,
+      "step": 257
+    },
+    {
+      "epoch": 0.00258,
+      "grad_norm": 0.8888953354311911,
+      "learning_rate": 0.0007740000000000001,
+      "loss": 6.497,
+      "step": 258
+    },
+    {
+      "epoch": 0.00259,
+      "grad_norm": 0.6556759438280303,
+      "learning_rate": 0.000777,
+      "loss": 6.491,
+      "step": 259
+    },
+    {
+      "epoch": 0.0026,
+      "grad_norm": 0.5556045743581063,
+      "learning_rate": 0.0007800000000000001,
+      "loss": 6.4736,
+      "step": 260
+    },
+    {
+      "epoch": 0.00261,
+      "grad_norm": 0.5757853625150707,
+      "learning_rate": 0.0007830000000000001,
+      "loss": 6.4835,
+      "step": 261
+    },
+    {
+      "epoch": 0.00262,
+      "grad_norm": 0.4702991355160636,
+      "learning_rate": 0.000786,
+      "loss": 6.4605,
+      "step": 262
+    },
+    {
+      "epoch": 0.00263,
+      "grad_norm": 0.5236270398277312,
+      "learning_rate": 0.0007890000000000001,
+      "loss": 6.4668,
+      "step": 263
+    },
+    {
+      "epoch": 0.00264,
+      "grad_norm": 0.57918798311576,
+      "learning_rate": 0.0007920000000000001,
+      "loss": 6.4494,
+      "step": 264
+    },
+    {
+      "epoch": 0.00265,
+      "grad_norm": 0.48386144230733374,
+      "learning_rate": 0.000795,
+      "loss": 6.4394,
+      "step": 265
+    },
+    {
+      "epoch": 0.00266,
+      "grad_norm": 0.44170660781498655,
+      "learning_rate": 0.0007980000000000001,
+      "loss": 6.4372,
+      "step": 266
+    },
+    {
+      "epoch": 0.00267,
+      "grad_norm": 0.35002993788441544,
+      "learning_rate": 0.0008010000000000001,
+      "loss": 6.4273,
+      "step": 267
+    },
+    {
+      "epoch": 0.00268,
+      "grad_norm": 0.49450621569184094,
+      "learning_rate": 0.000804,
+      "loss": 6.426,
+      "step": 268
+    },
+    {
+      "epoch": 0.00269,
+      "grad_norm": 0.5917094821390496,
+      "learning_rate": 0.0008070000000000001,
+      "loss": 6.4104,
+      "step": 269
+    },
+    {
+      "epoch": 0.0027,
+      "grad_norm": 0.8610573273382283,
+      "learning_rate": 0.0008100000000000001,
+      "loss": 6.4165,
+      "step": 270
+    },
+    {
+      "epoch": 0.00271,
+      "grad_norm": 0.9687752227819599,
+      "learning_rate": 0.000813,
+      "loss": 6.4342,
+      "step": 271
+    },
+    {
+      "epoch": 0.00272,
+      "grad_norm": 0.8252297813091672,
+      "learning_rate": 0.0008160000000000001,
+      "loss": 6.4109,
+      "step": 272
+    },
+    {
+      "epoch": 0.00273,
+      "grad_norm": 1.0509340869742279,
+      "learning_rate": 0.0008190000000000001,
+      "loss": 6.4091,
+      "step": 273
+    },
+    {
+      "epoch": 0.00274,
+      "grad_norm": 1.2492890299831212,
+      "learning_rate": 0.000822,
+      "loss": 6.3984,
+      "step": 274
+    },
+    {
+      "epoch": 0.00275,
+      "grad_norm": 0.7449833128353471,
+      "learning_rate": 0.0008250000000000001,
+      "loss": 6.3937,
+      "step": 275
+    },
+    {
+      "epoch": 0.00276,
+      "grad_norm": 0.650217944416595,
+      "learning_rate": 0.0008280000000000001,
+      "loss": 6.383,
+      "step": 276
+    },
+    {
+      "epoch": 0.00277,
+      "grad_norm": 0.6897831829588578,
+      "learning_rate": 0.0008310000000000001,
+      "loss": 6.3791,
+      "step": 277
+    },
+    {
+      "epoch": 0.00278,
+      "grad_norm": 0.6845751052836879,
+      "learning_rate": 0.0008340000000000001,
+      "loss": 6.3625,
+      "step": 278
+    },
+    {
+      "epoch": 0.00279,
+      "grad_norm": 0.6895680336476154,
+      "learning_rate": 0.0008370000000000001,
+      "loss": 6.3607,
+      "step": 279
+    },
+    {
+      "epoch": 0.0028,
+      "grad_norm": 0.7962805776833526,
+      "learning_rate": 0.0008400000000000001,
+      "loss": 6.3497,
+      "step": 280
+    },
+    {
+      "epoch": 0.00281,
+      "grad_norm": 0.7281805890288613,
+      "learning_rate": 0.0008430000000000001,
+      "loss": 6.3358,
+      "step": 281
+    },
+    {
+      "epoch": 0.00282,
+      "grad_norm": 0.6077351466654652,
+      "learning_rate": 0.000846,
+      "loss": 6.3343,
+      "step": 282
+    },
+    {
+      "epoch": 0.00283,
+      "grad_norm": 0.7021452115418912,
+      "learning_rate": 0.0008489999999999999,
+      "loss": 6.3268,
+      "step": 283
+    },
+    {
+      "epoch": 0.00284,
+      "grad_norm": 0.8067589645945545,
+      "learning_rate": 0.0008519999999999999,
+      "loss": 6.3217,
+      "step": 284
+    },
+    {
+      "epoch": 0.00285,
+      "grad_norm": 0.7238661685202905,
+      "learning_rate": 0.000855,
+      "loss": 6.3229,
+      "step": 285
+    },
+    {
+      "epoch": 0.00286,
+      "grad_norm": 0.6566110557704579,
+      "learning_rate": 0.0008579999999999999,
+      "loss": 6.3002,
+      "step": 286
+    },
+    {
+      "epoch": 0.00287,
+      "grad_norm": 0.5448339024770075,
+      "learning_rate": 0.000861,
+      "loss": 6.3016,
+      "step": 287
+    },
+    {
+      "epoch": 0.00288,
+      "grad_norm": 0.4516947530472597,
+      "learning_rate": 0.000864,
+      "loss": 6.279,
+      "step": 288
+    },
+    {
+      "epoch": 0.00289,
+      "grad_norm": 0.5415480649305122,
+      "learning_rate": 0.0008669999999999999,
+      "loss": 6.2929,
+      "step": 289
+    },
+    {
+      "epoch": 0.0029,
+      "grad_norm": 0.5812113854990302,
+      "learning_rate": 0.00087,
+      "loss": 6.2743,
+      "step": 290
+    },
+    {
+      "epoch": 0.00291,
+      "grad_norm": 0.5620937871349196,
+      "learning_rate": 0.000873,
+      "loss": 6.2679,
+      "step": 291
+    },
+    {
+      "epoch": 0.00292,
+      "grad_norm": 0.43362286284353735,
+      "learning_rate": 0.0008759999999999999,
+      "loss": 6.2618,
+      "step": 292
+    },
+    {
+      "epoch": 0.00293,
+      "grad_norm": 0.563953479113573,
+      "learning_rate": 0.000879,
+      "loss": 6.2491,
+      "step": 293
+    },
+    {
+      "epoch": 0.00294,
+      "grad_norm": 0.5784107789505917,
+      "learning_rate": 0.000882,
+      "loss": 6.2636,
+      "step": 294
+    },
+    {
+      "epoch": 0.00295,
+      "grad_norm": 0.5410292555611181,
+      "learning_rate": 0.0008849999999999999,
+      "loss": 6.2293,
+      "step": 295
+    },
+    {
+      "epoch": 0.00296,
+      "grad_norm": 0.5413778994680785,
+      "learning_rate": 0.000888,
+      "loss": 6.2428,
+      "step": 296
+    },
+    {
+      "epoch": 0.00297,
+      "grad_norm": 0.7587824592105622,
+      "learning_rate": 0.000891,
+      "loss": 6.2186,
+      "step": 297
+    },
+    {
+      "epoch": 0.00298,
+      "grad_norm": 1.2390223277612324,
+      "learning_rate": 0.0008939999999999999,
+      "loss": 6.2454,
+      "step": 298
+    },
+    {
+      "epoch": 0.00299,
+      "grad_norm": 1.0051162436627608,
+      "learning_rate": 0.000897,
+      "loss": 6.2195,
+      "step": 299
+    },
+    {
+      "epoch": 0.003,
+      "grad_norm": 0.9301924220055711,
+      "learning_rate": 0.0009,
+      "loss": 6.2223,
+      "step": 300
+    },
+    {
+      "epoch": 0.00301,
+      "grad_norm": 0.9092669241988458,
+      "learning_rate": 0.0009029999999999999,
+      "loss": 6.2133,
+      "step": 301
+    },
+    {
+      "epoch": 0.00302,
+      "grad_norm": 1.0280476917286576,
+      "learning_rate": 0.000906,
+      "loss": 6.2045,
+      "step": 302
+    },
+    {
+      "epoch": 0.00303,
+      "grad_norm": 0.955686909103224,
+      "learning_rate": 0.000909,
+      "loss": 6.2159,
+      "step": 303
+    },
+    {
+      "epoch": 0.00304,
+      "grad_norm": 0.924153033838841,
+      "learning_rate": 0.000912,
+      "loss": 6.2068,
+      "step": 304
+    },
+    {
+      "epoch": 0.00305,
+      "grad_norm": 1.0129569357421315,
+      "learning_rate": 0.000915,
+      "loss": 6.2162,
+      "step": 305
+    },
+    {
+      "epoch": 0.00306,
+      "grad_norm": 0.908196615580651,
+      "learning_rate": 0.000918,
+      "loss": 6.1982,
+      "step": 306
+    },
+    {
+      "epoch": 0.00307,
+      "grad_norm": 0.9412115799039344,
+      "learning_rate": 0.000921,
+      "loss": 6.1922,
+      "step": 307
+    },
+    {
+      "epoch": 0.00308,
+      "grad_norm": 1.0133210251014897,
+      "learning_rate": 0.000924,
+      "loss": 6.1801,
+      "step": 308
+    },
+    {
+      "epoch": 0.00309,
+      "grad_norm": 0.7885110954474147,
+      "learning_rate": 0.000927,
+      "loss": 6.1916,
+      "step": 309
+    },
+    {
+      "epoch": 0.0031,
+      "grad_norm": 0.8502930513548597,
+      "learning_rate": 0.00093,
+      "loss": 6.1767,
+      "step": 310
+    },
+    {
+      "epoch": 0.00311,
+      "grad_norm": 0.8830498260903274,
+      "learning_rate": 0.000933,
+      "loss": 6.1706,
+      "step": 311
+    },
+    {
+      "epoch": 0.00312,
+      "grad_norm": 0.7731682808518218,
+      "learning_rate": 0.000936,
+      "loss": 6.1483,
+      "step": 312
+    },
+    {
+      "epoch": 0.00313,
+      "grad_norm": 0.5034693382195244,
+      "learning_rate": 0.0009390000000000001,
+      "loss": 6.1541,
+      "step": 313
+    },
+    {
+      "epoch": 0.00314,
+      "grad_norm": 0.5944206771214186,
+      "learning_rate": 0.000942,
+      "loss": 6.1615,
+      "step": 314
+    },
+    {
+      "epoch": 0.00315,
+      "grad_norm": 0.6198949865082991,
+      "learning_rate": 0.000945,
+      "loss": 6.1382,
+      "step": 315
+    },
+    {
+      "epoch": 0.00316,
+      "grad_norm": 0.6322301787116348,
+      "learning_rate": 0.0009480000000000001,
+      "loss": 6.112,
+      "step": 316
+    },
+    {
+      "epoch": 0.00317,
+      "grad_norm": 0.6034799354522895,
+      "learning_rate": 0.000951,
+      "loss": 6.1294,
+      "step": 317
+    },
+    {
+      "epoch": 0.00318,
+      "grad_norm": 0.594812791501401,
+      "learning_rate": 0.000954,
+      "loss": 6.1243,
+      "step": 318
+    },
+    {
+      "epoch": 0.00319,
+      "grad_norm": 0.5121880287121718,
+      "learning_rate": 0.0009570000000000001,
+      "loss": 6.1104,
+      "step": 319
+    },
+    {
+      "epoch": 0.0032,
+      "grad_norm": 0.4383869814323231,
+      "learning_rate": 0.00096,
+      "loss": 6.1078,
+      "step": 320
+    },
+    {
+      "epoch": 0.00321,
+      "grad_norm": 0.46253996577569373,
+      "learning_rate": 0.000963,
+      "loss": 6.0957,
+      "step": 321
+    },
+    {
+      "epoch": 0.00322,
+      "grad_norm": 0.5330974668531825,
+      "learning_rate": 0.0009660000000000001,
+      "loss": 6.0852,
+      "step": 322
+    },
+    {
+      "epoch": 0.00323,
+      "grad_norm": 0.4269664457046623,
+      "learning_rate": 0.000969,
+      "loss": 6.0805,
+      "step": 323
+    },
+    {
+      "epoch": 0.00324,
+      "grad_norm": 0.4120363869720864,
+      "learning_rate": 0.0009720000000000001,
+      "loss": 6.0833,
+      "step": 324
+    },
+    {
+      "epoch": 0.00325,
+      "grad_norm": 0.43852524771556284,
+      "learning_rate": 0.0009750000000000001,
+      "loss": 6.0494,
+      "step": 325
+    },
+    {
+      "epoch": 0.00326,
+      "grad_norm": 0.5390182954459198,
+      "learning_rate": 0.0009780000000000001,
+      "loss": 6.0675,
+      "step": 326
+    },
+    {
+      "epoch": 0.00327,
+      "grad_norm": 0.7655676469366879,
+      "learning_rate": 0.000981,
+      "loss": 6.0729,
+      "step": 327
+    },
+    {
+      "epoch": 0.00328,
+      "grad_norm": 1.3095809996152918,
+      "learning_rate": 0.000984,
+      "loss": 6.0661,
+      "step": 328
+    },
+    {
+      "epoch": 0.00329,
+      "grad_norm": 0.9398963728555475,
+      "learning_rate": 0.000987,
+      "loss": 6.0593,
+      "step": 329
+    },
+    {
+      "epoch": 0.0033,
+      "grad_norm": 0.9687365266328093,
+      "learning_rate": 0.00099,
+      "loss": 6.0595,
+      "step": 330
+    },
+    {
+      "epoch": 0.00331,
+      "grad_norm": 1.0572725295338647,
+      "learning_rate": 0.0009930000000000002,
+      "loss": 6.0709,
+      "step": 331
+    },
+    {
+      "epoch": 0.00332,
+      "grad_norm": 0.7804253243852415,
+      "learning_rate": 0.0009960000000000001,
+      "loss": 6.0449,
+      "step": 332
+    },
+    {
+      "epoch": 0.00333,
+      "grad_norm": 0.9477444786898163,
+      "learning_rate": 0.000999,
+      "loss": 6.0352,
+      "step": 333
+    },
+    {
+      "epoch": 0.00334,
+      "grad_norm": 1.119636024931192,
+      "learning_rate": 0.001002,
+      "loss": 6.0509,
+      "step": 334
+    },
+    {
+      "epoch": 0.00335,
+      "grad_norm": 1.0398096390896505,
+      "learning_rate": 0.001005,
+      "loss": 6.0495,
+      "step": 335
+    },
+    {
+      "epoch": 0.00336,
+      "grad_norm": 0.8946811171098662,
+      "learning_rate": 0.001008,
+      "loss": 6.0348,
+      "step": 336
+    },
+    {
+      "epoch": 0.00337,
+      "grad_norm": 1.0240126692017184,
+      "learning_rate": 0.0010110000000000002,
+      "loss": 6.0215,
+      "step": 337
+    },
+    {
+      "epoch": 0.00338,
+      "grad_norm": 1.0245564805630587,
+      "learning_rate": 0.0010140000000000001,
+      "loss": 6.0414,
+      "step": 338
+    },
+    {
+      "epoch": 0.00339,
+      "grad_norm": 0.9606863659347135,
+      "learning_rate": 0.0010170000000000001,
+      "loss": 6.0439,
+      "step": 339
+    },
+    {
+      "epoch": 0.0034,
+      "grad_norm": 0.7549695201896788,
+      "learning_rate": 0.00102,
+      "loss": 6.0067,
+      "step": 340
+    },
+    {
+      "epoch": 0.00341,
+      "grad_norm": 0.7981881225838676,
+      "learning_rate": 0.001023,
+      "loss": 5.9935,
+      "step": 341
+    },
+    {
+      "epoch": 0.00342,
+      "grad_norm": 0.6476806075689499,
+      "learning_rate": 0.001026,
+      "loss": 5.9919,
+      "step": 342
+    },
+    {
+      "epoch": 0.00343,
+      "grad_norm": 0.6180530813345404,
+      "learning_rate": 0.0010290000000000002,
+      "loss": 5.9943,
+      "step": 343
+    },
+    {
+      "epoch": 0.00344,
+      "grad_norm": 0.5904235085071222,
+      "learning_rate": 0.001032,
+      "loss": 5.9824,
+      "step": 344
+    },
+    {
+      "epoch": 0.00345,
+      "grad_norm": 0.5662246103658737,
+      "learning_rate": 0.001035,
+      "loss": 5.9736,
+      "step": 345
+    },
+    {
+      "epoch": 0.00346,
+      "grad_norm": 0.5815647759401746,
+      "learning_rate": 0.0010379999999999999,
+      "loss": 5.9787,
+      "step": 346
+    },
+    {
+      "epoch": 0.00347,
+      "grad_norm": 0.5922585456492798,
+      "learning_rate": 0.001041,
+      "loss": 5.9505,
+      "step": 347
+    },
+    {
+      "epoch": 0.00348,
+      "grad_norm": 0.546770754957902,
+      "learning_rate": 0.001044,
+      "loss": 5.9566,
+      "step": 348
+    },
+    {
+      "epoch": 0.00349,
+      "grad_norm": 0.5259270224752429,
+      "learning_rate": 0.001047,
+      "loss": 5.9388,
+      "step": 349
+    },
+    {
+      "epoch": 0.0035,
+      "grad_norm": 0.5639453872626701,
+      "learning_rate": 0.00105,
+      "loss": 5.96,
+      "step": 350
+    },
+    {
+      "epoch": 0.00351,
+      "grad_norm": 0.562716119705,
+      "learning_rate": 0.001053,
+      "loss": 5.9575,
+      "step": 351
+    },
+    {
+      "epoch": 0.00352,
+      "grad_norm": 0.6276157353012132,
+      "learning_rate": 0.0010559999999999999,
+      "loss": 5.9357,
+      "step": 352
+    },
+    {
+      "epoch": 0.00353,
+      "grad_norm": 0.6899105072430818,
+      "learning_rate": 0.001059,
+      "loss": 5.939,
+      "step": 353
+    },
+    {
+      "epoch": 0.00354,
+      "grad_norm": 0.7414855002893123,
+      "learning_rate": 0.001062,
+      "loss": 5.9142,
+      "step": 354
+    },
+    {
+      "epoch": 0.00355,
+      "grad_norm": 0.8275245559939801,
+      "learning_rate": 0.001065,
+      "loss": 5.932,
+      "step": 355
+    },
+    {
+      "epoch": 0.00356,
+      "grad_norm": 0.8334699202442165,
+      "learning_rate": 0.001068,
+      "loss": 5.925,
+      "step": 356
+    },
+    {
+      "epoch": 0.00357,
+      "grad_norm": 0.733463396370079,
+      "learning_rate": 0.001071,
+      "loss": 5.9147,
+      "step": 357
+    },
+    {
+      "epoch": 0.00358,
+      "grad_norm": 0.5898600387972718,
+      "learning_rate": 0.001074,
+      "loss": 5.9022,
+      "step": 358
+    },
+    {
+      "epoch": 0.00359,
+      "grad_norm": 0.53807120274405,
+      "learning_rate": 0.001077,
+      "loss": 5.8933,
+      "step": 359
+    },
+    {
+      "epoch": 0.0036,
+      "grad_norm": 0.5868526923580266,
+      "learning_rate": 0.00108,
+      "loss": 5.8984,
+      "step": 360
+    },
+    {
+      "epoch": 0.00361,
+      "grad_norm": 0.6561371025439784,
+      "learning_rate": 0.001083,
+      "loss": 5.8927,
+      "step": 361
+    },
+    {
+      "epoch": 0.00362,
+      "grad_norm": 1.003392418604553,
+      "learning_rate": 0.001086,
+      "loss": 5.8918,
+      "step": 362
+    },
+    {
+      "epoch": 0.00363,
+      "grad_norm": 1.2318504963643775,
+      "learning_rate": 0.001089,
+      "loss": 5.899,
+      "step": 363
+    },
+    {
+      "epoch": 0.00364,
+      "grad_norm": 0.7073595215993886,
+      "learning_rate": 0.001092,
+      "loss": 5.8956,
+      "step": 364
+    },
+    {
+      "epoch": 0.00365,
+      "grad_norm": 0.6834408409212124,
+      "learning_rate": 0.001095,
+      "loss": 5.8924,
+      "step": 365
+    },
+    {
+      "epoch": 0.00366,
+      "grad_norm": 0.808409762735137,
+      "learning_rate": 0.001098,
+      "loss": 5.8732,
+      "step": 366
+    },
+    {
+      "epoch": 0.00367,
+      "grad_norm": 1.150363353399194,
+      "learning_rate": 0.001101,
+      "loss": 5.8871,
+      "step": 367
+    },
+    {
+      "epoch": 0.00368,
+      "grad_norm": 1.5759048240116487,
+      "learning_rate": 0.001104,
+      "loss": 5.9113,
+      "step": 368
+    },
+    {
+      "epoch": 0.00369,
+      "grad_norm": 0.6706600035773841,
+      "learning_rate": 0.001107,
+      "loss": 5.8826,
+      "step": 369
+    },
+    {
+      "epoch": 0.0037,
+      "grad_norm": 1.0165372867603926,
+      "learning_rate": 0.00111,
+      "loss": 5.8773,
+      "step": 370
+    },
+    {
+      "epoch": 0.00371,
+      "grad_norm": 1.199271508025637,
+      "learning_rate": 0.001113,
+      "loss": 5.8595,
+      "step": 371
+    },
+    {
+      "epoch": 0.00372,
+      "grad_norm": 0.8890339964058547,
+      "learning_rate": 0.001116,
+      "loss": 5.8693,
+      "step": 372
+    },
+    {
+      "epoch": 0.00373,
+      "grad_norm": 1.0853395614111072,
+      "learning_rate": 0.001119,
+      "loss": 5.8674,
+      "step": 373
+    },
+    {
+      "epoch": 0.00374,
+      "grad_norm": 0.9530307319001724,
+      "learning_rate": 0.001122,
+      "loss": 5.8703,
+      "step": 374
+    },
+    {
+      "epoch": 0.00375,
+      "grad_norm": 0.934750801302772,
+      "learning_rate": 0.0011250000000000001,
+      "loss": 5.8719,
+      "step": 375
+    },
+    {
+      "epoch": 0.00376,
+      "grad_norm": 0.7986123477309184,
+      "learning_rate": 0.001128,
+      "loss": 5.863,
+      "step": 376
+    },
+    {
+      "epoch": 0.00377,
+      "grad_norm": 1.034328210427756,
+      "learning_rate": 0.001131,
+      "loss": 5.8644,
+      "step": 377
+    },
+    {
+      "epoch": 0.00378,
+      "grad_norm": 1.1394107219722458,
+      "learning_rate": 0.001134,
+      "loss": 5.8702,
+      "step": 378
+    },
+    {
+      "epoch": 0.00379,
+      "grad_norm": 0.6778105205682995,
+      "learning_rate": 0.001137,
+      "loss": 5.8386,
+      "step": 379
+    },
+    {
+      "epoch": 0.0038,
+      "grad_norm": 0.7735792074169382,
+      "learning_rate": 0.00114,
+      "loss": 5.8619,
+      "step": 380
+    },
+    {
+      "epoch": 0.00381,
+      "grad_norm": 0.882590402307869,
+      "learning_rate": 0.0011430000000000001,
+      "loss": 5.83,
+      "step": 381
+    },
+    {
+      "epoch": 0.00382,
+      "grad_norm": 0.8860425089624719,
+      "learning_rate": 0.001146,
+      "loss": 5.8197,
+      "step": 382
+    },
+    {
+      "epoch": 0.00383,
+      "grad_norm": 1.0396684860073488,
+      "learning_rate": 0.001149,
+      "loss": 5.836,
+      "step": 383
+    },
+    {
+      "epoch": 0.00384,
+      "grad_norm": 0.9683520339794217,
+      "learning_rate": 0.001152,
+      "loss": 5.8386,
+      "step": 384
+    },
+    {
+      "epoch": 0.00385,
+      "grad_norm": 0.8210443052214895,
+      "learning_rate": 0.001155,
+      "loss": 5.8215,
+      "step": 385
+    },
+    {
+      "epoch": 0.00386,
+      "grad_norm": 1.030920073327714,
+      "learning_rate": 0.001158,
+      "loss": 5.8336,
+      "step": 386
+    },
+    {
+      "epoch": 0.00387,
+      "grad_norm": 0.8032114385497527,
+      "learning_rate": 0.0011610000000000001,
+      "loss": 5.8025,
+      "step": 387
+    },
+    {
+      "epoch": 0.00388,
+      "grad_norm": 0.6803620347459473,
+      "learning_rate": 0.001164,
+      "loss": 5.7978,
+      "step": 388
+    },
+    {
+      "epoch": 0.00389,
+      "grad_norm": 0.6752304208768743,
+      "learning_rate": 0.001167,
+      "loss": 5.8027,
+      "step": 389
+    },
+    {
+      "epoch": 0.0039,
+      "grad_norm": 0.6054825081153106,
+      "learning_rate": 0.00117,
+      "loss": 5.7828,
+      "step": 390
+    },
+    {
+      "epoch": 0.00391,
+      "grad_norm": 0.5156470046541872,
+      "learning_rate": 0.001173,
+      "loss": 5.7863,
+      "step": 391
+    },
+    {
+      "epoch": 0.00392,
+      "grad_norm": 0.49609211852516366,
+      "learning_rate": 0.001176,
+      "loss": 5.7945,
+      "step": 392
+    },
+    {
+      "epoch": 0.00393,
+      "grad_norm": 0.42817727304572534,
+      "learning_rate": 0.0011790000000000001,
+      "loss": 5.7664,
+      "step": 393
+    },
+    {
+      "epoch": 0.00394,
+      "grad_norm": 0.43894767278563757,
+      "learning_rate": 0.001182,
+      "loss": 5.7539,
+      "step": 394
+    },
+    {
+      "epoch": 0.00395,
+      "grad_norm": 0.41850660912289844,
+      "learning_rate": 0.001185,
+      "loss": 5.7366,
+      "step": 395
+    },
+    {
+      "epoch": 0.00396,
+      "grad_norm": 0.41745423249833347,
+      "learning_rate": 0.001188,
+      "loss": 5.7516,
+      "step": 396
+    },
+    {
+      "epoch": 0.00397,
+      "grad_norm": 0.40474070631964676,
+      "learning_rate": 0.001191,
+      "loss": 5.7433,
+      "step": 397
+    },
+    {
+      "epoch": 0.00398,
+      "grad_norm": 0.3677722018443306,
+      "learning_rate": 0.0011940000000000002,
+      "loss": 5.7479,
+      "step": 398
+    },
+    {
+      "epoch": 0.00399,
+      "grad_norm": 0.3480206547108819,
+      "learning_rate": 0.0011970000000000001,
+      "loss": 5.7478,
+      "step": 399
+    },
+    {
+      "epoch": 0.004,
+      "grad_norm": 0.3434828622202681,
+      "learning_rate": 0.0012000000000000001,
+      "loss": 5.7345,
+      "step": 400
+    },
+    {
+      "epoch": 0.00401,
+      "grad_norm": 0.34918136204349326,
+      "learning_rate": 0.001203,
+      "loss": 5.7155,
+      "step": 401
+    },
+    {
+      "epoch": 0.00402,
+      "grad_norm": 0.30554980038341767,
+      "learning_rate": 0.001206,
+      "loss": 5.7215,
+      "step": 402
+    },
+    {
+      "epoch": 0.00403,
+      "grad_norm": 0.38840665384838735,
+      "learning_rate": 0.001209,
+      "loss": 5.7184,
+      "step": 403
+    },
+    {
+      "epoch": 0.00404,
+      "grad_norm": 0.5409513056663879,
+      "learning_rate": 0.0012120000000000002,
+      "loss": 5.714,
+      "step": 404
+    },
+    {
+      "epoch": 0.00405,
+      "grad_norm": 0.845020924848713,
+      "learning_rate": 0.0012150000000000002,
+      "loss": 5.7181,
+      "step": 405
+    },
+    {
+      "epoch": 0.00406,
+      "grad_norm": 1.1911410915070972,
+      "learning_rate": 0.0012180000000000001,
+      "loss": 5.7205,
+      "step": 406
+    },
+    {
+      "epoch": 0.00407,
+      "grad_norm": 0.6577283880630926,
+      "learning_rate": 0.0012209999999999999,
+      "loss": 5.6994,
+      "step": 407
+    },
+    {
+      "epoch": 0.00408,
+      "grad_norm": 0.7475745975098248,
+      "learning_rate": 0.001224,
+      "loss": 5.7213,
+      "step": 408
+    },
+    {
+      "epoch": 0.00409,
+      "grad_norm": 0.9872391126413178,
+      "learning_rate": 0.001227,
+      "loss": 5.7126,
+      "step": 409
+    },
+    {
+      "epoch": 0.0041,
+      "grad_norm": 1.1359034721668335,
+      "learning_rate": 0.00123,
+      "loss": 5.7088,
+      "step": 410
+    },
+    {
+      "epoch": 0.00411,
+      "grad_norm": 1.3596329145222696,
+      "learning_rate": 0.001233,
+      "loss": 5.7402,
+      "step": 411
+    },
+    {
+      "epoch": 0.00412,
+      "grad_norm": 0.7538358474928969,
+      "learning_rate": 0.001236,
+      "loss": 5.7066,
+      "step": 412
+    },
+    {
+      "epoch": 0.00413,
+      "grad_norm": 0.9465320539051596,
+      "learning_rate": 0.0012389999999999999,
+      "loss": 5.7197,
+      "step": 413
+    },
+    {
+      "epoch": 0.00414,
+      "grad_norm": 0.9262933655624658,
+      "learning_rate": 0.001242,
+      "loss": 5.6978,
+      "step": 414
+    },
+    {
+      "epoch": 0.00415,
+      "grad_norm": 1.1564175286146172,
+      "learning_rate": 0.001245,
+      "loss": 5.7105,
+      "step": 415
+    },
+    {
+      "epoch": 0.00416,
+      "grad_norm": 1.1001247072345506,
+      "learning_rate": 0.001248,
+      "loss": 5.6929,
+      "step": 416
+    },
+    {
+      "epoch": 0.00417,
+      "grad_norm": 1.0416153435685582,
+      "learning_rate": 0.001251,
+      "loss": 5.7199,
+      "step": 417
+    },
+    {
+      "epoch": 0.00418,
+      "grad_norm": 1.0281555694116995,
+      "learning_rate": 0.001254,
+      "loss": 5.6999,
+      "step": 418
+    },
+    {
+      "epoch": 0.00419,
+      "grad_norm": 1.1154617103247704,
+      "learning_rate": 0.0012569999999999999,
+      "loss": 5.6876,
+      "step": 419
+    },
+    {
+      "epoch": 0.0042,
+      "grad_norm": 0.9999912825556322,
+      "learning_rate": 0.00126,
+      "loss": 5.7045,
+      "step": 420
+    },
+    {
+      "epoch": 0.00421,
+      "grad_norm": 0.9729638313238949,
+      "learning_rate": 0.001263,
+      "loss": 5.6933,
+      "step": 421
+    },
+    {
+      "epoch": 0.00422,
+      "grad_norm": 1.0297208241186608,
+      "learning_rate": 0.001266,
+      "loss": 5.6894,
+      "step": 422
+    },
+    {
+      "epoch": 0.00423,
+      "grad_norm": 0.8381062456476874,
+      "learning_rate": 0.001269,
+      "loss": 5.6811,
+      "step": 423
+    },
+    {
+      "epoch": 0.00424,
+      "grad_norm": 0.73825928337582,
+      "learning_rate": 0.001272,
+      "loss": 5.6773,
+      "step": 424
+    },
+    {
+      "epoch": 0.00425,
+      "grad_norm": 0.8151693610653118,
+      "learning_rate": 0.001275,
+      "loss": 5.6815,
+      "step": 425
+    },
+    {
+      "epoch": 0.00426,
+      "grad_norm": 0.9877723869544494,
+      "learning_rate": 0.001278,
+      "loss": 5.6881,
+      "step": 426
+    },
+    {
+      "epoch": 0.00427,
+      "grad_norm": 0.8875063218226354,
+      "learning_rate": 0.001281,
+      "loss": 5.6676,
+      "step": 427
+    },
+    {
+      "epoch": 0.00428,
+      "grad_norm": 0.8761493100474917,
+      "learning_rate": 0.001284,
+      "loss": 5.6437,
+      "step": 428
+    },
+    {
+      "epoch": 0.00429,
+      "grad_norm": 1.109439755296158,
+      "learning_rate": 0.001287,
+      "loss": 5.668,
+      "step": 429
+    },
+    {
+      "epoch": 0.0043,
+      "grad_norm": 0.8726361797071612,
+      "learning_rate": 0.00129,
+      "loss": 5.6611,
+      "step": 430
+    },
+    {
+      "epoch": 0.00431,
+      "grad_norm": 0.5601358924900582,
+      "learning_rate": 0.001293,
+      "loss": 5.6342,
+      "step": 431
+    },
+    {
+      "epoch": 0.00432,
+      "grad_norm": 0.5940988397687699,
+      "learning_rate": 0.001296,
+      "loss": 5.6448,
+      "step": 432
+    },
+    {
+      "epoch": 0.00433,
+      "grad_norm": 0.480538337754196,
+      "learning_rate": 0.001299,
+      "loss": 5.6336,
+      "step": 433
+    },
+    {
+      "epoch": 0.00434,
+      "grad_norm": 0.5069699406038057,
+      "learning_rate": 0.001302,
+      "loss": 5.6331,
+      "step": 434
+    },
+    {
+      "epoch": 0.00435,
+      "grad_norm": 0.417220569666801,
+      "learning_rate": 0.001305,
+      "loss": 5.6227,
+      "step": 435
+    },
+    {
+      "epoch": 0.00436,
+      "grad_norm": 0.46597565539240443,
+      "learning_rate": 0.001308,
+      "loss": 5.6039,
+      "step": 436
+    },
+    {
+      "epoch": 0.00437,
+      "grad_norm": 0.37605303006482044,
+      "learning_rate": 0.001311,
+      "loss": 5.6161,
+      "step": 437
+    },
+    {
+      "epoch": 0.00438,
+      "grad_norm": 0.3526339213940271,
+      "learning_rate": 0.001314,
+      "loss": 5.5977,
+      "step": 438
+    },
+    {
+      "epoch": 0.00439,
+      "grad_norm": 0.3618369277094543,
+      "learning_rate": 0.001317,
+      "loss": 5.6246,
+      "step": 439
+    },
+    {
+      "epoch": 0.0044,
+      "grad_norm": 0.3223158135938896,
+      "learning_rate": 0.00132,
+      "loss": 5.5938,
+      "step": 440
+    },
+    {
+      "epoch": 0.00441,
+      "grad_norm": 0.3386640445759432,
+      "learning_rate": 0.001323,
+      "loss": 5.5905,
+      "step": 441
+    },
+    {
+      "epoch": 0.00442,
+      "grad_norm": 0.3397360216396013,
+      "learning_rate": 0.0013260000000000001,
+      "loss": 5.5838,
+      "step": 442
+    },
+    {
+      "epoch": 0.00443,
+      "grad_norm": 0.34532354892574607,
+      "learning_rate": 0.001329,
+      "loss": 5.5832,
+      "step": 443
+    },
+    {
+      "epoch": 0.00444,
+      "grad_norm": 0.37928556611065656,
+      "learning_rate": 0.001332,
+      "loss": 5.5853,
+      "step": 444
+    },
+    {
+      "epoch": 0.00445,
+      "grad_norm": 0.44947894711961484,
+      "learning_rate": 0.001335,
+      "loss": 5.5723,
+      "step": 445
+    },
+    {
+      "epoch": 0.00446,
+      "grad_norm": 0.5239308309237933,
+      "learning_rate": 0.001338,
+      "loss": 5.5751,
+      "step": 446
+    },
+    {
+      "epoch": 0.00447,
+      "grad_norm": 0.60889528761374,
+      "learning_rate": 0.001341,
+      "loss": 5.5777,
+      "step": 447
+    },
+    {
+      "epoch": 0.00448,
+      "grad_norm": 0.6150556040535831,
+      "learning_rate": 0.0013440000000000001,
+      "loss": 5.561,
+      "step": 448
+    },
+    {
+      "epoch": 0.00449,
+      "grad_norm": 0.5444850551876294,
+      "learning_rate": 0.001347,
+      "loss": 5.5623,
+      "step": 449
+    },
+    {
+      "epoch": 0.0045,
+      "grad_norm": 0.6163688303714219,
+      "learning_rate": 0.00135,
+      "loss": 5.5617,
+      "step": 450
+    },
+    {
+      "epoch": 0.00451,
+      "grad_norm": 0.7972728836658292,
+      "learning_rate": 0.001353,
+      "loss": 5.5614,
+      "step": 451
+    },
+    {
+      "epoch": 0.00452,
+      "grad_norm": 0.7711764764543457,
+      "learning_rate": 0.001356,
+      "loss": 5.5454,
+      "step": 452
+    },
+    {
+      "epoch": 0.00453,
+      "grad_norm": 0.7702356986189732,
+      "learning_rate": 0.001359,
+      "loss": 5.5379,
+      "step": 453
+    },
+    {
+      "epoch": 0.00454,
+      "grad_norm": 1.0838160417982272,
+      "learning_rate": 0.0013620000000000001,
+      "loss": 5.5686,
+      "step": 454
+    },
+    {
+      "epoch": 0.00455,
+      "grad_norm": 1.152655585803101,
+      "learning_rate": 0.0013650000000000001,
+      "loss": 5.5757,
+      "step": 455
+    },
+    {
+      "epoch": 0.00456,
+      "grad_norm": 1.2776355160689266,
+      "learning_rate": 0.001368,
+      "loss": 5.5831,
+      "step": 456
+    },
+    {
+      "epoch": 0.00457,
+      "grad_norm": 1.091292995937963,
+      "learning_rate": 0.001371,
+      "loss": 5.5727,
+      "step": 457
+    },
+    {
+      "epoch": 0.00458,
+      "grad_norm": 0.8963405103823251,
+      "learning_rate": 0.001374,
+      "loss": 5.5735,
+      "step": 458
+    },
+    {
+      "epoch": 0.00459,
+      "grad_norm": 1.0168648046101516,
+      "learning_rate": 0.0013770000000000002,
+      "loss": 5.5669,
+      "step": 459
+    },
+    {
+      "epoch": 0.0046,
+      "grad_norm": 1.296131601782866,
+      "learning_rate": 0.0013800000000000002,
+      "loss": 5.5665,
+      "step": 460
+    },
+    {
+      "epoch": 0.00461,
+      "grad_norm": 0.9634420565591739,
+      "learning_rate": 0.0013830000000000001,
+      "loss": 5.556,
+      "step": 461
+    },
+    {
+      "epoch": 0.00462,
+      "grad_norm": 0.9383281224355017,
+      "learning_rate": 0.001386,
+      "loss": 5.5634,
+      "step": 462
+    },
+    {
+      "epoch": 0.00463,
+      "grad_norm": 1.2569585597421309,
+      "learning_rate": 0.001389,
+      "loss": 5.5542,
+      "step": 463
+    },
+    {
+      "epoch": 0.00464,
+      "grad_norm": 0.9874595595654581,
+      "learning_rate": 0.001392,
+      "loss": 5.5689,
+      "step": 464
+    },
+    {
+      "epoch": 0.00465,
+      "grad_norm": 1.085595749506429,
+      "learning_rate": 0.0013950000000000002,
+      "loss": 5.5385,
+      "step": 465
+    },
+    {
+      "epoch": 0.00466,
+      "grad_norm": 1.0673943770446899,
+      "learning_rate": 0.0013980000000000002,
+      "loss": 5.5603,
+      "step": 466
+    },
+    {
+      "epoch": 0.00467,
+      "grad_norm": 0.8139501043376736,
+      "learning_rate": 0.0014010000000000001,
+      "loss": 5.5432,
+      "step": 467
+    },
+    {
+      "epoch": 0.00468,
+      "grad_norm": 0.7494382793960519,
+      "learning_rate": 0.001404,
+      "loss": 5.5245,
+      "step": 468
+    },
+    {
+      "epoch": 0.00469,
+      "grad_norm": 0.7634992086588068,
+      "learning_rate": 0.001407,
+      "loss": 5.5282,
+      "step": 469
+    },
+    {
+      "epoch": 0.0047,
+      "grad_norm": 0.8018093758476836,
+      "learning_rate": 0.00141,
+      "loss": 5.5404,
+      "step": 470
+    },
+    {
+      "epoch": 0.00471,
+      "grad_norm": 0.7418690809708749,
+      "learning_rate": 0.001413,
+      "loss": 5.5115,
+      "step": 471
+    },
+    {
+      "epoch": 0.00472,
+      "grad_norm": 0.7355325431039438,
+      "learning_rate": 0.001416,
+      "loss": 5.5216,
+      "step": 472
+    },
+    {
+      "epoch": 0.00473,
+      "grad_norm": 0.709026539269664,
+      "learning_rate": 0.001419,
+      "loss": 5.5305,
+      "step": 473
+    },
+    {
+      "epoch": 0.00474,
+      "grad_norm": 0.5742329758009745,
+      "learning_rate": 0.0014219999999999999,
+      "loss": 5.5064,
+      "step": 474
+    },
+    {
+      "epoch": 0.00475,
+      "grad_norm": 0.5859758403725885,
+      "learning_rate": 0.001425,
+      "loss": 5.4971,
+      "step": 475
+    },
+    {
+      "epoch": 0.00476,
+      "grad_norm": 0.6365922795308678,
+      "learning_rate": 0.001428,
+      "loss": 5.5308,
+      "step": 476
+    },
+    {
+      "epoch": 0.00477,
+      "grad_norm": 0.6539516343537074,
+      "learning_rate": 0.001431,
+      "loss": 5.4846,
+      "step": 477
+    },
+    {
+      "epoch": 0.00478,
+      "grad_norm": 0.6446859909585969,
+      "learning_rate": 0.001434,
+      "loss": 5.5074,
+      "step": 478
+    },
+    {
+      "epoch": 0.00479,
+      "grad_norm": 0.7791938725908187,
+      "learning_rate": 0.001437,
+      "loss": 5.4897,
+      "step": 479
+    },
+    {
+      "epoch": 0.0048,
+      "grad_norm": 0.7608695568360718,
+      "learning_rate": 0.0014399999999999999,
+      "loss": 5.4855,
+      "step": 480
+    },
+    {
+      "epoch": 0.00481,
+      "grad_norm": 0.5435552532069989,
+      "learning_rate": 0.001443,
+      "loss": 5.4813,
+      "step": 481
+    },
+    {
+      "epoch": 0.00482,
+      "grad_norm": 0.4961185149512517,
+      "learning_rate": 0.001446,
+      "loss": 5.4538,
+      "step": 482
+    },
+    {
+      "epoch": 0.00483,
+      "grad_norm": 0.5120902463904886,
+      "learning_rate": 0.001449,
+      "loss": 5.4636,
+      "step": 483
+    },
+    {
+      "epoch": 0.00484,
+      "grad_norm": 0.418197369903841,
+      "learning_rate": 0.001452,
+      "loss": 5.464,
+      "step": 484
+    },
+    {
+      "epoch": 0.00485,
+      "grad_norm": 0.36311406822078424,
+      "learning_rate": 0.001455,
+      "loss": 5.4671,
+      "step": 485
+    },
+    {
+      "epoch": 0.00486,
+      "grad_norm": 0.3913777576995821,
+      "learning_rate": 0.001458,
+      "loss": 5.4393,
+      "step": 486
+    },
+    {
+      "epoch": 0.00487,
+      "grad_norm": 0.36874474197662527,
+      "learning_rate": 0.001461,
+      "loss": 5.4517,
+      "step": 487
+    },
+    {
+      "epoch": 0.00488,
+      "grad_norm": 0.41593519822402414,
+      "learning_rate": 0.001464,
+      "loss": 5.43,
+      "step": 488
+    },
+    {
+      "epoch": 0.00489,
+      "grad_norm": 0.4051383375955623,
+      "learning_rate": 0.001467,
+      "loss": 5.4204,
+      "step": 489
+    },
+    {
+      "epoch": 0.0049,
+      "grad_norm": 0.4948319932753325,
+      "learning_rate": 0.00147,
+      "loss": 5.434,
+      "step": 490
+    },
+    {
+      "epoch": 0.00491,
+      "grad_norm": 0.6887805434617323,
+      "learning_rate": 0.001473,
+      "loss": 5.4357,
+      "step": 491
+    },
+    {
+      "epoch": 0.00492,
+      "grad_norm": 0.912047432270828,
+      "learning_rate": 0.001476,
+      "loss": 5.4432,
+      "step": 492
+    },
+    {
+      "epoch": 0.00493,
+      "grad_norm": 0.8950009277905591,
+      "learning_rate": 0.001479,
+      "loss": 5.4415,
+      "step": 493
+    },
+    {
+      "epoch": 0.00494,
+      "grad_norm": 0.807683799932231,
+      "learning_rate": 0.001482,
+      "loss": 5.4427,
+      "step": 494
+    },
+    {
+      "epoch": 0.00495,
+      "grad_norm": 0.976722169908224,
+      "learning_rate": 0.001485,
+      "loss": 5.4601,
+      "step": 495
+    },
+    {
+      "epoch": 0.00496,
+      "grad_norm": 0.7224780529872387,
+      "learning_rate": 0.001488,
+      "loss": 5.4314,
+      "step": 496
+    },
+    {
+      "epoch": 0.00497,
+      "grad_norm": 0.7139613919522917,
+      "learning_rate": 0.001491,
+      "loss": 5.4172,
+      "step": 497
+    },
+    {
+      "epoch": 0.00498,
+      "grad_norm": 0.7556637282468179,
+      "learning_rate": 0.001494,
+      "loss": 5.4443,
+      "step": 498
+    },
+    {
+      "epoch": 0.00499,
+      "grad_norm": 0.8519321120342865,
+      "learning_rate": 0.001497,
+      "loss": 5.4223,
+      "step": 499
+    },
+    {
+      "epoch": 0.005,
+      "grad_norm": 0.9868888032280079,
+      "learning_rate": 0.0015,
+      "loss": 5.4308,
+      "step": 500
+    },
+    {
+      "epoch": 0.00501,
+      "grad_norm": 1.2028724714017198,
+      "learning_rate": 0.001503,
+      "loss": 5.4458,
+      "step": 501
+    },
+    {
+      "epoch": 0.00502,
+      "grad_norm": 0.9548534640519003,
+      "learning_rate": 0.001506,
+      "loss": 5.4405,
+      "step": 502
+    },
+    {
+      "epoch": 0.00503,
+      "grad_norm": 0.9423651398338494,
+      "learning_rate": 0.0015090000000000001,
+      "loss": 5.4484,
+      "step": 503
+    },
+    {
+      "epoch": 0.00504,
+      "grad_norm": 0.9672711526274779,
+      "learning_rate": 0.001512,
+      "loss": 5.4328,
+      "step": 504
+    },
+    {
+      "epoch": 0.00505,
+      "grad_norm": 0.9474889600476256,
+      "learning_rate": 0.001515,
+      "loss": 5.4283,
+      "step": 505
+    },
+    {
+      "epoch": 0.00506,
+      "grad_norm": 1.3039451405080307,
+      "learning_rate": 0.001518,
+      "loss": 5.4432,
+      "step": 506
+    },
+    {
+      "epoch": 0.00507,
+      "grad_norm": 1.2269782118632737,
+      "learning_rate": 0.001521,
+      "loss": 5.4452,
+      "step": 507
+    },
+    {
+      "epoch": 0.00508,
+      "grad_norm": 0.8597426683067237,
+      "learning_rate": 0.001524,
+      "loss": 5.4268,
+      "step": 508
+    },
+    {
+      "epoch": 0.00509,
+      "grad_norm": 0.9388657090373522,
+      "learning_rate": 0.0015270000000000001,
+      "loss": 5.4178,
+      "step": 509
+    },
+    {
+      "epoch": 0.0051,
+      "grad_norm": 0.9594427813189665,
+      "learning_rate": 0.0015300000000000001,
+      "loss": 5.4356,
+      "step": 510
+    },
+    {
+      "epoch": 0.00511,
+      "grad_norm": 1.04563577032056,
+      "learning_rate": 0.001533,
+      "loss": 5.4212,
+      "step": 511
+    },
+    {
+      "epoch": 0.00512,
+      "grad_norm": 0.733703407645156,
+      "learning_rate": 0.001536,
+      "loss": 5.4003,
+      "step": 512
+    },
+    {
+      "epoch": 0.00513,
+      "grad_norm": 0.8415210942026606,
+      "learning_rate": 0.001539,
+      "loss": 5.423,
+      "step": 513
+    },
+    {
+      "epoch": 0.00514,
+      "grad_norm": 0.8791751992621939,
+      "learning_rate": 0.001542,
+      "loss": 5.4064,
+      "step": 514
+    },
+    {
+      "epoch": 0.00515,
+      "grad_norm": 0.8161499995578689,
+      "learning_rate": 0.0015450000000000001,
+      "loss": 5.4094,
+      "step": 515
+    },
+    {
+      "epoch": 0.00516,
+      "grad_norm": 0.878383695319614,
+      "learning_rate": 0.0015480000000000001,
+      "loss": 5.4087,
+      "step": 516
+    },
+    {
+      "epoch": 0.00517,
+      "grad_norm": 0.9768725869756134,
+      "learning_rate": 0.001551,
+      "loss": 5.4055,
+      "step": 517
+    },
+    {
+      "epoch": 0.00518,
+      "grad_norm": 0.8865017723772849,
+      "learning_rate": 0.001554,
+      "loss": 5.3907,
+      "step": 518
+    },
+    {
+      "epoch": 0.00519,
+      "grad_norm": 0.8308797688973832,
+      "learning_rate": 0.001557,
+      "loss": 5.3905,
+      "step": 519
+    },
+    {
+      "epoch": 0.0052,
+      "grad_norm": 0.6978413162257922,
+      "learning_rate": 0.0015600000000000002,
+      "loss": 5.3938,
+      "step": 520
+    },
+    {
+      "epoch": 0.00521,
+      "grad_norm": 0.6562689530690187,
+      "learning_rate": 0.0015630000000000002,
+      "loss": 5.3676,
+      "step": 521
+    },
+    {
+      "epoch": 0.00522,
+      "grad_norm": 0.5577523148431155,
+      "learning_rate": 0.0015660000000000001,
+      "loss": 5.3673,
+      "step": 522
+    },
+    {
+      "epoch": 0.00523,
+      "grad_norm": 0.5298728018270966,
+      "learning_rate": 0.001569,
+      "loss": 5.3784,
+      "step": 523
+    },
+    {
+      "epoch": 0.00524,
+      "grad_norm": 0.44216623389663734,
+      "learning_rate": 0.001572,
+      "loss": 5.3811,
+      "step": 524
+    },
+    {
+      "epoch": 0.00525,
+      "grad_norm": 0.3702182111689363,
+      "learning_rate": 0.001575,
+      "loss": 5.369,
+      "step": 525
+    },
+    {
+      "epoch": 0.00526,
+      "grad_norm": 0.40289552198632295,
+      "learning_rate": 0.0015780000000000002,
+      "loss": 5.3444,
+      "step": 526
+    },
+    {
+      "epoch": 0.00527,
+      "grad_norm": 0.38284579814689895,
+      "learning_rate": 0.0015810000000000002,
+      "loss": 5.3551,
+      "step": 527
+    },
+    {
+      "epoch": 0.00528,
+      "grad_norm": 0.33950473031510653,
+      "learning_rate": 0.0015840000000000001,
+      "loss": 5.3339,
+      "step": 528
+    },
+    {
+      "epoch": 0.00529,
+      "grad_norm": 0.3777758983585419,
+      "learning_rate": 0.001587,
+      "loss": 5.343,
+      "step": 529
+    },
+    {
+      "epoch": 0.0053,
+      "grad_norm": 0.4257451161382566,
+      "learning_rate": 0.00159,
+      "loss": 5.3483,
+      "step": 530
+    },
+    {
+      "epoch": 0.00531,
+      "grad_norm": 0.5472045550610978,
+      "learning_rate": 0.001593,
+      "loss": 5.3387,
+      "step": 531
+    },
+    {
+      "epoch": 0.00532,
+      "grad_norm": 0.696856419507981,
+      "learning_rate": 0.0015960000000000002,
+      "loss": 5.3455,
+      "step": 532
+    },
+    {
+      "epoch": 0.00533,
+      "grad_norm": 0.7975941430607876,
+      "learning_rate": 0.0015990000000000002,
+      "loss": 5.3447,
+      "step": 533
+    },
+    {
+      "epoch": 0.00534,
+      "grad_norm": 0.6576622452461942,
+      "learning_rate": 0.0016020000000000001,
+      "loss": 5.3339,
+      "step": 534
+    },
+    {
+      "epoch": 0.00535,
+      "grad_norm": 0.5879483820814444,
+      "learning_rate": 0.001605,
+      "loss": 5.3347,
+      "step": 535
+    },
+    {
+      "epoch": 0.00536,
+      "grad_norm": 0.8212636469840171,
+      "learning_rate": 0.001608,
+      "loss": 5.3454,
+      "step": 536
+    },
+    {
+      "epoch": 0.00537,
+      "grad_norm": 0.7740232031924225,
+      "learning_rate": 0.0016110000000000002,
+      "loss": 5.3418,
+      "step": 537
+    },
+    {
+      "epoch": 0.00538,
+      "grad_norm": 0.683098129060214,
+      "learning_rate": 0.0016140000000000002,
+      "loss": 5.3143,
+      "step": 538
+    },
+    {
+      "epoch": 0.00539,
+      "grad_norm": 0.9170551975741953,
+      "learning_rate": 0.0016170000000000002,
+      "loss": 5.3245,
+      "step": 539
+    },
+    {
+      "epoch": 0.0054,
+      "grad_norm": 0.765080131514484,
+      "learning_rate": 0.0016200000000000001,
+      "loss": 5.3262,
+      "step": 540
+    },
+    {
+      "epoch": 0.00541,
+      "grad_norm": 0.658457026305436,
+      "learning_rate": 0.001623,
+      "loss": 5.3137,
+      "step": 541
+    },
+    {
+      "epoch": 0.00542,
+      "grad_norm": 0.601942869875084,
+      "learning_rate": 0.001626,
+      "loss": 5.3315,
+      "step": 542
+    },
+    {
+      "epoch": 0.00543,
+      "grad_norm": 0.6751097730454854,
+      "learning_rate": 0.0016290000000000002,
+      "loss": 5.2998,
+      "step": 543
+    },
+    {
+      "epoch": 0.00544,
+      "grad_norm": 0.6943293389301006,
+      "learning_rate": 0.0016320000000000002,
+      "loss": 5.3191,
+      "step": 544
+    },
+    {
+      "epoch": 0.00545,
+      "grad_norm": 0.757591621302123,
+      "learning_rate": 0.0016350000000000002,
+      "loss": 5.3293,
+      "step": 545
+    },
+    {
+      "epoch": 0.00546,
+      "grad_norm": 0.825685624372282,
+      "learning_rate": 0.0016380000000000001,
+      "loss": 5.3061,
+      "step": 546
+    },
+    {
+      "epoch": 0.00547,
+      "grad_norm": 0.8411824339962438,
+      "learning_rate": 0.001641,
+      "loss": 5.3051,
+      "step": 547
+    },
+    {
+      "epoch": 0.00548,
+      "grad_norm": 0.8225125547688507,
+      "learning_rate": 0.001644,
+      "loss": 5.303,
+      "step": 548
+    },
+    {
+      "epoch": 0.00549,
+      "grad_norm": 0.7692066087665821,
+      "learning_rate": 0.0016470000000000002,
+      "loss": 5.3016,
+      "step": 549
+    },
+    {
+      "epoch": 0.0055,
+      "grad_norm": 0.7541680263658305,
+      "learning_rate": 0.0016500000000000002,
+      "loss": 5.3099,
+      "step": 550
+    },
+    {
+      "epoch": 0.00551,
+      "grad_norm": 0.8632990593818363,
+      "learning_rate": 0.0016530000000000002,
+      "loss": 5.3123,
+      "step": 551
+    },
+    {
+      "epoch": 0.00552,
+      "grad_norm": 0.9083583396116485,
+      "learning_rate": 0.0016560000000000001,
+      "loss": 5.3063,
+      "step": 552
+    },
+    {
+      "epoch": 0.00553,
+      "grad_norm": 0.9125681148017897,
+      "learning_rate": 0.001659,
+      "loss": 5.304,
+      "step": 553
+    },
+    {
+      "epoch": 0.00554,
+      "grad_norm": 0.9360850923631763,
+      "learning_rate": 0.0016620000000000003,
+      "loss": 5.3021,
+      "step": 554
+    },
+    {
+      "epoch": 0.00555,
+      "grad_norm": 0.9519012966124738,
+      "learning_rate": 0.0016650000000000002,
+      "loss": 5.3154,
+      "step": 555
+    },
+    {
+      "epoch": 0.00556,
+      "grad_norm": 0.9297335237671114,
+      "learning_rate": 0.0016680000000000002,
+      "loss": 5.2962,
+      "step": 556
+    },
+    {
+      "epoch": 0.00557,
+      "grad_norm": 1.1187109821510643,
+      "learning_rate": 0.0016710000000000002,
+      "loss": 5.2935,
+      "step": 557
+    },
+    {
+      "epoch": 0.00558,
+      "grad_norm": 0.9868599839032521,
+      "learning_rate": 0.0016740000000000001,
+      "loss": 5.3219,
+      "step": 558
+    },
+    {
+      "epoch": 0.00559,
+      "grad_norm": 0.8786610310513777,
+      "learning_rate": 0.001677,
+      "loss": 5.2938,
+      "step": 559
+    },
+    {
+      "epoch": 0.0056,
+      "grad_norm": 0.9118163270219521,
+      "learning_rate": 0.0016800000000000003,
+      "loss": 5.2767,
+      "step": 560
+    },
+    {
+      "epoch": 0.00561,
+      "grad_norm": 0.8649386991384733,
+      "learning_rate": 0.0016830000000000003,
+      "loss": 5.3004,
+      "step": 561
+    },
+    {
+      "epoch": 0.00562,
+      "grad_norm": 0.5982058914442256,
+      "learning_rate": 0.0016860000000000002,
+      "loss": 5.279,
+      "step": 562
+    },
+    {
+      "epoch": 0.00563,
+      "grad_norm": 0.5830709316445766,
+      "learning_rate": 0.001689,
+      "loss": 5.2821,
+      "step": 563
+    },
+    {
+      "epoch": 0.00564,
+      "grad_norm": 0.5564672250442253,
+      "learning_rate": 0.001692,
+      "loss": 5.2582,
+      "step": 564
+    },
+    {
+      "epoch": 0.00565,
+      "grad_norm": 0.5903966881939692,
+      "learning_rate": 0.001695,
+      "loss": 5.2619,
+      "step": 565
+    },
+    {
+      "epoch": 0.00566,
+      "grad_norm": 0.5777761648359326,
+      "learning_rate": 0.0016979999999999999,
+      "loss": 5.2552,
+      "step": 566
+    },
+    {
+      "epoch": 0.00567,
+      "grad_norm": 0.6069030603134064,
+      "learning_rate": 0.0017009999999999998,
+      "loss": 5.2491,
+      "step": 567
+    },
+    {
+      "epoch": 0.00568,
+      "grad_norm": 0.5693132087018719,
+      "learning_rate": 0.0017039999999999998,
+      "loss": 5.2604,
+      "step": 568
+    },
+    {
+      "epoch": 0.00569,
+      "grad_norm": 0.47662406778838745,
+      "learning_rate": 0.001707,
+      "loss": 5.2359,
+      "step": 569
+    },
+    {
+      "epoch": 0.0057,
+      "grad_norm": 0.49231021705037487,
+      "learning_rate": 0.00171,
+      "loss": 5.253,
+      "step": 570
+    },
+    {
+      "epoch": 0.00571,
+      "grad_norm": 0.4167352661720621,
+      "learning_rate": 0.001713,
+      "loss": 5.2491,
+      "step": 571
+    },
+    {
+      "epoch": 0.00572,
+      "grad_norm": 0.39011854138215074,
+      "learning_rate": 0.0017159999999999999,
+      "loss": 5.2258,
+      "step": 572
+    },
+    {
+      "epoch": 0.00573,
+      "grad_norm": 0.45971389455263184,
+      "learning_rate": 0.0017189999999999998,
+      "loss": 5.2313,
+      "step": 573
+    },
+    {
+      "epoch": 0.00574,
+      "grad_norm": 0.4643694238461635,
+      "learning_rate": 0.001722,
+      "loss": 5.2399,
+      "step": 574
+    },
+    {
+      "epoch": 0.00575,
+      "grad_norm": 0.4439840100405838,
+      "learning_rate": 0.001725,
+      "loss": 5.2168,
+      "step": 575
+    },
+    {
+      "epoch": 0.00576,
+      "grad_norm": 0.4794795740566699,
+      "learning_rate": 0.001728,
+      "loss": 5.2109,
+      "step": 576
+    },
+    {
+      "epoch": 0.00577,
+      "grad_norm": 0.5445912068882687,
+      "learning_rate": 0.001731,
+      "loss": 5.2354,
+      "step": 577
+    },
+    {
+      "epoch": 0.00578,
+      "grad_norm": 0.5881619532039025,
+      "learning_rate": 0.0017339999999999999,
+      "loss": 5.2156,
+      "step": 578
+    },
+    {
+      "epoch": 0.00579,
+      "grad_norm": 0.6688964155004951,
+      "learning_rate": 0.0017369999999999998,
+      "loss": 5.2064,
+      "step": 579
+    },
+    {
+      "epoch": 0.0058,
+      "grad_norm": 0.6446644513052245,
+      "learning_rate": 0.00174,
+      "loss": 5.2223,
+      "step": 580
+    },
+    {
+      "epoch": 0.00581,
+      "grad_norm": 0.7385299343409043,
+      "learning_rate": 0.001743,
+      "loss": 5.2222,
+      "step": 581
+    },
+    {
+      "epoch": 0.00582,
+      "grad_norm": 1.0486135149054512,
+      "learning_rate": 0.001746,
+      "loss": 5.2114,
+      "step": 582
+    },
+    {
+      "epoch": 0.00583,
+      "grad_norm": 0.9184260585056472,
+      "learning_rate": 0.001749,
+      "loss": 5.2374,
+      "step": 583
+    },
+    {
+      "epoch": 0.00584,
+      "grad_norm": 0.7724931135788974,
+      "learning_rate": 0.0017519999999999999,
+      "loss": 5.2235,
+      "step": 584
+    },
+    {
+      "epoch": 0.00585,
+      "grad_norm": 0.903942587317279,
+      "learning_rate": 0.0017549999999999998,
+      "loss": 5.2218,
+      "step": 585
+    },
+    {
+      "epoch": 0.00586,
+      "grad_norm": 0.8496888678331875,
+      "learning_rate": 0.001758,
+      "loss": 5.2272,
+      "step": 586
+    },
+    {
+      "epoch": 0.00587,
+      "grad_norm": 0.8580070219006531,
+      "learning_rate": 0.001761,
+      "loss": 5.2094,
+      "step": 587
+    },
+    {
+      "epoch": 0.00588,
+      "grad_norm": 1.0100984711915582,
+      "learning_rate": 0.001764,
+      "loss": 5.233,
+      "step": 588
+    },
+    {
+      "epoch": 0.00589,
+      "grad_norm": 0.9311738464832717,
+      "learning_rate": 0.001767,
+      "loss": 5.2289,
+      "step": 589
+    },
+    {
+      "epoch": 0.0059,
+      "grad_norm": 0.9363758859588256,
+      "learning_rate": 0.0017699999999999999,
+      "loss": 5.2155,
+      "step": 590
+    },
+    {
+      "epoch": 0.00591,
+      "grad_norm": 0.9695025163620248,
+      "learning_rate": 0.001773,
+      "loss": 5.2165,
+      "step": 591
+    },
+    {
+      "epoch": 0.00592,
+      "grad_norm": 0.9805216711841677,
+      "learning_rate": 0.001776,
+      "loss": 5.2417,
+      "step": 592
+    },
+    {
+      "epoch": 0.00593,
+      "grad_norm": 0.9827241162755458,
+      "learning_rate": 0.001779,
+      "loss": 5.2392,
+      "step": 593
+    },
+    {
+      "epoch": 0.00594,
+      "grad_norm": 1.1445433178010636,
+      "learning_rate": 0.001782,
+      "loss": 5.2522,
+      "step": 594
+    },
+    {
+      "epoch": 0.00595,
+      "grad_norm": 0.950699335886284,
+      "learning_rate": 0.001785,
+      "loss": 5.2351,
+      "step": 595
+    },
+    {
+      "epoch": 0.00596,
+      "grad_norm": 0.8373647963925929,
+      "learning_rate": 0.0017879999999999999,
+      "loss": 5.2269,
+      "step": 596
+    },
+    {
+      "epoch": 0.00597,
+      "grad_norm": 0.9799263768822638,
+      "learning_rate": 0.001791,
+      "loss": 5.2118,
+      "step": 597
+    },
+    {
+      "epoch": 0.00598,
+      "grad_norm": 0.9413379648187816,
+      "learning_rate": 0.001794,
+      "loss": 5.2215,
+      "step": 598
+    },
+    {
+      "epoch": 0.00599,
+      "grad_norm": 0.8392983563516706,
+      "learning_rate": 0.001797,
+      "loss": 5.2191,
+      "step": 599
+    },
+    {
+      "epoch": 0.006,
+      "grad_norm": 0.8414286288934597,
+      "learning_rate": 0.0018,
+      "loss": 5.2134,
+      "step": 600
+    },
+    {
+      "epoch": 0.00601,
+      "grad_norm": 0.8993201270126511,
+      "learning_rate": 0.001803,
+      "loss": 5.2089,
+      "step": 601
+    },
+    {
+      "epoch": 0.00602,
+      "grad_norm": 0.8678008919316418,
+      "learning_rate": 0.0018059999999999999,
+      "loss": 5.1957,
+      "step": 602
+    },
+    {
+      "epoch": 0.00603,
+      "grad_norm": 0.6967382011990112,
+      "learning_rate": 0.001809,
+      "loss": 5.2122,
+      "step": 603
+    },
+    {
+      "epoch": 0.00604,
+      "grad_norm": 0.5312401482691863,
+      "learning_rate": 0.001812,
+      "loss": 5.1788,
+      "step": 604
+    },
+    {
+      "epoch": 0.00605,
+      "grad_norm": 0.5795052729398412,
+      "learning_rate": 0.001815,
+      "loss": 5.1915,
+      "step": 605
+    },
+    {
+      "epoch": 0.00606,
+      "grad_norm": 0.6345039250131549,
+      "learning_rate": 0.001818,
+      "loss": 5.1894,
+      "step": 606
+    },
+    {
+      "epoch": 0.00607,
+      "grad_norm": 0.6716049737606092,
+      "learning_rate": 0.001821,
+      "loss": 5.1916,
+      "step": 607
+    },
+    {
+      "epoch": 0.00608,
+      "grad_norm": 0.631863512626499,
+      "learning_rate": 0.001824,
+      "loss": 5.1762,
+      "step": 608
+    },
+    {
+      "epoch": 0.00609,
+      "grad_norm": 0.5345713642167306,
+      "learning_rate": 0.001827,
+      "loss": 5.177,
+      "step": 609
+    },
+    {
+      "epoch": 0.0061,
+      "grad_norm": 0.455626202663485,
+      "learning_rate": 0.00183,
+      "loss": 5.1434,
+      "step": 610
+    },
+    {
+      "epoch": 0.00611,
+      "grad_norm": 0.46197362206927406,
+      "learning_rate": 0.001833,
+      "loss": 5.1489,
+      "step": 611
+    },
+    {
+      "epoch": 0.00612,
+      "grad_norm": 0.3943653846553885,
+      "learning_rate": 0.001836,
+      "loss": 5.1488,
+      "step": 612
+    },
+    {
+      "epoch": 0.00613,
+      "grad_norm": 0.4053887102586243,
+      "learning_rate": 0.001839,
+      "loss": 5.1409,
+      "step": 613
+    },
+    {
+      "epoch": 0.00614,
+      "grad_norm": 0.44363743610311057,
+      "learning_rate": 0.001842,
+      "loss": 5.1611,
+      "step": 614
+    },
+    {
+      "epoch": 0.00615,
+      "grad_norm": 0.40025757691173514,
+      "learning_rate": 0.001845,
+      "loss": 5.145,
+      "step": 615
+    },
+    {
+      "epoch": 0.00616,
+      "grad_norm": 0.4067321452673224,
+      "learning_rate": 0.001848,
+      "loss": 5.1299,
+      "step": 616
+    },
+    {
+      "epoch": 0.00617,
+      "grad_norm": 0.42403873536126996,
+      "learning_rate": 0.001851,
+      "loss": 5.1397,
+      "step": 617
+    },
+    {
+      "epoch": 0.00618,
+      "grad_norm": 0.44192084147381183,
+      "learning_rate": 0.001854,
+      "loss": 5.1265,
+      "step": 618
+    },
+    {
+      "epoch": 0.00619,
+      "grad_norm": 0.5104655865853382,
+      "learning_rate": 0.001857,
+      "loss": 5.1299,
+      "step": 619
+    },
+    {
+      "epoch": 0.0062,
+      "grad_norm": 0.6226145292933671,
+      "learning_rate": 0.00186,
+      "loss": 5.118,
+      "step": 620
+    },
+    {
+      "epoch": 0.00621,
+      "grad_norm": 0.745237667035521,
+      "learning_rate": 0.001863,
+      "loss": 5.1237,
+      "step": 621
+    },
+    {
+      "epoch": 0.00622,
+      "grad_norm": 0.7986865810762169,
+      "learning_rate": 0.001866,
+      "loss": 5.1267,
+      "step": 622
+    },
+    {
+      "epoch": 0.00623,
+      "grad_norm": 0.7723208155309241,
+      "learning_rate": 0.001869,
+      "loss": 5.1084,
+      "step": 623
+    },
+    {
+      "epoch": 0.00624,
+      "grad_norm": 0.8171878766946273,
+      "learning_rate": 0.001872,
+      "loss": 5.1246,
+      "step": 624
+    },
+    {
+      "epoch": 0.00625,
+      "grad_norm": 0.7607424856276187,
+      "learning_rate": 0.001875,
+      "loss": 5.1359,
+      "step": 625
+    },
+    {
+      "epoch": 0.00626,
+      "grad_norm": 0.7005729458198662,
+      "learning_rate": 0.0018780000000000001,
+      "loss": 5.1104,
+      "step": 626
+    },
+    {
+      "epoch": 0.00627,
+      "grad_norm": 0.6735260479679158,
+      "learning_rate": 0.001881,
+      "loss": 5.1015,
+      "step": 627
+    },
+    {
+      "epoch": 0.00628,
+      "grad_norm": 0.780486998163919,
+      "learning_rate": 0.001884,
+      "loss": 5.1349,
+      "step": 628
+    },
+    {
+      "epoch": 0.00629,
+      "grad_norm": 0.8746830311438225,
+      "learning_rate": 0.001887,
+      "loss": 5.1097,
+      "step": 629
+    },
+    {
+      "epoch": 0.0063,
+      "grad_norm": 0.9536711950620466,
+      "learning_rate": 0.00189,
+      "loss": 5.1137,
+      "step": 630
+    },
+    {
+      "epoch": 0.00631,
+      "grad_norm": 0.8628106812141149,
+      "learning_rate": 0.0018930000000000002,
+      "loss": 5.1035,
+      "step": 631
+    },
+    {
+      "epoch": 0.00632,
+      "grad_norm": 0.9320890333045916,
+      "learning_rate": 0.0018960000000000001,
+      "loss": 5.1281,
+      "step": 632
+    },
+    {
+      "epoch": 0.00633,
+      "grad_norm": 1.0593968691082751,
+      "learning_rate": 0.001899,
+      "loss": 5.1321,
+      "step": 633
+    },
+    {
+      "epoch": 0.00634,
+      "grad_norm": 0.943443350801409,
+      "learning_rate": 0.001902,
+      "loss": 5.1182,
+      "step": 634
+    },
+    {
+      "epoch": 0.00635,
+      "grad_norm": 0.9628484504873114,
+      "learning_rate": 0.001905,
+      "loss": 5.1089,
+      "step": 635
+    },
+    {
+      "epoch": 0.00636,
+      "grad_norm": 1.1043383962750646,
+      "learning_rate": 0.001908,
+      "loss": 5.1289,
+      "step": 636
+    },
+    {
+      "epoch": 0.00637,
+      "grad_norm": 0.8987493866500654,
+      "learning_rate": 0.0019110000000000002,
+      "loss": 5.1351,
+      "step": 637
+    },
+    {
+      "epoch": 0.00638,
+      "grad_norm": 0.9251804377428581,
+      "learning_rate": 0.0019140000000000001,
+      "loss": 5.1288,
+      "step": 638
+    },
+    {
+      "epoch": 0.00639,
+      "grad_norm": 0.854418425044198,
+      "learning_rate": 0.001917,
+      "loss": 5.0998,
+      "step": 639
+    },
+    {
+      "epoch": 0.0064,
+      "grad_norm": 0.9324816679284724,
+      "learning_rate": 0.00192,
+      "loss": 5.1038,
+      "step": 640
+    },
+    {
+      "epoch": 0.00641,
+      "grad_norm": 0.9892889234371413,
+      "learning_rate": 0.001923,
+      "loss": 5.1163,
+      "step": 641
+    },
+    {
+      "epoch": 0.00642,
+      "grad_norm": 1.0346602459121752,
+      "learning_rate": 0.001926,
+      "loss": 5.106,
+      "step": 642
+    },
+    {
+      "epoch": 0.00643,
+      "grad_norm": 0.8661994645957561,
+      "learning_rate": 0.0019290000000000002,
+      "loss": 5.117,
+      "step": 643
+    },
+    {
+      "epoch": 0.00644,
+      "grad_norm": 0.8724056100423225,
+      "learning_rate": 0.0019320000000000001,
+      "loss": 5.0889,
+      "step": 644
+    },
+    {
+      "epoch": 0.00645,
+      "grad_norm": 0.8584186184200229,
+      "learning_rate": 0.001935,
+      "loss": 5.1004,
+      "step": 645
+    },
+    {
+      "epoch": 0.00646,
+      "grad_norm": 0.7360558672224548,
+      "learning_rate": 0.001938,
+      "loss": 5.0955,
+      "step": 646
+    },
+    {
+      "epoch": 0.00647,
+      "grad_norm": 0.7977702647925389,
+      "learning_rate": 0.001941,
+      "loss": 5.1058,
+      "step": 647
+    },
+    {
+      "epoch": 0.00648,
+      "grad_norm": 0.7872116543506851,
+      "learning_rate": 0.0019440000000000002,
+      "loss": 5.0908,
+      "step": 648
+    },
+    {
+      "epoch": 0.00649,
+      "grad_norm": 0.7104658813349117,
+      "learning_rate": 0.0019470000000000002,
+      "loss": 5.0718,
+      "step": 649
+    },
+    {
+      "epoch": 0.0065,
+      "grad_norm": 0.7453763255239747,
+      "learning_rate": 0.0019500000000000001,
+      "loss": 5.0953,
+      "step": 650
+    },
+    {
+      "epoch": 0.00651,
+      "grad_norm": 0.7781624388594444,
+      "learning_rate": 0.001953,
+      "loss": 5.0758,
+      "step": 651
+    },
+    {
+      "epoch": 0.00652,
+      "grad_norm": 0.7616046275009601,
+      "learning_rate": 0.0019560000000000003,
+      "loss": 5.0661,
+      "step": 652
+    },
+    {
+      "epoch": 0.00653,
+      "grad_norm": 0.5945469625366651,
+      "learning_rate": 0.0019590000000000002,
+      "loss": 5.0539,
+      "step": 653
+    },
+    {
+      "epoch": 0.00654,
+      "grad_norm": 0.6024408595794577,
+      "learning_rate": 0.001962,
+      "loss": 5.0374,
+      "step": 654
+    },
+    {
+      "epoch": 0.00655,
+      "grad_norm": 0.5905307565923603,
+      "learning_rate": 0.001965,
+      "loss": 5.048,
+      "step": 655
+    },
+    {
+      "epoch": 0.00656,
+      "grad_norm": 0.5236322372626927,
+      "learning_rate": 0.001968,
+      "loss": 5.04,
+      "step": 656
+    },
+    {
+      "epoch": 0.00657,
+      "grad_norm": 0.5283416618835216,
+      "learning_rate": 0.001971,
+      "loss": 5.0223,
+      "step": 657
+    },
+    {
+      "epoch": 0.00658,
+      "grad_norm": 0.5563146586062104,
+      "learning_rate": 0.001974,
+      "loss": 5.0415,
+      "step": 658
+    },
+    {
+      "epoch": 0.00659,
+      "grad_norm": 0.6297873363395704,
+      "learning_rate": 0.001977,
+      "loss": 5.0241,
+      "step": 659
+    },
+    {
+      "epoch": 0.0066,
+      "grad_norm": 0.5780538180580159,
+      "learning_rate": 0.00198,
+      "loss": 5.0197,
+      "step": 660
+    },
+    {
+      "epoch": 0.00661,
+      "grad_norm": 0.5505266007864265,
+      "learning_rate": 0.001983,
+      "loss": 4.9938,
+      "step": 661
+    },
+    {
+      "epoch": 0.00662,
+      "grad_norm": 0.6077995116545319,
+      "learning_rate": 0.0019860000000000004,
+      "loss": 4.9946,
+      "step": 662
+    },
+    {
+      "epoch": 0.00663,
+      "grad_norm": 0.693544741809416,
+      "learning_rate": 0.0019890000000000003,
+      "loss": 5.01,
+      "step": 663
+    },
+    {
+      "epoch": 0.00664,
+      "grad_norm": 0.801492715102265,
+      "learning_rate": 0.0019920000000000003,
+      "loss": 5.0017,
+      "step": 664
+    },
+    {
+      "epoch": 0.00665,
+      "grad_norm": 0.8803866066519176,
+      "learning_rate": 0.0019950000000000002,
+      "loss": 4.9845,
+      "step": 665
+    },
+    {
+      "epoch": 0.00666,
+      "grad_norm": 0.9133314823416234,
+      "learning_rate": 0.001998,
+      "loss": 5.0048,
+      "step": 666
+    },
+    {
+      "epoch": 0.00667,
+      "grad_norm": 0.9006055773427947,
+      "learning_rate": 0.002001,
+      "loss": 4.9967,
+      "step": 667
+    },
+    {
+      "epoch": 0.00668,
+      "grad_norm": 0.8268536663020751,
+      "learning_rate": 0.002004,
+      "loss": 5.0164,
+      "step": 668
+    },
+    {
+      "epoch": 0.00669,
+      "grad_norm": 1.1034218430158187,
+      "learning_rate": 0.002007,
+      "loss": 5.0358,
+      "step": 669
+    },
+    {
+      "epoch": 0.0067,
+      "grad_norm": 1.0710112933622913,
+      "learning_rate": 0.00201,
+      "loss": 5.0286,
+      "step": 670
+    },
+    {
+      "epoch": 0.00671,
+      "grad_norm": 1.0384170566197124,
+      "learning_rate": 0.002013,
+      "loss": 5.0085,
+      "step": 671
+    },
+    {
+      "epoch": 0.00672,
+      "grad_norm": 0.9247071512184438,
+      "learning_rate": 0.002016,
+      "loss": 4.9973,
+      "step": 672
+    },
+    {
+      "epoch": 0.00673,
+      "grad_norm": 0.9905174678816935,
+      "learning_rate": 0.002019,
+      "loss": 5.0151,
+      "step": 673
+    },
+    {
+      "epoch": 0.00674,
+      "grad_norm": 0.8930692065183647,
+      "learning_rate": 0.0020220000000000004,
+      "loss": 5.0014,
+      "step": 674
+    },
+    {
+      "epoch": 0.00675,
+      "grad_norm": 1.0067908289229996,
+      "learning_rate": 0.0020250000000000003,
+      "loss": 5.0182,
+      "step": 675
+    },
+    {
+      "epoch": 0.00676,
+      "grad_norm": 0.972209366764115,
+      "learning_rate": 0.0020280000000000003,
+      "loss": 5.0082,
+      "step": 676
+    },
+    {
+      "epoch": 0.00677,
+      "grad_norm": 1.0440090424594235,
+      "learning_rate": 0.0020310000000000003,
+      "loss": 5.0109,
+      "step": 677
+    },
+    {
+      "epoch": 0.00678,
+      "grad_norm": 1.1292207401563255,
+      "learning_rate": 0.0020340000000000002,
+      "loss": 5.0278,
+      "step": 678
+    },
+    {
+      "epoch": 0.00679,
+      "grad_norm": 0.9872089241398577,
+      "learning_rate": 0.002037,
+      "loss": 5.0054,
+      "step": 679
+    },
+    {
+      "epoch": 0.0068,
+      "grad_norm": 0.8748535789650627,
+      "learning_rate": 0.00204,
+      "loss": 5.0004,
+      "step": 680
+    },
+    {
+      "epoch": 0.00681,
+      "grad_norm": 1.0039141852392888,
+      "learning_rate": 0.002043,
+      "loss": 5.0018,
+      "step": 681
+    },
+    {
+      "epoch": 0.00682,
+      "grad_norm": 1.0414085118011525,
+      "learning_rate": 0.002046,
+      "loss": 5.0026,
+      "step": 682
+    },
+    {
+      "epoch": 0.00683,
+      "grad_norm": 0.8083819733286619,
+      "learning_rate": 0.002049,
+      "loss": 4.9746,
+      "step": 683
+    },
+    {
+      "epoch": 0.00684,
+      "grad_norm": 0.618682103661302,
+      "learning_rate": 0.002052,
+      "loss": 4.9803,
+      "step": 684
+    },
+    {
+      "epoch": 0.00685,
+      "grad_norm": 0.5775508276061283,
+      "learning_rate": 0.0020550000000000004,
+      "loss": 4.9792,
+      "step": 685
+    },
+    {
+      "epoch": 0.00686,
+      "grad_norm": 0.5359292821422027,
+      "learning_rate": 0.0020580000000000004,
+      "loss": 4.9508,
+      "step": 686
+    },
+    {
+      "epoch": 0.00687,
+      "grad_norm": 0.5439874489161504,
+      "learning_rate": 0.0020610000000000003,
+      "loss": 4.9456,
+      "step": 687
+    },
+    {
+      "epoch": 0.00688,
+      "grad_norm": 0.6521058673701751,
+      "learning_rate": 0.002064,
+      "loss": 4.9472,
+      "step": 688
+    },
+    {
+      "epoch": 0.00689,
+      "grad_norm": 0.7201992210148584,
+      "learning_rate": 0.002067,
+      "loss": 4.9514,
+      "step": 689
+    },
+    {
+      "epoch": 0.0069,
+      "grad_norm": 0.6329359289093167,
+      "learning_rate": 0.00207,
+      "loss": 4.9393,
+      "step": 690
+    },
+    {
+      "epoch": 0.00691,
+      "grad_norm": 0.5452341441548267,
+      "learning_rate": 0.0020729999999999998,
+      "loss": 4.9065,
+      "step": 691
+    },
+    {
+      "epoch": 0.00692,
+      "grad_norm": 0.524421270821715,
+      "learning_rate": 0.0020759999999999997,
+      "loss": 4.9264,
+      "step": 692
+    },
+    {
+      "epoch": 0.00693,
+      "grad_norm": 0.5237198583423548,
+      "learning_rate": 0.0020789999999999997,
+      "loss": 4.9275,
+      "step": 693
+    },
+    {
+      "epoch": 0.00694,
+      "grad_norm": 0.5943284166970615,
+      "learning_rate": 0.002082,
+      "loss": 4.9065,
+      "step": 694
+    },
+    {
+      "epoch": 0.00695,
+      "grad_norm": 0.6331066584123409,
+      "learning_rate": 0.002085,
+      "loss": 4.9096,
+      "step": 695
+    },
+    {
+      "epoch": 0.00696,
+      "grad_norm": 0.6552141599381052,
+      "learning_rate": 0.002088,
+      "loss": 4.9103,
+      "step": 696
+    },
+    {
+      "epoch": 0.00697,
+      "grad_norm": 0.7459258980378775,
+      "learning_rate": 0.002091,
+      "loss": 4.9188,
+      "step": 697
+    },
+    {
+      "epoch": 0.00698,
+      "grad_norm": 0.902413176094811,
+      "learning_rate": 0.002094,
+      "loss": 4.9109,
+      "step": 698
+    },
+    {
+      "epoch": 0.00699,
+      "grad_norm": 0.9488326760888935,
+      "learning_rate": 0.002097,
+      "loss": 4.9116,
+      "step": 699
+    },
+    {
+      "epoch": 0.007,
+      "grad_norm": 0.9189763967394823,
+      "learning_rate": 0.0021,
+      "loss": 4.929,
+      "step": 700
+    },
+    {
+      "epoch": 0.00701,
+      "grad_norm": 0.8924436046521577,
+      "learning_rate": 0.002103,
+      "loss": 4.9083,
+      "step": 701
+    },
+    {
+      "epoch": 0.00702,
+      "grad_norm": 0.8863614629240012,
+      "learning_rate": 0.002106,
+      "loss": 4.9092,
+      "step": 702
+    },
+    {
+      "epoch": 0.00703,
+      "grad_norm": 0.7940709541538681,
+      "learning_rate": 0.0021089999999999998,
+      "loss": 4.891,
+      "step": 703
+    },
+    {
+      "epoch": 0.00704,
+      "grad_norm": 0.7939787736751149,
+      "learning_rate": 0.0021119999999999997,
+      "loss": 4.8964,
+      "step": 704
+    },
+    {
+      "epoch": 0.00705,
+      "grad_norm": 1.0444656030359551,
+      "learning_rate": 0.002115,
+      "loss": 4.9151,
+      "step": 705
+    },
+    {
+      "epoch": 0.00706,
+      "grad_norm": 1.1431311909042268,
+      "learning_rate": 0.002118,
+      "loss": 4.9234,
+      "step": 706
+    },
+    {
+      "epoch": 0.00707,
+      "grad_norm": 0.8384635334186645,
+      "learning_rate": 0.002121,
+      "loss": 4.8902,
+      "step": 707
+    },
+    {
+      "epoch": 0.00708,
+      "grad_norm": 0.9594405079672866,
+      "learning_rate": 0.002124,
+      "loss": 4.8745,
+      "step": 708
+    },
+    {
+      "epoch": 0.00709,
+      "grad_norm": 0.8900382869322284,
+      "learning_rate": 0.002127,
+      "loss": 4.9161,
+      "step": 709
+    },
+    {
+      "epoch": 0.0071,
+      "grad_norm": 0.8868570415283396,
+      "learning_rate": 0.00213,
+      "loss": 4.9065,
+      "step": 710
+    },
+    {
+      "epoch": 0.00711,
+      "grad_norm": 0.8610490944817158,
+      "learning_rate": 0.002133,
+      "loss": 4.8804,
+      "step": 711
+    },
+    {
+      "epoch": 0.00712,
+      "grad_norm": 0.8727164938852855,
+      "learning_rate": 0.002136,
+      "loss": 4.9046,
+      "step": 712
+    },
+    {
+      "epoch": 0.00713,
+      "grad_norm": 0.8857525308493206,
+      "learning_rate": 0.002139,
+      "loss": 4.9135,
+      "step": 713
+    },
+    {
+      "epoch": 0.00714,
+      "grad_norm": 0.9495661806955594,
+      "learning_rate": 0.002142,
+      "loss": 4.918,
+      "step": 714
+    },
+    {
+      "epoch": 0.00715,
+      "grad_norm": 1.0263024097609161,
+      "learning_rate": 0.0021449999999999998,
+      "loss": 4.8857,
+      "step": 715
+    },
+    {
+      "epoch": 0.00716,
+      "grad_norm": 0.8876358680026493,
+      "learning_rate": 0.002148,
+      "loss": 4.8749,
+      "step": 716
+    },
+    {
+      "epoch": 0.00717,
+      "grad_norm": 0.8225498605776377,
+      "learning_rate": 0.002151,
+      "loss": 4.8925,
+      "step": 717
+    },
+    {
+      "epoch": 0.00718,
+      "grad_norm": 0.628552485265691,
+      "learning_rate": 0.002154,
+      "loss": 4.8659,
+      "step": 718
+    },
+    {
+      "epoch": 0.00719,
+      "grad_norm": 0.6584104654465238,
+      "learning_rate": 0.002157,
+      "loss": 4.8747,
+      "step": 719
+    },
+    {
+      "epoch": 0.0072,
+      "grad_norm": 0.6698592474865601,
+      "learning_rate": 0.00216,
+      "loss": 4.8635,
+      "step": 720
+    },
+    {
+      "epoch": 0.00721,
+      "grad_norm": 0.6673590176314685,
+      "learning_rate": 0.002163,
+      "loss": 4.8639,
+      "step": 721
+    },
+    {
+      "epoch": 0.00722,
+      "grad_norm": 0.6674098225397388,
+      "learning_rate": 0.002166,
+      "loss": 4.8386,
+      "step": 722
+    },
+    {
+      "epoch": 0.00723,
+      "grad_norm": 0.6090726175552883,
+      "learning_rate": 0.002169,
+      "loss": 4.8464,
+      "step": 723
+    },
+    {
+      "epoch": 0.00724,
+      "grad_norm": 0.6325507361418539,
+      "learning_rate": 0.002172,
+      "loss": 4.8403,
+      "step": 724
+    },
+    {
+      "epoch": 0.00725,
+      "grad_norm": 0.6927587431932604,
+      "learning_rate": 0.002175,
+      "loss": 4.8341,
+      "step": 725
+    },
+    {
+      "epoch": 0.00726,
+      "grad_norm": 0.7422551683158218,
+      "learning_rate": 0.002178,
+      "loss": 4.8448,
+      "step": 726
+    },
+    {
+      "epoch": 0.00727,
+      "grad_norm": 0.7946686392459241,
+      "learning_rate": 0.0021809999999999998,
+      "loss": 4.8264,
+      "step": 727
+    },
+    {
+      "epoch": 0.00728,
+      "grad_norm": 0.651194780867581,
+      "learning_rate": 0.002184,
+      "loss": 4.8373,
+      "step": 728
+    },
+    {
+      "epoch": 0.00729,
+      "grad_norm": 0.5507866158426874,
+      "learning_rate": 0.002187,
+      "loss": 4.8279,
+      "step": 729
+    },
+    {
+      "epoch": 0.0073,
+      "grad_norm": 0.5770531279665235,
+      "learning_rate": 0.00219,
+      "loss": 4.8256,
+      "step": 730
+    },
+    {
+      "epoch": 0.00731,
+      "grad_norm": 0.6604772562967653,
+      "learning_rate": 0.002193,
+      "loss": 4.8198,
+      "step": 731
+    },
+    {
+      "epoch": 0.00732,
+      "grad_norm": 0.7902754963422924,
+      "learning_rate": 0.002196,
+      "loss": 4.832,
+      "step": 732
+    },
+    {
+      "epoch": 0.00733,
+      "grad_norm": 0.8530754775219535,
+      "learning_rate": 0.002199,
+      "loss": 4.814,
+      "step": 733
+    },
+    {
+      "epoch": 0.00734,
+      "grad_norm": 0.808835939559151,
+      "learning_rate": 0.002202,
+      "loss": 4.8365,
+      "step": 734
+    },
+    {
+      "epoch": 0.00735,
+      "grad_norm": 0.7793455630355729,
+      "learning_rate": 0.002205,
+      "loss": 4.8484,
+      "step": 735
+    },
+    {
+      "epoch": 0.00736,
+      "grad_norm": 0.8899384636665534,
+      "learning_rate": 0.002208,
+      "loss": 4.8322,
+      "step": 736
+    },
+    {
+      "epoch": 0.00737,
+      "grad_norm": 1.0166382867407526,
+      "learning_rate": 0.002211,
+      "loss": 4.8145,
+      "step": 737
+    },
+    {
+      "epoch": 0.00738,
+      "grad_norm": 0.9857126416807526,
+      "learning_rate": 0.002214,
+      "loss": 4.8213,
+      "step": 738
+    },
+    {
+      "epoch": 0.00739,
+      "grad_norm": 0.8982016923721464,
+      "learning_rate": 0.0022170000000000002,
+      "loss": 4.809,
+      "step": 739
+    },
+    {
+      "epoch": 0.0074,
+      "grad_norm": 0.8517808259905928,
+      "learning_rate": 0.00222,
+      "loss": 4.8138,
+      "step": 740
+    },
+    {
+      "epoch": 0.00741,
+      "grad_norm": 0.6571636698582773,
+      "learning_rate": 0.002223,
+      "loss": 4.778,
+      "step": 741
+    },
+    {
+      "epoch": 0.00742,
+      "grad_norm": 0.5983530041008951,
+      "learning_rate": 0.002226,
+      "loss": 4.8043,
+      "step": 742
+    },
+    {
+      "epoch": 0.00743,
+      "grad_norm": 0.613767022264535,
+      "learning_rate": 0.002229,
+      "loss": 4.772,
+      "step": 743
+    },
+    {
+      "epoch": 0.00744,
+      "grad_norm": 0.5746332772613801,
+      "learning_rate": 0.002232,
+      "loss": 4.7755,
+      "step": 744
+    },
+    {
+      "epoch": 0.00745,
+      "grad_norm": 0.5491916195482714,
+      "learning_rate": 0.002235,
+      "loss": 4.7792,
+      "step": 745
+    },
+    {
+      "epoch": 0.00746,
+      "grad_norm": 0.6224704559098753,
+      "learning_rate": 0.002238,
+      "loss": 4.7785,
+      "step": 746
+    },
+    {
+      "epoch": 0.00747,
+      "grad_norm": 0.721423929849636,
+      "learning_rate": 0.002241,
+      "loss": 4.766,
+      "step": 747
+    },
+    {
+      "epoch": 0.00748,
+      "grad_norm": 0.8211263973312402,
+      "learning_rate": 0.002244,
+      "loss": 4.7825,
+      "step": 748
+    },
+    {
+      "epoch": 0.00749,
+      "grad_norm": 0.9630311220772746,
+      "learning_rate": 0.002247,
+      "loss": 4.7687,
+      "step": 749
+    },
+    {
+      "epoch": 0.0075,
+      "grad_norm": 1.0671208539128567,
+      "learning_rate": 0.0022500000000000003,
+      "loss": 4.8201,
+      "step": 750
+    },
+    {
+      "epoch": 0.00751,
+      "grad_norm": 1.0285377005887373,
+      "learning_rate": 0.0022530000000000002,
+      "loss": 4.8053,
+      "step": 751
+    },
+    {
+      "epoch": 0.00752,
+      "grad_norm": 0.9937387135055332,
+      "learning_rate": 0.002256,
+      "loss": 4.7635,
+      "step": 752
+    },
+    {
+      "epoch": 0.00753,
+      "grad_norm": 0.8939099354397223,
+      "learning_rate": 0.002259,
+      "loss": 4.8048,
+      "step": 753
+    },
+    {
+      "epoch": 0.00754,
+      "grad_norm": 0.958657384547811,
+      "learning_rate": 0.002262,
+      "loss": 4.816,
+      "step": 754
+    },
+    {
+      "epoch": 0.00755,
+      "grad_norm": 0.8579165829387244,
+      "learning_rate": 0.002265,
+      "loss": 4.7809,
+      "step": 755
+    },
+    {
+      "epoch": 0.00756,
+      "grad_norm": 0.8147493051985796,
+      "learning_rate": 0.002268,
+      "loss": 4.7687,
+      "step": 756
+    },
+    {
+      "epoch": 0.00757,
+      "grad_norm": 0.9899674342411924,
+      "learning_rate": 0.002271,
+      "loss": 4.7845,
+      "step": 757
+    },
+    {
+      "epoch": 0.00758,
+      "grad_norm": 1.2327208130150207,
+      "learning_rate": 0.002274,
+      "loss": 4.7967,
+      "step": 758
+    },
+    {
+      "epoch": 0.00759,
+      "grad_norm": 0.8760473410677282,
+      "learning_rate": 0.002277,
+      "loss": 4.7951,
+      "step": 759
+    },
+    {
+      "epoch": 0.0076,
+      "grad_norm": 0.9553393765090987,
+      "learning_rate": 0.00228,
+      "loss": 4.7842,
+      "step": 760
+    },
+    {
+      "epoch": 0.00761,
+      "grad_norm": 1.088855928225056,
+      "learning_rate": 0.002283,
+      "loss": 4.769,
+      "step": 761
+    },
+    {
+      "epoch": 0.00762,
+      "grad_norm": 0.9818480683263884,
+      "learning_rate": 0.0022860000000000003,
+      "loss": 4.7512,
+      "step": 762
+    },
+    {
+      "epoch": 0.00763,
+      "grad_norm": 0.9731540924632093,
+      "learning_rate": 0.0022890000000000002,
+      "loss": 4.7931,
+      "step": 763
+    },
+    {
+      "epoch": 0.00764,
+      "grad_norm": 1.0508884929557651,
+      "learning_rate": 0.002292,
+      "loss": 4.8167,
+      "step": 764
+    },
+    {
+      "epoch": 0.00765,
+      "grad_norm": 1.0020863769727308,
+      "learning_rate": 0.002295,
+      "loss": 4.7984,
+      "step": 765
+    },
+    {
+      "epoch": 0.00766,
+      "grad_norm": 1.1527463652354557,
+      "learning_rate": 0.002298,
+      "loss": 4.8085,
+      "step": 766
+    },
+    {
+      "epoch": 0.00767,
+      "grad_norm": 0.9657952239159258,
+      "learning_rate": 0.002301,
+      "loss": 4.7959,
+      "step": 767
+    },
+    {
+      "epoch": 0.00768,
+      "grad_norm": 1.0234917976922082,
+      "learning_rate": 0.002304,
+      "loss": 4.8012,
+      "step": 768
+    },
+    {
+      "epoch": 0.00769,
+      "grad_norm": 0.9850893067060651,
+      "learning_rate": 0.002307,
+      "loss": 4.8144,
+      "step": 769
+    },
+    {
+      "epoch": 0.0077,
+      "grad_norm": 0.9062134932024389,
+      "learning_rate": 0.00231,
+      "loss": 4.7653,
+      "step": 770
+    },
+    {
+      "epoch": 0.00771,
+      "grad_norm": 0.8476285286232204,
+      "learning_rate": 0.002313,
+      "loss": 4.7979,
+      "step": 771
+    },
+    {
+      "epoch": 0.00772,
+      "grad_norm": 0.9122213123018311,
+      "learning_rate": 0.002316,
+      "loss": 4.7851,
+      "step": 772
+    },
+    {
+      "epoch": 0.00773,
+      "grad_norm": 1.0718910624781612,
+      "learning_rate": 0.0023190000000000003,
+      "loss": 4.8052,
+      "step": 773
+    },
+    {
+      "epoch": 0.00774,
+      "grad_norm": 0.7792131883523417,
+      "learning_rate": 0.0023220000000000003,
+      "loss": 4.7945,
+      "step": 774
+    },
+    {
+      "epoch": 0.00775,
+      "grad_norm": 0.7995411986928386,
+      "learning_rate": 0.0023250000000000002,
+      "loss": 4.7914,
+      "step": 775
+    },
+    {
+      "epoch": 0.00776,
+      "grad_norm": 0.7054590225014301,
+      "learning_rate": 0.002328,
+      "loss": 4.7883,
+      "step": 776
+    },
+    {
+      "epoch": 0.00777,
+      "grad_norm": 0.6505869359405926,
+      "learning_rate": 0.002331,
+      "loss": 4.7585,
+      "step": 777
+    },
+    {
+      "epoch": 0.00778,
+      "grad_norm": 0.6484695284206986,
+      "learning_rate": 0.002334,
+      "loss": 4.7652,
+      "step": 778
+    },
+    {
+      "epoch": 0.00779,
+      "grad_norm": 0.6047799586124271,
+      "learning_rate": 0.002337,
+      "loss": 4.7239,
+      "step": 779
+    },
+    {
+      "epoch": 0.0078,
+      "grad_norm": 0.5436502526586032,
+      "learning_rate": 0.00234,
+      "loss": 4.7364,
+      "step": 780
+    },
+    {
+      "epoch": 0.00781,
+      "grad_norm": 0.5682167623371829,
+      "learning_rate": 0.002343,
+      "loss": 4.7355,
+      "step": 781
+    },
+    {
+      "epoch": 0.00782,
+      "grad_norm": 0.628910610744215,
+      "learning_rate": 0.002346,
+      "loss": 4.7356,
+      "step": 782
+    },
+    {
+      "epoch": 0.00783,
+      "grad_norm": 0.5845457532196663,
+      "learning_rate": 0.002349,
+      "loss": 4.7086,
+      "step": 783
+    },
+    {
+      "epoch": 0.00784,
+      "grad_norm": 0.6345965495249546,
+      "learning_rate": 0.002352,
+      "loss": 4.723,
+      "step": 784
+    },
+    {
+      "epoch": 0.00785,
+      "grad_norm": 0.4959017064429773,
+      "learning_rate": 0.0023550000000000003,
+      "loss": 4.7138,
+      "step": 785
+    },
+    {
+      "epoch": 0.00786,
+      "grad_norm": 0.4358915945164792,
+      "learning_rate": 0.0023580000000000003,
+      "loss": 4.7131,
+      "step": 786
+    },
+    {
+      "epoch": 0.00787,
+      "grad_norm": 0.4259891850743534,
+      "learning_rate": 0.0023610000000000003,
+      "loss": 4.7113,
+      "step": 787
+    },
+    {
+      "epoch": 0.00788,
+      "grad_norm": 0.4529949912379573,
+      "learning_rate": 0.002364,
+      "loss": 4.7077,
+      "step": 788
+    },
+    {
+      "epoch": 0.00789,
+      "grad_norm": 0.4944753699230628,
+      "learning_rate": 0.002367,
+      "loss": 4.7334,
+      "step": 789
+    },
+    {
+      "epoch": 0.0079,
+      "grad_norm": 0.6295031827770177,
+      "learning_rate": 0.00237,
+      "loss": 4.7144,
+      "step": 790
+    },
+    {
+      "epoch": 0.00791,
+      "grad_norm": 0.8474014146600959,
+      "learning_rate": 0.002373,
+      "loss": 4.6777,
+      "step": 791
+    },
+    {
+      "epoch": 0.00792,
+      "grad_norm": 0.8040402155973354,
+      "learning_rate": 0.002376,
+      "loss": 4.7267,
+      "step": 792
+    },
+    {
+      "epoch": 0.00793,
+      "grad_norm": 0.5568568781614048,
+      "learning_rate": 0.002379,
+      "loss": 4.7006,
+      "step": 793
+    },
+    {
+      "epoch": 0.00794,
+      "grad_norm": 0.8220014797505664,
+      "learning_rate": 0.002382,
+      "loss": 4.7246,
+      "step": 794
+    },
+    {
+      "epoch": 0.00795,
+      "grad_norm": 0.6551332917875898,
+      "learning_rate": 0.002385,
+      "loss": 4.6938,
+      "step": 795
+    },
+    {
+      "epoch": 0.00796,
+      "grad_norm": 0.5666814801389223,
+      "learning_rate": 0.0023880000000000004,
+      "loss": 4.6826,
+      "step": 796
+    },
+    {
+      "epoch": 0.00797,
+      "grad_norm": 0.654430341218369,
+      "learning_rate": 0.0023910000000000003,
+      "loss": 4.7248,
+      "step": 797
+    },
+    {
+      "epoch": 0.00798,
+      "grad_norm": 0.5511512463730408,
+      "learning_rate": 0.0023940000000000003,
+      "loss": 4.6864,
+      "step": 798
+    },
+    {
+      "epoch": 0.00799,
+      "grad_norm": 0.5084174359945534,
+      "learning_rate": 0.0023970000000000003,
+      "loss": 4.7161,
+      "step": 799
+    },
+    {
+      "epoch": 0.008,
+      "grad_norm": 0.41968398472595075,
+      "learning_rate": 0.0024000000000000002,
+      "loss": 4.6652,
+      "step": 800
+    },
+    {
+      "epoch": 0.00801,
+      "grad_norm": 0.4546309911468048,
+      "learning_rate": 0.002403,
+      "loss": 4.6967,
+      "step": 801
+    },
+    {
+      "epoch": 0.00802,
+      "grad_norm": 0.430018994115786,
+      "learning_rate": 0.002406,
+      "loss": 4.6827,
+      "step": 802
+    },
+    {
+      "epoch": 0.00803,
+      "grad_norm": 0.44614830716967085,
+      "learning_rate": 0.002409,
+      "loss": 4.6617,
+      "step": 803
+    },
+    {
+      "epoch": 0.00804,
+      "grad_norm": 0.44722400844593674,
+      "learning_rate": 0.002412,
+      "loss": 4.6598,
+      "step": 804
+    },
+    {
+      "epoch": 0.00805,
+      "grad_norm": 0.5179127215582825,
+      "learning_rate": 0.002415,
+      "loss": 4.6599,
+      "step": 805
+    },
+    {
+      "epoch": 0.00806,
+      "grad_norm": 0.5610832008078775,
+      "learning_rate": 0.002418,
+      "loss": 4.677,
+      "step": 806
+    },
+    {
+      "epoch": 0.00807,
+      "grad_norm": 0.5167453223410896,
+      "learning_rate": 0.0024210000000000004,
+      "loss": 4.6671,
+      "step": 807
+    },
+    {
+      "epoch": 0.00808,
+      "grad_norm": 0.46468933196331563,
+      "learning_rate": 0.0024240000000000004,
+      "loss": 4.6511,
+      "step": 808
+    },
+    {
+      "epoch": 0.00809,
+      "grad_norm": 0.5221883532574668,
+      "learning_rate": 0.0024270000000000003,
+      "loss": 4.6468,
+      "step": 809
+    },
+    {
+      "epoch": 0.0081,
+      "grad_norm": 0.4992566900849729,
+      "learning_rate": 0.0024300000000000003,
+      "loss": 4.6744,
+      "step": 810
+    },
+    {
+      "epoch": 0.00811,
+      "grad_norm": 0.4854147467055134,
+      "learning_rate": 0.0024330000000000003,
+      "loss": 4.646,
+      "step": 811
+    },
+    {
+      "epoch": 0.00812,
+      "grad_norm": 0.650970729431075,
+      "learning_rate": 0.0024360000000000002,
+      "loss": 4.6307,
+      "step": 812
+    },
+    {
+      "epoch": 0.00813,
+      "grad_norm": 0.8160691589494683,
+      "learning_rate": 0.0024389999999999998,
+      "loss": 4.6711,
+      "step": 813
+    },
+    {
+      "epoch": 0.00814,
+      "grad_norm": 0.9918101747931352,
+      "learning_rate": 0.0024419999999999997,
+      "loss": 4.6946,
+      "step": 814
+    },
+    {
+      "epoch": 0.00815,
+      "grad_norm": 1.247963175893729,
+      "learning_rate": 0.0024449999999999997,
+      "loss": 4.7226,
+      "step": 815
+    },
+    {
+      "epoch": 0.00816,
+      "grad_norm": 0.8376200515557375,
+      "learning_rate": 0.002448,
+      "loss": 4.6777,
+      "step": 816
+    },
+    {
+      "epoch": 0.00817,
+      "grad_norm": 0.9161032619759178,
+      "learning_rate": 0.002451,
+      "loss": 4.6939,
+      "step": 817
+    },
+    {
+      "epoch": 0.00818,
+      "grad_norm": 1.0914649908014256,
+      "learning_rate": 0.002454,
+      "loss": 4.6886,
+      "step": 818
+    },
+    {
+      "epoch": 0.00819,
+      "grad_norm": 0.9806171410774952,
+      "learning_rate": 0.002457,
+      "loss": 4.712,
+      "step": 819
+    },
+    {
+      "epoch": 0.0082,
+      "grad_norm": 0.992236077471004,
+      "learning_rate": 0.00246,
+      "loss": 4.6918,
+      "step": 820
+    },
+    {
+      "epoch": 0.00821,
+      "grad_norm": 1.0594557870263281,
+      "learning_rate": 0.002463,
+      "loss": 4.6759,
+      "step": 821
+    },
+    {
+      "epoch": 0.00822,
+      "grad_norm": 1.0346800919438124,
+      "learning_rate": 0.002466,
+      "loss": 4.6853,
+      "step": 822
+    },
+    {
+      "epoch": 0.00823,
+      "grad_norm": 0.9573573191186882,
+      "learning_rate": 0.002469,
+      "loss": 4.6833,
+      "step": 823
+    },
+    {
+      "epoch": 0.00824,
+      "grad_norm": 1.1123514933123841,
+      "learning_rate": 0.002472,
+      "loss": 4.714,
+      "step": 824
+    },
+    {
+      "epoch": 0.00825,
+      "grad_norm": 0.8463845700248506,
+      "learning_rate": 0.0024749999999999998,
+      "loss": 4.7191,
+      "step": 825
+    },
+    {
+      "epoch": 0.00826,
+      "grad_norm": 0.8444785606085857,
+      "learning_rate": 0.0024779999999999997,
+      "loss": 4.672,
+      "step": 826
+    },
+    {
+      "epoch": 0.00827,
+      "grad_norm": 0.9726341870117121,
+      "learning_rate": 0.002481,
+      "loss": 4.7078,
+      "step": 827
+    },
+    {
+      "epoch": 0.00828,
+      "grad_norm": 0.9106448417621353,
+      "learning_rate": 0.002484,
+      "loss": 4.7003,
+      "step": 828
+    },
+    {
+      "epoch": 0.00829,
+      "grad_norm": 0.7565680418878746,
+      "learning_rate": 0.002487,
+      "loss": 4.6856,
+      "step": 829
+    },
+    {
+      "epoch": 0.0083,
+      "grad_norm": 0.8537774465977133,
+      "learning_rate": 0.00249,
+      "loss": 4.7017,
+      "step": 830
+    },
+    {
+      "epoch": 0.00831,
+      "grad_norm": 0.9023323948099834,
+      "learning_rate": 0.002493,
+      "loss": 4.6871,
+      "step": 831
+    },
+    {
+      "epoch": 0.00832,
+      "grad_norm": 0.8524529451127855,
+      "learning_rate": 0.002496,
+      "loss": 4.6815,
+      "step": 832
+    },
+    {
+      "epoch": 0.00833,
+      "grad_norm": 0.9428655185832147,
+      "learning_rate": 0.002499,
+      "loss": 4.6808,
+      "step": 833
+    },
+    {
+      "epoch": 0.00834,
+      "grad_norm": 0.9597220185428569,
+      "learning_rate": 0.002502,
+      "loss": 4.6879,
+      "step": 834
+    },
+    {
+      "epoch": 0.00835,
+      "grad_norm": 0.7735101632354252,
+      "learning_rate": 0.002505,
+      "loss": 4.6724,
+      "step": 835
+    },
+    {
+      "epoch": 0.00836,
+      "grad_norm": 0.9597202731139803,
+      "learning_rate": 0.002508,
+      "loss": 4.7061,
+      "step": 836
+    },
+    {
+      "epoch": 0.00837,
+      "grad_norm": 0.9520863539431935,
+      "learning_rate": 0.0025109999999999998,
+      "loss": 4.6636,
+      "step": 837
+    },
+    {
+      "epoch": 0.00838,
+      "grad_norm": 0.7800128524395746,
+      "learning_rate": 0.0025139999999999997,
+      "loss": 4.6721,
+      "step": 838
+    },
+    {
+      "epoch": 0.00839,
+      "grad_norm": 0.8122589832425033,
+      "learning_rate": 0.002517,
+      "loss": 4.675,
+      "step": 839
+    },
+    {
+      "epoch": 0.0084,
+      "grad_norm": 0.8183344402395425,
+      "learning_rate": 0.00252,
+      "loss": 4.6669,
+      "step": 840
+    },
+    {
+      "epoch": 0.00841,
+      "grad_norm": 0.6918735110390536,
+      "learning_rate": 0.002523,
+      "loss": 4.6489,
+      "step": 841
+    },
+    {
+      "epoch": 0.00842,
+      "grad_norm": 0.6201385747244391,
+      "learning_rate": 0.002526,
+      "loss": 4.6423,
+      "step": 842
+    },
+    {
+      "epoch": 0.00843,
+      "grad_norm": 0.606127970479136,
+      "learning_rate": 0.002529,
+      "loss": 4.6465,
+      "step": 843
+    },
+    {
+      "epoch": 0.00844,
+      "grad_norm": 0.5515773209874846,
+      "learning_rate": 0.002532,
+      "loss": 4.6607,
+      "step": 844
+    },
+    {
+      "epoch": 0.00845,
+      "grad_norm": 0.6203742299859808,
+      "learning_rate": 0.002535,
+      "loss": 4.6293,
+      "step": 845
+    },
+    {
+      "epoch": 0.00846,
+      "grad_norm": 0.5875832865020281,
+      "learning_rate": 0.002538,
+      "loss": 4.6474,
+      "step": 846
+    },
+    {
+      "epoch": 0.00847,
+      "grad_norm": 0.5703256353430879,
+      "learning_rate": 0.002541,
+      "loss": 4.6282,
+      "step": 847
+    },
+    {
+      "epoch": 0.00848,
+      "grad_norm": 0.602830367643936,
+      "learning_rate": 0.002544,
+      "loss": 4.6269,
+      "step": 848
+    },
+    {
+      "epoch": 0.00849,
+      "grad_norm": 0.6741507039909044,
+      "learning_rate": 0.002547,
+      "loss": 4.6233,
+      "step": 849
+    },
+    {
+      "epoch": 0.0085,
+      "grad_norm": 0.6288739006540759,
+      "learning_rate": 0.00255,
+      "loss": 4.6341,
+      "step": 850
+    },
+    {
+      "epoch": 0.00851,
+      "grad_norm": 0.5820099008678455,
+      "learning_rate": 0.002553,
+      "loss": 4.644,
+      "step": 851
+    },
+    {
+      "epoch": 0.00852,
+      "grad_norm": 0.586123912558797,
+      "learning_rate": 0.002556,
+      "loss": 4.6367,
+      "step": 852
+    },
+    {
+      "epoch": 0.00853,
+      "grad_norm": 0.5127813487098001,
+      "learning_rate": 0.002559,
+      "loss": 4.6085,
+      "step": 853
+    },
+    {
+      "epoch": 0.00854,
+      "grad_norm": 0.4730499644759358,
+      "learning_rate": 0.002562,
+      "loss": 4.6029,
+      "step": 854
+    },
+    {
+      "epoch": 0.00855,
+      "grad_norm": 0.44708869980986227,
+      "learning_rate": 0.002565,
+      "loss": 4.5799,
+      "step": 855
+    },
+    {
+      "epoch": 0.00856,
+      "grad_norm": 0.466044480858233,
+      "learning_rate": 0.002568,
+      "loss": 4.6142,
+      "step": 856
+    },
+    {
+      "epoch": 0.00857,
+      "grad_norm": 0.5382201915945353,
+      "learning_rate": 0.002571,
+      "loss": 4.6036,
+      "step": 857
+    },
+    {
+      "epoch": 0.00858,
+      "grad_norm": 0.6780662034295477,
+      "learning_rate": 0.002574,
+      "loss": 4.609,
+      "step": 858
+    },
+    {
+      "epoch": 0.00859,
+      "grad_norm": 0.9086610382483981,
+      "learning_rate": 0.002577,
+      "loss": 4.6039,
+      "step": 859
+    },
+    {
+      "epoch": 0.0086,
+      "grad_norm": 0.8563688949272525,
+      "learning_rate": 0.00258,
+      "loss": 4.6531,
+      "step": 860
+    },
+    {
+      "epoch": 0.00861,
+      "grad_norm": 0.5965670098126366,
+      "learning_rate": 0.0025830000000000002,
+      "loss": 4.6461,
+      "step": 861
+    },
+    {
+      "epoch": 0.00862,
+      "grad_norm": 0.7975052365958228,
+      "learning_rate": 0.002586,
+      "loss": 4.611,
+      "step": 862
+    },
+    {
+      "epoch": 0.00863,
+      "grad_norm": 0.650099032572018,
+      "learning_rate": 0.002589,
+      "loss": 4.6284,
+      "step": 863
+    },
+    {
+      "epoch": 0.00864,
+      "grad_norm": 0.6277114763068243,
+      "learning_rate": 0.002592,
+      "loss": 4.5809,
+      "step": 864
+    },
+    {
+      "epoch": 0.00865,
+      "grad_norm": 0.7499269309750987,
+      "learning_rate": 0.002595,
+      "loss": 4.6024,
+      "step": 865
+    },
+    {
+      "epoch": 0.00866,
+      "grad_norm": 0.7085973518403954,
+      "learning_rate": 0.002598,
+      "loss": 4.6233,
+      "step": 866
+    },
+    {
+      "epoch": 0.00867,
+      "grad_norm": 0.605538925445329,
+      "learning_rate": 0.002601,
+      "loss": 4.6101,
+      "step": 867
+    },
+    {
+      "epoch": 0.00868,
+      "grad_norm": 0.48200776305054654,
+      "learning_rate": 0.002604,
+      "loss": 4.586,
+      "step": 868
+    },
+    {
+      "epoch": 0.00869,
+      "grad_norm": 0.5266950965425763,
+      "learning_rate": 0.002607,
+      "loss": 4.5966,
+      "step": 869
+    },
+    {
+      "epoch": 0.0087,
+      "grad_norm": 0.48953699231607295,
+      "learning_rate": 0.00261,
+      "loss": 4.5832,
+      "step": 870
+    },
+    {
+      "epoch": 0.00871,
+      "grad_norm": 0.5478274928438833,
+      "learning_rate": 0.002613,
+      "loss": 4.5992,
+      "step": 871
+    },
+    {
+      "epoch": 0.00872,
+      "grad_norm": 0.6263670684952429,
+      "learning_rate": 0.002616,
+      "loss": 4.5864,
+      "step": 872
+    },
+    {
+      "epoch": 0.00873,
+      "grad_norm": 0.6672951132542,
+      "learning_rate": 0.0026190000000000002,
+      "loss": 4.5977,
+      "step": 873
+    },
+    {
+      "epoch": 0.00874,
+      "grad_norm": 0.677096167715366,
+      "learning_rate": 0.002622,
+      "loss": 4.617,
+      "step": 874
+    },
+    {
+      "epoch": 0.00875,
+      "grad_norm": 0.6959913524482387,
+      "learning_rate": 0.002625,
+      "loss": 4.5696,
+      "step": 875
+    },
+    {
+      "epoch": 0.00876,
+      "grad_norm": 0.6682762743495083,
+      "learning_rate": 0.002628,
+      "loss": 4.5958,
+      "step": 876
+    },
+    {
+      "epoch": 0.00877,
+      "grad_norm": 0.6375266502117092,
+      "learning_rate": 0.002631,
+      "loss": 4.612,
+      "step": 877
+    },
+    {
+      "epoch": 0.00878,
+      "grad_norm": 0.7079418482290942,
+      "learning_rate": 0.002634,
+      "loss": 4.5486,
+      "step": 878
+    },
+    {
+      "epoch": 0.00879,
+      "grad_norm": 0.6282689223941402,
+      "learning_rate": 0.002637,
+      "loss": 4.578,
+      "step": 879
+    },
+    {
+      "epoch": 0.0088,
+      "grad_norm": 0.5460943252882049,
+      "learning_rate": 0.00264,
+      "loss": 4.5852,
+      "step": 880
+    },
+    {
+      "epoch": 0.00881,
+      "grad_norm": 0.5723972494402886,
+      "learning_rate": 0.002643,
+      "loss": 4.5869,
+      "step": 881
+    },
+    {
+      "epoch": 0.00882,
+      "grad_norm": 0.647818443655113,
+      "learning_rate": 0.002646,
+      "loss": 4.588,
+      "step": 882
+    },
+    {
+      "epoch": 0.00883,
+      "grad_norm": 0.8827068805337381,
+      "learning_rate": 0.002649,
+      "loss": 4.5935,
+      "step": 883
+    },
+    {
+      "epoch": 0.00884,
+      "grad_norm": 1.3000201706023533,
+      "learning_rate": 0.0026520000000000003,
+      "loss": 4.6052,
+      "step": 884
+    },
+    {
+      "epoch": 0.00885,
+      "grad_norm": 0.7527768384442359,
+      "learning_rate": 0.0026550000000000002,
+      "loss": 4.5797,
+      "step": 885
+    },
+    {
+      "epoch": 0.00886,
+      "grad_norm": 0.7313595200920677,
+      "learning_rate": 0.002658,
+      "loss": 4.6019,
+      "step": 886
+    },
+    {
+      "epoch": 0.00887,
+      "grad_norm": 0.553327654847044,
+      "learning_rate": 0.002661,
+      "loss": 4.5828,
+      "step": 887
+    },
+    {
+      "epoch": 0.00888,
+      "grad_norm": 0.6064219625843388,
+      "learning_rate": 0.002664,
+      "loss": 4.5894,
+      "step": 888
+    },
+    {
+      "epoch": 0.00889,
+      "grad_norm": 0.6392357596846293,
+      "learning_rate": 0.002667,
+      "loss": 4.5422,
+      "step": 889
+    },
+    {
+      "epoch": 0.0089,
+      "grad_norm": 0.6860123914477424,
+      "learning_rate": 0.00267,
+      "loss": 4.5989,
+      "step": 890
+    },
+    {
+      "epoch": 0.00891,
+      "grad_norm": 0.7088960904364014,
+      "learning_rate": 0.002673,
+      "loss": 4.5822,
+      "step": 891
+    },
+    {
+      "epoch": 0.00892,
+      "grad_norm": 0.7157207147763361,
+      "learning_rate": 0.002676,
+      "loss": 4.5934,
+      "step": 892
+    },
+    {
+      "epoch": 0.00893,
+      "grad_norm": 0.7412527752908875,
+      "learning_rate": 0.002679,
+      "loss": 4.5709,
+      "step": 893
+    },
+    {
+      "epoch": 0.00894,
+      "grad_norm": 0.8084836835989728,
+      "learning_rate": 0.002682,
+      "loss": 4.5639,
+      "step": 894
+    },
+    {
+      "epoch": 0.00895,
+      "grad_norm": 0.9923307111818513,
+      "learning_rate": 0.0026850000000000003,
+      "loss": 4.5864,
+      "step": 895
+    },
+    {
+      "epoch": 0.00896,
+      "grad_norm": 1.2171682577354312,
+      "learning_rate": 0.0026880000000000003,
+      "loss": 4.6057,
+      "step": 896
+    },
+    {
+      "epoch": 0.00897,
+      "grad_norm": 0.797478427208377,
+      "learning_rate": 0.0026910000000000002,
+      "loss": 4.5989,
+      "step": 897
+    },
+    {
+      "epoch": 0.00898,
+      "grad_norm": 0.7928728804117916,
+      "learning_rate": 0.002694,
+      "loss": 4.594,
+      "step": 898
+    },
+    {
+      "epoch": 0.00899,
+      "grad_norm": 0.8357403035452178,
+      "learning_rate": 0.002697,
+      "loss": 4.5983,
+      "step": 899
+    },
+    {
+      "epoch": 0.009,
+      "grad_norm": 0.8448290091163538,
+      "learning_rate": 0.0027,
+      "loss": 4.6292,
+      "step": 900
+    },
+    {
+      "epoch": 0.00901,
+      "grad_norm": 0.9488092229670547,
+      "learning_rate": 0.002703,
+      "loss": 4.5868,
+      "step": 901
+    },
+    {
+      "epoch": 0.00902,
+      "grad_norm": 0.9434404658743749,
+      "learning_rate": 0.002706,
+      "loss": 4.5999,
+      "step": 902
+    },
+    {
+      "epoch": 0.00903,
+      "grad_norm": 1.0122099567822476,
+      "learning_rate": 0.002709,
+      "loss": 4.6102,
+      "step": 903
+    },
+    {
+      "epoch": 0.00904,
+      "grad_norm": 0.9358691681287052,
+      "learning_rate": 0.002712,
+      "loss": 4.5848,
+      "step": 904
+    },
+    {
+      "epoch": 0.00905,
+      "grad_norm": 0.8321510442485943,
+      "learning_rate": 0.002715,
+      "loss": 4.5984,
+      "step": 905
+    },
+    {
+      "epoch": 0.00906,
+      "grad_norm": 0.8914473393947665,
+      "learning_rate": 0.002718,
+      "loss": 4.6112,
+      "step": 906
+    },
+    {
+      "epoch": 0.00907,
+      "grad_norm": 0.9883982303638487,
+      "learning_rate": 0.0027210000000000003,
+      "loss": 4.6386,
+      "step": 907
+    },
+    {
+      "epoch": 0.00908,
+      "grad_norm": 0.86073203349026,
+      "learning_rate": 0.0027240000000000003,
+      "loss": 4.6116,
+      "step": 908
+    },
+    {
+      "epoch": 0.00909,
+      "grad_norm": 0.7773747412069614,
+      "learning_rate": 0.0027270000000000003,
+      "loss": 4.6163,
+      "step": 909
+    },
+    {
+      "epoch": 0.0091,
+      "grad_norm": 0.7370585718531062,
+      "learning_rate": 0.0027300000000000002,
+      "loss": 4.6234,
+      "step": 910
+    },
+    {
+      "epoch": 0.00911,
+      "grad_norm": 0.6906269071273593,
+      "learning_rate": 0.002733,
+      "loss": 4.5785,
+      "step": 911
+    },
+    {
+      "epoch": 0.00912,
+      "grad_norm": 0.6578032292778252,
+      "learning_rate": 0.002736,
+      "loss": 4.5778,
+      "step": 912
+    },
+    {
+      "epoch": 0.00913,
+      "grad_norm": 0.6528626059582382,
+      "learning_rate": 0.002739,
+      "loss": 4.5704,
+      "step": 913
+    },
+    {
+      "epoch": 0.00914,
+      "grad_norm": 0.599731896856576,
+      "learning_rate": 0.002742,
+      "loss": 4.595,
+      "step": 914
+    },
+    {
+      "epoch": 0.00915,
+      "grad_norm": 0.5922054086035364,
+      "learning_rate": 0.002745,
+      "loss": 4.5555,
+      "step": 915
+    },
+    {
+      "epoch": 0.00916,
+      "grad_norm": 0.528646140228931,
+      "learning_rate": 0.002748,
+      "loss": 4.5304,
+      "step": 916
+    },
+    {
+      "epoch": 0.00917,
+      "grad_norm": 0.5305158198561161,
+      "learning_rate": 0.002751,
+      "loss": 4.5419,
+      "step": 917
+    },
+    {
+      "epoch": 0.00918,
+      "grad_norm": 0.4736382884122071,
+      "learning_rate": 0.0027540000000000004,
+      "loss": 4.5569,
+      "step": 918
+    },
+    {
+      "epoch": 0.00919,
+      "grad_norm": 0.45838817911808083,
+      "learning_rate": 0.0027570000000000003,
+      "loss": 4.5357,
+      "step": 919
+    },
+    {
+      "epoch": 0.0092,
+      "grad_norm": 0.4361472184478695,
+      "learning_rate": 0.0027600000000000003,
+      "loss": 4.5493,
+      "step": 920
+    },
+    {
+      "epoch": 0.00921,
+      "grad_norm": 0.43931126757575867,
+      "learning_rate": 0.0027630000000000003,
+      "loss": 4.5336,
+      "step": 921
+    },
+    {
+      "epoch": 0.00922,
+      "grad_norm": 0.46920285200840567,
+      "learning_rate": 0.0027660000000000002,
+      "loss": 4.5412,
+      "step": 922
+    },
+    {
+      "epoch": 0.00923,
+      "grad_norm": 0.5996209745093354,
+      "learning_rate": 0.002769,
+      "loss": 4.5218,
+      "step": 923
+    },
+    {
+      "epoch": 0.00924,
+      "grad_norm": 0.702528530852281,
+      "learning_rate": 0.002772,
+      "loss": 4.5564,
+      "step": 924
+    },
+    {
+      "epoch": 0.00925,
+      "grad_norm": 0.6651779892024297,
+      "learning_rate": 0.002775,
+      "loss": 4.5288,
+      "step": 925
+    },
+    {
+      "epoch": 0.00926,
+      "grad_norm": 0.5723855300218565,
+      "learning_rate": 0.002778,
+      "loss": 4.5426,
+      "step": 926
+    },
+    {
+      "epoch": 0.00927,
+      "grad_norm": 0.6080148945846373,
+      "learning_rate": 0.002781,
+      "loss": 4.5043,
+      "step": 927
+    },
+    {
+      "epoch": 0.00928,
+      "grad_norm": 0.6957035289314476,
+      "learning_rate": 0.002784,
+      "loss": 4.5502,
+      "step": 928
+    },
+    {
+      "epoch": 0.00929,
+      "grad_norm": 0.7366634335117053,
+      "learning_rate": 0.0027870000000000004,
+      "loss": 4.5403,
+      "step": 929
+    },
+    {
+      "epoch": 0.0093,
+      "grad_norm": 0.7796102692352119,
+      "learning_rate": 0.0027900000000000004,
+      "loss": 4.5265,
+      "step": 930
+    },
+    {
+      "epoch": 0.00931,
+      "grad_norm": 0.7116909602884058,
+      "learning_rate": 0.0027930000000000003,
+      "loss": 4.5333,
+      "step": 931
+    },
+    {
+      "epoch": 0.00932,
+      "grad_norm": 0.8319854024158545,
+      "learning_rate": 0.0027960000000000003,
+      "loss": 4.5481,
+      "step": 932
+    },
+    {
+      "epoch": 0.00933,
+      "grad_norm": 0.8001418279766108,
+      "learning_rate": 0.0027990000000000003,
+      "loss": 4.5286,
+      "step": 933
+    },
+    {
+      "epoch": 0.00934,
+      "grad_norm": 0.6519619232143173,
+      "learning_rate": 0.0028020000000000002,
+      "loss": 4.5422,
+      "step": 934
+    },
+    {
+      "epoch": 0.00935,
+      "grad_norm": 0.7580737482550882,
+      "learning_rate": 0.002805,
+      "loss": 4.5559,
+      "step": 935
+    },
+    {
+      "epoch": 0.00936,
+      "grad_norm": 0.8499038627491867,
+      "learning_rate": 0.002808,
+      "loss": 4.5679,
+      "step": 936
+    },
+    {
+      "epoch": 0.00937,
+      "grad_norm": 0.7597302495348821,
+      "learning_rate": 0.002811,
+      "loss": 4.5708,
+      "step": 937
+    },
+    {
+      "epoch": 0.00938,
+      "grad_norm": 0.9878821641788273,
+      "learning_rate": 0.002814,
+      "loss": 4.5609,
+      "step": 938
+    },
+    {
+      "epoch": 0.00939,
+      "grad_norm": 0.9691729918808772,
+      "learning_rate": 0.002817,
+      "loss": 4.563,
+      "step": 939
+    },
+    {
+      "epoch": 0.0094,
+      "grad_norm": 0.8937843559478598,
+      "learning_rate": 0.00282,
+      "loss": 4.56,
+      "step": 940
+    },
+    {
+      "epoch": 0.00941,
+      "grad_norm": 0.9477839045288606,
+      "learning_rate": 0.002823,
+      "loss": 4.5409,
+      "step": 941
+    },
+    {
+      "epoch": 0.00942,
+      "grad_norm": 1.088560613057821,
+      "learning_rate": 0.002826,
+      "loss": 4.5819,
+      "step": 942
+    },
+    {
+      "epoch": 0.00943,
+      "grad_norm": 0.8020128186220904,
+      "learning_rate": 0.002829,
+      "loss": 4.556,
+      "step": 943
+    },
+    {
+      "epoch": 0.00944,
+      "grad_norm": 0.7970499406732843,
+      "learning_rate": 0.002832,
+      "loss": 4.5652,
+      "step": 944
+    },
+    {
+      "epoch": 0.00945,
+      "grad_norm": 0.760430287307007,
+      "learning_rate": 0.002835,
+      "loss": 4.567,
+      "step": 945
+    },
+    {
+      "epoch": 0.00946,
+      "grad_norm": 0.8410168172764453,
+      "learning_rate": 0.002838,
+      "loss": 4.5808,
+      "step": 946
+    },
+    {
+      "epoch": 0.00947,
+      "grad_norm": 0.8502364092306604,
+      "learning_rate": 0.0028409999999999998,
+      "loss": 4.5581,
+      "step": 947
+    },
+    {
+      "epoch": 0.00948,
+      "grad_norm": 0.7534324730199542,
+      "learning_rate": 0.0028439999999999997,
+      "loss": 4.533,
+      "step": 948
+    },
+    {
+      "epoch": 0.00949,
+      "grad_norm": 0.8075715283027973,
+      "learning_rate": 0.002847,
+      "loss": 4.5789,
+      "step": 949
+    },
+    {
+      "epoch": 0.0095,
+      "grad_norm": 0.8790685187514339,
+      "learning_rate": 0.00285,
+      "loss": 4.5764,
+      "step": 950
+    },
+    {
+      "epoch": 0.00951,
+      "grad_norm": 0.8527621336415785,
+      "learning_rate": 0.002853,
+      "loss": 4.552,
+      "step": 951
+    },
+    {
+      "epoch": 0.00952,
+      "grad_norm": 0.793648162843131,
+      "learning_rate": 0.002856,
+      "loss": 4.5771,
+      "step": 952
+    },
+    {
+      "epoch": 0.00953,
+      "grad_norm": 0.7100823051409002,
+      "learning_rate": 0.002859,
+      "loss": 4.5151,
+      "step": 953
+    },
+    {
+      "epoch": 0.00954,
+      "grad_norm": 0.776086010454581,
+      "learning_rate": 0.002862,
+      "loss": 4.5748,
+      "step": 954
+    },
+    {
+      "epoch": 0.00955,
+      "grad_norm": 0.7357834745016256,
+      "learning_rate": 0.002865,
+      "loss": 4.5651,
+      "step": 955
+    },
+    {
+      "epoch": 0.00956,
+      "grad_norm": 0.6871788604084053,
+      "learning_rate": 0.002868,
+      "loss": 4.5378,
+      "step": 956
+    },
+    {
+      "epoch": 0.00957,
+      "grad_norm": 0.6293704920093642,
+      "learning_rate": 0.002871,
+      "loss": 4.5585,
+      "step": 957
+    },
+    {
+      "epoch": 0.00958,
+      "grad_norm": 0.6933721545151298,
+      "learning_rate": 0.002874,
+      "loss": 4.5402,
+      "step": 958
+    },
+    {
+      "epoch": 0.00959,
+      "grad_norm": 0.6216290945191316,
+      "learning_rate": 0.002877,
+      "loss": 4.5294,
+      "step": 959
+    },
+    {
+      "epoch": 0.0096,
+      "grad_norm": 0.44090482568449035,
+      "learning_rate": 0.0028799999999999997,
+      "loss": 4.5205,
+      "step": 960
+    },
+    {
+      "epoch": 0.00961,
+      "grad_norm": 0.5026549244936088,
+      "learning_rate": 0.002883,
+      "loss": 4.4973,
+      "step": 961
+    },
+    {
+      "epoch": 0.00962,
+      "grad_norm": 0.46550744372429714,
+      "learning_rate": 0.002886,
+      "loss": 4.5199,
+      "step": 962
+    },
+    {
+      "epoch": 0.00963,
+      "grad_norm": 0.4817462883709995,
+      "learning_rate": 0.002889,
+      "loss": 4.5204,
+      "step": 963
+    },
+    {
+      "epoch": 0.00964,
+      "grad_norm": 0.5021989893794454,
+      "learning_rate": 0.002892,
+      "loss": 4.5105,
+      "step": 964
+    },
+    {
+      "epoch": 0.00965,
+      "grad_norm": 0.6331237702649058,
+      "learning_rate": 0.002895,
+      "loss": 4.4888,
+      "step": 965
+    },
+    {
+      "epoch": 0.00966,
+      "grad_norm": 0.7186463225121739,
+      "learning_rate": 0.002898,
+      "loss": 4.5122,
+      "step": 966
+    },
+    {
+      "epoch": 0.00967,
+      "grad_norm": 0.835541291398658,
+      "learning_rate": 0.002901,
+      "loss": 4.5497,
+      "step": 967
+    },
+    {
+      "epoch": 0.00968,
+      "grad_norm": 0.7770950591910699,
+      "learning_rate": 0.002904,
+      "loss": 4.5246,
+      "step": 968
+    },
+    {
+      "epoch": 0.00969,
+      "grad_norm": 0.6421972738290654,
+      "learning_rate": 0.002907,
+      "loss": 4.5465,
+      "step": 969
+    },
+    {
+      "epoch": 0.0097,
+      "grad_norm": 0.6170493579190435,
+      "learning_rate": 0.00291,
+      "loss": 4.5014,
+      "step": 970
+    },
+    {
+      "epoch": 0.00971,
+      "grad_norm": 0.671893763395282,
+      "learning_rate": 0.002913,
+      "loss": 4.5134,
+      "step": 971
+    },
+    {
+      "epoch": 0.00972,
+      "grad_norm": 0.5384349268117217,
+      "learning_rate": 0.002916,
+      "loss": 4.51,
+      "step": 972
+    },
+    {
+      "epoch": 0.00973,
+      "grad_norm": 0.6570052261370841,
+      "learning_rate": 0.002919,
+      "loss": 4.5075,
+      "step": 973
+    },
+    {
+      "epoch": 0.00974,
+      "grad_norm": 0.6469437996214488,
+      "learning_rate": 0.002922,
+      "loss": 4.5042,
+      "step": 974
+    },
+    {
+      "epoch": 0.00975,
+      "grad_norm": 0.5139434995269291,
+      "learning_rate": 0.002925,
+      "loss": 4.5141,
+      "step": 975
+    },
+    {
+      "epoch": 0.00976,
+      "grad_norm": 0.540350404123188,
+      "learning_rate": 0.002928,
+      "loss": 4.4984,
+      "step": 976
+    },
+    {
+      "epoch": 0.00977,
+      "grad_norm": 0.5640158884340003,
+      "learning_rate": 0.002931,
+      "loss": 4.5359,
+      "step": 977
+    },
+    {
+      "epoch": 0.00978,
+      "grad_norm": 0.5471232322596488,
+      "learning_rate": 0.002934,
+      "loss": 4.5069,
+      "step": 978
+    },
+    {
+      "epoch": 0.00979,
+      "grad_norm": 0.6391692840311302,
+      "learning_rate": 0.002937,
+      "loss": 4.4972,
+      "step": 979
+    },
+    {
+      "epoch": 0.0098,
+      "grad_norm": 0.7065424241899814,
+      "learning_rate": 0.00294,
+      "loss": 4.4963,
+      "step": 980
+    },
+    {
+      "epoch": 0.00981,
+      "grad_norm": 0.6903724510426201,
+      "learning_rate": 0.002943,
+      "loss": 4.5078,
+      "step": 981
+    },
+    {
+      "epoch": 0.00982,
+      "grad_norm": 0.653302049548968,
+      "learning_rate": 0.002946,
+      "loss": 4.4991,
+      "step": 982
+    },
+    {
+      "epoch": 0.00983,
+      "grad_norm": 0.7786836590229197,
+      "learning_rate": 0.0029490000000000002,
+      "loss": 4.5061,
+      "step": 983
+    },
+    {
+      "epoch": 0.00984,
+      "grad_norm": 0.8944946184941699,
+      "learning_rate": 0.002952,
+      "loss": 4.5043,
+      "step": 984
+    },
+    {
+      "epoch": 0.00985,
+      "grad_norm": 0.9434362337974719,
+      "learning_rate": 0.002955,
+      "loss": 4.5335,
+      "step": 985
+    },
+    {
+      "epoch": 0.00986,
+      "grad_norm": 0.9312010498282314,
+      "learning_rate": 0.002958,
+      "loss": 4.4908,
+      "step": 986
+    },
+    {
+      "epoch": 0.00987,
+      "grad_norm": 0.7600059379317153,
+      "learning_rate": 0.002961,
+      "loss": 4.5038,
+      "step": 987
+    },
+    {
+      "epoch": 0.00988,
+      "grad_norm": 0.6949983731085813,
+      "learning_rate": 0.002964,
+      "loss": 4.533,
+      "step": 988
+    },
+    {
+      "epoch": 0.00989,
+      "grad_norm": 0.6958207764457225,
+      "learning_rate": 0.002967,
+      "loss": 4.4744,
+      "step": 989
+    },
+    {
+      "epoch": 0.0099,
+      "grad_norm": 0.8314159874496235,
+      "learning_rate": 0.00297,
+      "loss": 4.5218,
+      "step": 990
+    },
+    {
+      "epoch": 0.00991,
+      "grad_norm": 0.90109982550625,
+      "learning_rate": 0.002973,
+      "loss": 4.5224,
+      "step": 991
+    },
+    {
+      "epoch": 0.00992,
+      "grad_norm": 0.951208720196979,
+      "learning_rate": 0.002976,
+      "loss": 4.5281,
+      "step": 992
+    },
+    {
+      "epoch": 0.00993,
+      "grad_norm": 1.034654706401041,
+      "learning_rate": 0.002979,
+      "loss": 4.539,
+      "step": 993
+    },
+    {
+      "epoch": 0.00994,
+      "grad_norm": 1.0078120294806783,
+      "learning_rate": 0.002982,
+      "loss": 4.5459,
+      "step": 994
+    },
+    {
+      "epoch": 0.00995,
+      "grad_norm": 0.9905056664043064,
+      "learning_rate": 0.0029850000000000002,
+      "loss": 4.5461,
+      "step": 995
+    },
+    {
+      "epoch": 0.00996,
+      "grad_norm": 1.1603429750232241,
+      "learning_rate": 0.002988,
+      "loss": 4.5605,
+      "step": 996
+    },
+    {
+      "epoch": 0.00997,
+      "grad_norm": 0.8595677567729485,
+      "learning_rate": 0.002991,
+      "loss": 4.5366,
+      "step": 997
+    },
+    {
+      "epoch": 0.00998,
+      "grad_norm": 0.9672785170741734,
+      "learning_rate": 0.002994,
+      "loss": 4.5467,
+      "step": 998
+    },
+    {
+      "epoch": 0.00999,
+      "grad_norm": 0.8855476122040005,
+      "learning_rate": 0.002997,
+      "loss": 4.5306,
+      "step": 999
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7479993027173912,
+      "learning_rate": 0.003,
+      "loss": 4.5436,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01001,
+      "grad_norm": 0.7121555496393158,
+      "learning_rate": 0.003,
+      "loss": 4.5401,
+      "step": 1001
+    },
+    {
+      "epoch": 0.01002,
+      "grad_norm": 0.72505396659877,
+      "learning_rate": 0.003,
+      "loss": 4.5099,
+      "step": 1002
+    },
+    {
+      "epoch": 0.01003,
+      "grad_norm": 0.7482409624375888,
+      "learning_rate": 0.003,
+      "loss": 4.5346,
+      "step": 1003
+    },
+    {
+      "epoch": 0.01004,
+      "grad_norm": 0.5922239695560899,
+      "learning_rate": 0.003,
+      "loss": 4.513,
+      "step": 1004
+    },
+    {
+      "epoch": 0.01005,
+      "grad_norm": 0.7163386651042714,
+      "learning_rate": 0.003,
+      "loss": 4.4897,
+      "step": 1005
+    },
+    {
+      "epoch": 0.01006,
+      "grad_norm": 0.7664341729228217,
+      "learning_rate": 0.003,
+      "loss": 4.4943,
+      "step": 1006
+    },
+    {
+      "epoch": 0.01007,
+      "grad_norm": 0.6325498197242884,
+      "learning_rate": 0.003,
+      "loss": 4.5316,
+      "step": 1007
+    },
+    {
+      "epoch": 0.01008,
+      "grad_norm": 0.5334181570982156,
+      "learning_rate": 0.003,
+      "loss": 4.5325,
+      "step": 1008
+    },
+    {
+      "epoch": 0.01009,
+      "grad_norm": 0.4756687788935663,
+      "learning_rate": 0.003,
+      "loss": 4.4998,
+      "step": 1009
+    },
+    {
+      "epoch": 0.0101,
+      "grad_norm": 0.44081077367096977,
+      "learning_rate": 0.003,
+      "loss": 4.5025,
+      "step": 1010
+    },
+    {
+      "epoch": 0.01011,
+      "grad_norm": 0.4706704982402646,
+      "learning_rate": 0.003,
+      "loss": 4.5082,
+      "step": 1011
+    },
+    {
+      "epoch": 0.01012,
+      "grad_norm": 0.47223639219621605,
+      "learning_rate": 0.003,
+      "loss": 4.492,
+      "step": 1012
+    },
+    {
+      "epoch": 0.01013,
+      "grad_norm": 0.4611333961201937,
+      "learning_rate": 0.003,
+      "loss": 4.5135,
+      "step": 1013
+    },
+    {
+      "epoch": 0.01014,
+      "grad_norm": 0.36636268932008986,
+      "learning_rate": 0.003,
+      "loss": 4.4528,
+      "step": 1014
+    },
+    {
+      "epoch": 0.01015,
+      "grad_norm": 0.4484711002206511,
+      "learning_rate": 0.003,
+      "loss": 4.4948,
+      "step": 1015
+    },
+    {
+      "epoch": 0.01016,
+      "grad_norm": 0.553209882568818,
+      "learning_rate": 0.003,
+      "loss": 4.5082,
+      "step": 1016
+    },
+    {
+      "epoch": 0.01017,
+      "grad_norm": 0.8898002239967567,
+      "learning_rate": 0.003,
+      "loss": 4.4579,
+      "step": 1017
+    },
+    {
+      "epoch": 0.01018,
+      "grad_norm": 0.9938726896514534,
+      "learning_rate": 0.003,
+      "loss": 4.5557,
+      "step": 1018
+    },
+    {
+      "epoch": 0.01019,
+      "grad_norm": 0.7008138530645361,
+      "learning_rate": 0.003,
+      "loss": 4.5003,
+      "step": 1019
+    },
+    {
+      "epoch": 0.0102,
+      "grad_norm": 1.006685411589756,
+      "learning_rate": 0.003,
+      "loss": 4.5244,
+      "step": 1020
+    },
+    {
+      "epoch": 0.01021,
+      "grad_norm": 0.6295405897908188,
+      "learning_rate": 0.003,
+      "loss": 4.4915,
+      "step": 1021
+    },
+    {
+      "epoch": 0.01022,
+      "grad_norm": 0.6613412501940705,
+      "learning_rate": 0.003,
+      "loss": 4.5063,
+      "step": 1022
+    },
+    {
+      "epoch": 0.01023,
+      "grad_norm": 0.6284055579476905,
+      "learning_rate": 0.003,
+      "loss": 4.4914,
+      "step": 1023
+    },
+    {
+      "epoch": 0.01024,
+      "grad_norm": 0.6735486877543648,
+      "learning_rate": 0.003,
+      "loss": 4.4573,
+      "step": 1024
+    },
+    {
+      "epoch": 0.01025,
+      "grad_norm": 0.6253815453908939,
+      "learning_rate": 0.003,
+      "loss": 4.4832,
+      "step": 1025
+    },
+    {
+      "epoch": 0.01026,
+      "grad_norm": 0.5932239048767913,
+      "learning_rate": 0.003,
+      "loss": 4.4735,
+      "step": 1026
+    },
+    {
+      "epoch": 0.01027,
+      "grad_norm": 0.6318859057234201,
+      "learning_rate": 0.003,
+      "loss": 4.4952,
+      "step": 1027
+    },
+    {
+      "epoch": 0.01028,
+      "grad_norm": 0.6588163663127304,
+      "learning_rate": 0.003,
+      "loss": 4.4763,
+      "step": 1028
+    },
+    {
+      "epoch": 0.01029,
+      "grad_norm": 0.55721560335989,
+      "learning_rate": 0.003,
+      "loss": 4.4524,
+      "step": 1029
+    },
+    {
+      "epoch": 0.0103,
+      "grad_norm": 0.5468199872259095,
+      "learning_rate": 0.003,
+      "loss": 4.4809,
+      "step": 1030
+    },
+    {
+      "epoch": 0.01031,
+      "grad_norm": 0.5601179932704955,
+      "learning_rate": 0.003,
+      "loss": 4.4934,
+      "step": 1031
+    },
+    {
+      "epoch": 0.01032,
+      "grad_norm": 0.5514443283118416,
+      "learning_rate": 0.003,
+      "loss": 4.464,
+      "step": 1032
+    },
+    {
+      "epoch": 0.01033,
+      "grad_norm": 0.49777794102419176,
+      "learning_rate": 0.003,
+      "loss": 4.496,
+      "step": 1033
+    },
+    {
+      "epoch": 0.01034,
+      "grad_norm": 0.5163173300909295,
+      "learning_rate": 0.003,
+      "loss": 4.4625,
+      "step": 1034
+    },
+    {
+      "epoch": 0.01035,
+      "grad_norm": 0.4798196088286999,
+      "learning_rate": 0.003,
+      "loss": 4.4573,
+      "step": 1035
+    },
+    {
+      "epoch": 0.01036,
+      "grad_norm": 0.3958499097440651,
+      "learning_rate": 0.003,
+      "loss": 4.4459,
+      "step": 1036
+    },
+    {
+      "epoch": 0.01037,
+      "grad_norm": 0.3643003447603396,
+      "learning_rate": 0.003,
+      "loss": 4.4698,
+      "step": 1037
+    },
+    {
+      "epoch": 0.01038,
+      "grad_norm": 0.40989678679265595,
+      "learning_rate": 0.003,
+      "loss": 4.4667,
+      "step": 1038
+    },
+    {
+      "epoch": 0.01039,
+      "grad_norm": 0.3544910886173235,
+      "learning_rate": 0.003,
+      "loss": 4.4556,
+      "step": 1039
+    },
+    {
+      "epoch": 0.0104,
+      "grad_norm": 0.3719879786284374,
+      "learning_rate": 0.003,
+      "loss": 4.4667,
+      "step": 1040
+    },
+    {
+      "epoch": 0.01041,
+      "grad_norm": 0.3770068025965155,
+      "learning_rate": 0.003,
+      "loss": 4.4469,
+      "step": 1041
+    },
+    {
+      "epoch": 0.01042,
+      "grad_norm": 0.4469030721524434,
+      "learning_rate": 0.003,
+      "loss": 4.4752,
+      "step": 1042
+    },
+    {
+      "epoch": 0.01043,
+      "grad_norm": 0.7272268008417163,
+      "learning_rate": 0.003,
+      "loss": 4.4274,
+      "step": 1043
+    },
+    {
+      "epoch": 0.01044,
+      "grad_norm": 0.8992907359513324,
+      "learning_rate": 0.003,
+      "loss": 4.4949,
+      "step": 1044
+    },
+    {
+      "epoch": 0.01045,
+      "grad_norm": 0.8302310255488462,
+      "learning_rate": 0.003,
+      "loss": 4.4953,
+      "step": 1045
+    },
+    {
+      "epoch": 0.01046,
+      "grad_norm": 0.9975061123002056,
+      "learning_rate": 0.003,
+      "loss": 4.4816,
+      "step": 1046
+    },
+    {
+      "epoch": 0.01047,
+      "grad_norm": 0.8067658442263902,
+      "learning_rate": 0.003,
+      "loss": 4.4954,
+      "step": 1047
+    },
+    {
+      "epoch": 0.01048,
+      "grad_norm": 0.7572637033913173,
+      "learning_rate": 0.003,
+      "loss": 4.4741,
+      "step": 1048
+    },
+    {
+      "epoch": 0.01049,
+      "grad_norm": 0.8994748129621607,
+      "learning_rate": 0.003,
+      "loss": 4.4647,
+      "step": 1049
+    },
+    {
+      "epoch": 0.0105,
+      "grad_norm": 0.7962039985766673,
+      "learning_rate": 0.003,
+      "loss": 4.5006,
+      "step": 1050
+    },
+    {
+      "epoch": 0.01051,
+      "grad_norm": 0.8207643682537104,
+      "learning_rate": 0.003,
+      "loss": 4.4811,
+      "step": 1051
+    },
+    {
+      "epoch": 0.01052,
+      "grad_norm": 0.7458558375660994,
+      "learning_rate": 0.003,
+      "loss": 4.4881,
+      "step": 1052
+    },
+    {
+      "epoch": 0.01053,
+      "grad_norm": 0.6759075645027314,
+      "learning_rate": 0.003,
+      "loss": 4.4946,
+      "step": 1053
+    },
+    {
+      "epoch": 0.01054,
+      "grad_norm": 0.6790074754239308,
+      "learning_rate": 0.003,
+      "loss": 4.4852,
+      "step": 1054
+    },
+    {
+      "epoch": 0.01055,
+      "grad_norm": 0.6920845821894167,
+      "learning_rate": 0.003,
+      "loss": 4.4644,
+      "step": 1055
+    },
+    {
+      "epoch": 0.01056,
+      "grad_norm": 0.5219544124128492,
+      "learning_rate": 0.003,
+      "loss": 4.4569,
+      "step": 1056
+    },
+    {
+      "epoch": 0.01057,
+      "grad_norm": 0.6021767240952963,
+      "learning_rate": 0.003,
+      "loss": 4.4651,
+      "step": 1057
+    },
+    {
+      "epoch": 0.01058,
+      "grad_norm": 0.5219741420101096,
+      "learning_rate": 0.003,
+      "loss": 4.4552,
+      "step": 1058
+    },
+    {
+      "epoch": 0.01059,
+      "grad_norm": 0.509217286199775,
+      "learning_rate": 0.003,
+      "loss": 4.4557,
+      "step": 1059
+    },
+    {
+      "epoch": 0.0106,
+      "grad_norm": 0.6119681984085901,
+      "learning_rate": 0.003,
+      "loss": 4.4506,
+      "step": 1060
+    },
+    {
+      "epoch": 0.01061,
+      "grad_norm": 0.6018802462486371,
+      "learning_rate": 0.003,
+      "loss": 4.4451,
+      "step": 1061
+    },
+    {
+      "epoch": 0.01062,
+      "grad_norm": 0.6072220138777784,
+      "learning_rate": 0.003,
+      "loss": 4.4398,
+      "step": 1062
+    },
+    {
+      "epoch": 0.01063,
+      "grad_norm": 0.5793152524763621,
+      "learning_rate": 0.003,
+      "loss": 4.4372,
+      "step": 1063
+    },
+    {
+      "epoch": 0.01064,
+      "grad_norm": 0.740973950389461,
+      "learning_rate": 0.003,
+      "loss": 4.4576,
+      "step": 1064
+    },
+    {
+      "epoch": 0.01065,
+      "grad_norm": 0.8998946238241571,
+      "learning_rate": 0.003,
+      "loss": 4.4756,
+      "step": 1065
+    },
+    {
+      "epoch": 0.01066,
+      "grad_norm": 0.889057289027363,
+      "learning_rate": 0.003,
+      "loss": 4.4727,
+      "step": 1066
+    },
+    {
+      "epoch": 0.01067,
+      "grad_norm": 0.9442468447502006,
+      "learning_rate": 0.003,
+      "loss": 4.4902,
+      "step": 1067
+    },
+    {
+      "epoch": 0.01068,
+      "grad_norm": 1.018364097076677,
+      "learning_rate": 0.003,
+      "loss": 4.4857,
+      "step": 1068
+    },
+    {
+      "epoch": 0.01069,
+      "grad_norm": 0.9799434684826753,
+      "learning_rate": 0.003,
+      "loss": 4.4924,
+      "step": 1069
+    },
+    {
+      "epoch": 0.0107,
+      "grad_norm": 0.8992499083099482,
+      "learning_rate": 0.003,
+      "loss": 4.5446,
+      "step": 1070
+    },
+    {
+      "epoch": 0.01071,
+      "grad_norm": 0.8735440073637682,
+      "learning_rate": 0.003,
+      "loss": 4.4651,
+      "step": 1071
+    },
+    {
+      "epoch": 0.01072,
+      "grad_norm": 0.8443267804596191,
+      "learning_rate": 0.003,
+      "loss": 4.5031,
+      "step": 1072
+    },
+    {
+      "epoch": 0.01073,
+      "grad_norm": 0.7866639405001601,
+      "learning_rate": 0.003,
+      "loss": 4.4993,
+      "step": 1073
+    },
+    {
+      "epoch": 0.01074,
+      "grad_norm": 0.7412592488158714,
+      "learning_rate": 0.003,
+      "loss": 4.4612,
+      "step": 1074
+    },
+    {
+      "epoch": 0.01075,
+      "grad_norm": 0.7055777296919327,
+      "learning_rate": 0.003,
+      "loss": 4.4949,
+      "step": 1075
+    },
+    {
+      "epoch": 0.01076,
+      "grad_norm": 0.6602103525167045,
+      "learning_rate": 0.003,
+      "loss": 4.4903,
+      "step": 1076
+    },
+    {
+      "epoch": 0.01077,
+      "grad_norm": 0.6221585201435259,
+      "learning_rate": 0.003,
+      "loss": 4.4791,
+      "step": 1077
+    },
+    {
+      "epoch": 0.01078,
+      "grad_norm": 0.6312740380808073,
+      "learning_rate": 0.003,
+      "loss": 4.4373,
+      "step": 1078
+    },
+    {
+      "epoch": 0.01079,
+      "grad_norm": 0.5465751942660528,
+      "learning_rate": 0.003,
+      "loss": 4.4641,
+      "step": 1079
+    },
+    {
+      "epoch": 0.0108,
+      "grad_norm": 0.5040258102852178,
+      "learning_rate": 0.003,
+      "loss": 4.4392,
+      "step": 1080
+    },
+    {
+      "epoch": 0.01081,
+      "grad_norm": 0.4557353176092044,
+      "learning_rate": 0.003,
+      "loss": 4.4439,
+      "step": 1081
+    },
+    {
+      "epoch": 0.01082,
+      "grad_norm": 0.43657211030445375,
+      "learning_rate": 0.003,
+      "loss": 4.4598,
+      "step": 1082
+    },
+    {
+      "epoch": 0.01083,
+      "grad_norm": 0.37468227521199704,
+      "learning_rate": 0.003,
+      "loss": 4.4419,
+      "step": 1083
+    },
+    {
+      "epoch": 0.01084,
+      "grad_norm": 0.3702365616744669,
+      "learning_rate": 0.003,
+      "loss": 4.4465,
+      "step": 1084
+    },
+    {
+      "epoch": 0.01085,
+      "grad_norm": 0.3248887668649726,
+      "learning_rate": 0.003,
+      "loss": 4.4323,
+      "step": 1085
+    },
+    {
+      "epoch": 0.01086,
+      "grad_norm": 0.28646275942866467,
+      "learning_rate": 0.003,
+      "loss": 4.4059,
+      "step": 1086
+    },
+    {
+      "epoch": 0.01087,
+      "grad_norm": 0.3109295903469056,
+      "learning_rate": 0.003,
+      "loss": 4.4215,
+      "step": 1087
+    },
+    {
+      "epoch": 0.01088,
+      "grad_norm": 0.31305777007935864,
+      "learning_rate": 0.003,
+      "loss": 4.454,
+      "step": 1088
+    },
+    {
+      "epoch": 0.01089,
+      "grad_norm": 0.3202194647994524,
+      "learning_rate": 0.003,
+      "loss": 4.423,
+      "step": 1089
+    },
+    {
+      "epoch": 0.0109,
+      "grad_norm": 0.3546167101166988,
+      "learning_rate": 0.003,
+      "loss": 4.4466,
+      "step": 1090
+    },
+    {
+      "epoch": 0.01091,
+      "grad_norm": 0.40979643182380626,
+      "learning_rate": 0.003,
+      "loss": 4.4162,
+      "step": 1091
+    },
+    {
+      "epoch": 0.01092,
+      "grad_norm": 0.44852044367674554,
+      "learning_rate": 0.003,
+      "loss": 4.4131,
+      "step": 1092
+    },
+    {
+      "epoch": 0.01093,
+      "grad_norm": 0.5395428853759895,
+      "learning_rate": 0.003,
+      "loss": 4.4351,
+      "step": 1093
+    },
+    {
+      "epoch": 0.01094,
+      "grad_norm": 0.7419472173607777,
+      "learning_rate": 0.003,
+      "loss": 4.4344,
+      "step": 1094
+    },
+    {
+      "epoch": 0.01095,
+      "grad_norm": 1.3268112301670232,
+      "learning_rate": 0.003,
+      "loss": 4.4875,
+      "step": 1095
+    },
+    {
+      "epoch": 0.01096,
+      "grad_norm": 0.8107867356113418,
+      "learning_rate": 0.003,
+      "loss": 4.4576,
+      "step": 1096
+    },
+    {
+      "epoch": 0.01097,
+      "grad_norm": 0.6635063888056452,
+      "learning_rate": 0.003,
+      "loss": 4.4388,
+      "step": 1097
+    },
+    {
+      "epoch": 0.01098,
+      "grad_norm": 0.9662363707419442,
+      "learning_rate": 0.003,
+      "loss": 4.474,
+      "step": 1098
+    },
+    {
+      "epoch": 0.01099,
+      "grad_norm": 0.7696101380141932,
+      "learning_rate": 0.003,
+      "loss": 4.4405,
+      "step": 1099
+    },
+    {
+      "epoch": 0.011,
+      "grad_norm": 0.7450236895824817,
+      "learning_rate": 0.003,
+      "loss": 4.4451,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01101,
+      "grad_norm": 0.7254968126029079,
+      "learning_rate": 0.003,
+      "loss": 4.4513,
+      "step": 1101
+    },
+    {
+      "epoch": 0.01102,
+      "grad_norm": 0.7537628081039499,
+      "learning_rate": 0.003,
+      "loss": 4.4588,
+      "step": 1102
+    },
+    {
+      "epoch": 0.01103,
+      "grad_norm": 0.7123173358553468,
+      "learning_rate": 0.003,
+      "loss": 4.4551,
+      "step": 1103
+    },
+    {
+      "epoch": 0.01104,
+      "grad_norm": 0.6033148725235706,
+      "learning_rate": 0.003,
+      "loss": 4.4384,
+      "step": 1104
+    },
+    {
+      "epoch": 0.01105,
+      "grad_norm": 0.7223359575565769,
+      "learning_rate": 0.003,
+      "loss": 4.4643,
+      "step": 1105
+    },
+    {
+      "epoch": 0.01106,
+      "grad_norm": 0.6749769688136343,
+      "learning_rate": 0.003,
+      "loss": 4.4355,
+      "step": 1106
+    },
+    {
+      "epoch": 0.01107,
+      "grad_norm": 0.6519921729558845,
+      "learning_rate": 0.003,
+      "loss": 4.443,
+      "step": 1107
+    },
+    {
+      "epoch": 0.01108,
+      "grad_norm": 0.6442390747026363,
+      "learning_rate": 0.003,
+      "loss": 4.4155,
+      "step": 1108
+    },
+    {
+      "epoch": 0.01109,
+      "grad_norm": 0.5210761074063659,
+      "learning_rate": 0.003,
+      "loss": 4.4258,
+      "step": 1109
+    },
+    {
+      "epoch": 0.0111,
+      "grad_norm": 0.49112961811162964,
+      "learning_rate": 0.003,
+      "loss": 4.4439,
+      "step": 1110
+    },
+    {
+      "epoch": 0.01111,
+      "grad_norm": 0.5154689325771941,
+      "learning_rate": 0.003,
+      "loss": 4.4228,
+      "step": 1111
+    },
+    {
+      "epoch": 0.01112,
+      "grad_norm": 0.5147734027357755,
+      "learning_rate": 0.003,
+      "loss": 4.4383,
+      "step": 1112
+    },
+    {
+      "epoch": 0.01113,
+      "grad_norm": 0.4905886669705696,
+      "learning_rate": 0.003,
+      "loss": 4.4046,
+      "step": 1113
+    },
+    {
+      "epoch": 0.01114,
+      "grad_norm": 0.47555859430135783,
+      "learning_rate": 0.003,
+      "loss": 4.4166,
+      "step": 1114
+    },
+    {
+      "epoch": 0.01115,
+      "grad_norm": 0.42768329796719207,
+      "learning_rate": 0.003,
+      "loss": 4.4369,
+      "step": 1115
+    },
+    {
+      "epoch": 0.01116,
+      "grad_norm": 0.4281709726428523,
+      "learning_rate": 0.003,
+      "loss": 4.4286,
+      "step": 1116
+    },
+    {
+      "epoch": 0.01117,
+      "grad_norm": 0.4124788023594352,
+      "learning_rate": 0.003,
+      "loss": 4.414,
+      "step": 1117
+    },
+    {
+      "epoch": 0.01118,
+      "grad_norm": 0.4531459371875401,
+      "learning_rate": 0.003,
+      "loss": 4.415,
+      "step": 1118
+    },
+    {
+      "epoch": 0.01119,
+      "grad_norm": 0.5115410769007107,
+      "learning_rate": 0.003,
+      "loss": 4.4247,
+      "step": 1119
+    },
+    {
+      "epoch": 0.0112,
+      "grad_norm": 0.6598099600325713,
+      "learning_rate": 0.003,
+      "loss": 4.4115,
+      "step": 1120
+    },
+    {
+      "epoch": 0.01121,
+      "grad_norm": 0.7298362819040065,
+      "learning_rate": 0.003,
+      "loss": 4.4356,
+      "step": 1121
+    },
+    {
+      "epoch": 0.01122,
+      "grad_norm": 0.6383918217981295,
+      "learning_rate": 0.003,
+      "loss": 4.4021,
+      "step": 1122
+    },
+    {
+      "epoch": 0.01123,
+      "grad_norm": 0.7157274101955465,
+      "learning_rate": 0.003,
+      "loss": 4.3942,
+      "step": 1123
+    },
+    {
+      "epoch": 0.01124,
+      "grad_norm": 0.6763302773515354,
+      "learning_rate": 0.003,
+      "loss": 4.4444,
+      "step": 1124
+    },
+    {
+      "epoch": 0.01125,
+      "grad_norm": 0.5890253028490829,
+      "learning_rate": 0.003,
+      "loss": 4.4062,
+      "step": 1125
+    },
+    {
+      "epoch": 0.01126,
+      "grad_norm": 0.6745452077137704,
+      "learning_rate": 0.003,
+      "loss": 4.4204,
+      "step": 1126
+    },
+    {
+      "epoch": 0.01127,
+      "grad_norm": 0.6212265837956056,
+      "learning_rate": 0.003,
+      "loss": 4.446,
+      "step": 1127
+    },
+    {
+      "epoch": 0.01128,
+      "grad_norm": 0.6579375902442356,
+      "learning_rate": 0.003,
+      "loss": 4.4285,
+      "step": 1128
+    },
+    {
+      "epoch": 0.01129,
+      "grad_norm": 0.7038241088934594,
+      "learning_rate": 0.003,
+      "loss": 4.4197,
+      "step": 1129
+    },
+    {
+      "epoch": 0.0113,
+      "grad_norm": 0.8537912734504166,
+      "learning_rate": 0.003,
+      "loss": 4.4099,
+      "step": 1130
+    },
+    {
+      "epoch": 0.01131,
+      "grad_norm": 0.6676066708341881,
+      "learning_rate": 0.003,
+      "loss": 4.4184,
+      "step": 1131
+    },
+    {
+      "epoch": 0.01132,
+      "grad_norm": 0.6262944206433371,
+      "learning_rate": 0.003,
+      "loss": 4.4418,
+      "step": 1132
+    },
+    {
+      "epoch": 0.01133,
+      "grad_norm": 0.7096964978520276,
+      "learning_rate": 0.003,
+      "loss": 4.4077,
+      "step": 1133
+    },
+    {
+      "epoch": 0.01134,
+      "grad_norm": 0.7063046600590702,
+      "learning_rate": 0.003,
+      "loss": 4.4219,
+      "step": 1134
+    },
+    {
+      "epoch": 0.01135,
+      "grad_norm": 0.7389408975678705,
+      "learning_rate": 0.003,
+      "loss": 4.4511,
+      "step": 1135
+    },
+    {
+      "epoch": 0.01136,
+      "grad_norm": 0.902932943298548,
+      "learning_rate": 0.003,
+      "loss": 4.4334,
+      "step": 1136
+    },
+    {
+      "epoch": 0.01137,
+      "grad_norm": 0.8104098050855381,
+      "learning_rate": 0.003,
+      "loss": 4.4615,
+      "step": 1137
+    },
+    {
+      "epoch": 0.01138,
+      "grad_norm": 0.6304382373186352,
+      "learning_rate": 0.003,
+      "loss": 4.4061,
+      "step": 1138
+    },
+    {
+      "epoch": 0.01139,
+      "grad_norm": 0.6573985975094263,
+      "learning_rate": 0.003,
+      "loss": 4.4174,
+      "step": 1139
+    },
+    {
+      "epoch": 0.0114,
+      "grad_norm": 0.6243548176788066,
+      "learning_rate": 0.003,
+      "loss": 4.4015,
+      "step": 1140
+    },
+    {
+      "epoch": 0.01141,
+      "grad_norm": 0.6941761654842915,
+      "learning_rate": 0.003,
+      "loss": 4.4278,
+      "step": 1141
+    },
+    {
+      "epoch": 0.01142,
+      "grad_norm": 0.6422841400280502,
+      "learning_rate": 0.003,
+      "loss": 4.4042,
+      "step": 1142
+    },
+    {
+      "epoch": 0.01143,
+      "grad_norm": 0.6274623913648985,
+      "learning_rate": 0.003,
+      "loss": 4.4193,
+      "step": 1143
+    },
+    {
+      "epoch": 0.01144,
+      "grad_norm": 0.6449672978265372,
+      "learning_rate": 0.003,
+      "loss": 4.3861,
+      "step": 1144
+    },
+    {
+      "epoch": 0.01145,
+      "grad_norm": 0.8094555842673681,
+      "learning_rate": 0.003,
+      "loss": 4.4312,
+      "step": 1145
+    },
+    {
+      "epoch": 0.01146,
+      "grad_norm": 0.8440842425424007,
+      "learning_rate": 0.003,
+      "loss": 4.4247,
+      "step": 1146
+    },
+    {
+      "epoch": 0.01147,
+      "grad_norm": 0.8250289231813209,
+      "learning_rate": 0.003,
+      "loss": 4.4121,
+      "step": 1147
+    },
+    {
+      "epoch": 0.01148,
+      "grad_norm": 0.9842608879774175,
+      "learning_rate": 0.003,
+      "loss": 4.4248,
+      "step": 1148
+    },
+    {
+      "epoch": 0.01149,
+      "grad_norm": 1.140890319957312,
+      "learning_rate": 0.003,
+      "loss": 4.4515,
+      "step": 1149
+    },
+    {
+      "epoch": 0.0115,
+      "grad_norm": 1.0049560913231448,
+      "learning_rate": 0.003,
+      "loss": 4.454,
+      "step": 1150
+    },
+    {
+      "epoch": 0.01151,
+      "grad_norm": 0.9241096807179571,
+      "learning_rate": 0.003,
+      "loss": 4.4438,
+      "step": 1151
+    },
+    {
+      "epoch": 0.01152,
+      "grad_norm": 0.8197935874101315,
+      "learning_rate": 0.003,
+      "loss": 4.4496,
+      "step": 1152
+    },
+    {
+      "epoch": 0.01153,
+      "grad_norm": 0.8964418195355868,
+      "learning_rate": 0.003,
+      "loss": 4.4768,
+      "step": 1153
+    },
+    {
+      "epoch": 0.01154,
+      "grad_norm": 0.7052475574958695,
+      "learning_rate": 0.003,
+      "loss": 4.4381,
+      "step": 1154
+    },
+    {
+      "epoch": 0.01155,
+      "grad_norm": 0.6994161040159336,
+      "learning_rate": 0.003,
+      "loss": 4.425,
+      "step": 1155
+    },
+    {
+      "epoch": 0.01156,
+      "grad_norm": 0.5866991021029688,
+      "learning_rate": 0.003,
+      "loss": 4.4267,
+      "step": 1156
+    },
+    {
+      "epoch": 0.01157,
+      "grad_norm": 0.5890541796764595,
+      "learning_rate": 0.003,
+      "loss": 4.4503,
+      "step": 1157
+    },
+    {
+      "epoch": 0.01158,
+      "grad_norm": 0.5558417103751193,
+      "learning_rate": 0.003,
+      "loss": 4.4146,
+      "step": 1158
+    },
+    {
+      "epoch": 0.01159,
+      "grad_norm": 0.5494688252527891,
+      "learning_rate": 0.003,
+      "loss": 4.41,
+      "step": 1159
+    },
+    {
+      "epoch": 0.0116,
+      "grad_norm": 0.5721755744732883,
+      "learning_rate": 0.003,
+      "loss": 4.4228,
+      "step": 1160
+    },
+    {
+      "epoch": 0.01161,
+      "grad_norm": 0.6717218566295324,
+      "learning_rate": 0.003,
+      "loss": 4.4019,
+      "step": 1161
+    },
+    {
+      "epoch": 0.01162,
+      "grad_norm": 0.7268973886687413,
+      "learning_rate": 0.003,
+      "loss": 4.4138,
+      "step": 1162
+    },
+    {
+      "epoch": 0.01163,
+      "grad_norm": 0.5842283691456034,
+      "learning_rate": 0.003,
+      "loss": 4.4284,
+      "step": 1163
+    },
+    {
+      "epoch": 0.01164,
+      "grad_norm": 0.4855856513804374,
+      "learning_rate": 0.003,
+      "loss": 4.4134,
+      "step": 1164
+    },
+    {
+      "epoch": 0.01165,
+      "grad_norm": 0.4188604676147223,
+      "learning_rate": 0.003,
+      "loss": 4.4035,
+      "step": 1165
+    },
+    {
+      "epoch": 0.01166,
+      "grad_norm": 0.3540156009620338,
+      "learning_rate": 0.003,
+      "loss": 4.4019,
+      "step": 1166
+    },
+    {
+      "epoch": 0.01167,
+      "grad_norm": 0.34211903868248933,
+      "learning_rate": 0.003,
+      "loss": 4.3663,
+      "step": 1167
+    },
+    {
+      "epoch": 0.01168,
+      "grad_norm": 0.3891962127459521,
+      "learning_rate": 0.003,
+      "loss": 4.374,
+      "step": 1168
+    },
+    {
+      "epoch": 0.01169,
+      "grad_norm": 0.43062143986497,
+      "learning_rate": 0.003,
+      "loss": 4.4147,
+      "step": 1169
+    },
+    {
+      "epoch": 0.0117,
+      "grad_norm": 0.5082945055708723,
+      "learning_rate": 0.003,
+      "loss": 4.404,
+      "step": 1170
+    },
+    {
+      "epoch": 0.01171,
+      "grad_norm": 0.6470102757597544,
+      "learning_rate": 0.003,
+      "loss": 4.3885,
+      "step": 1171
+    },
+    {
+      "epoch": 0.01172,
+      "grad_norm": 0.7362820078056241,
+      "learning_rate": 0.003,
+      "loss": 4.3869,
+      "step": 1172
+    },
+    {
+      "epoch": 0.01173,
+      "grad_norm": 0.693269388291317,
+      "learning_rate": 0.003,
+      "loss": 4.4171,
+      "step": 1173
+    },
+    {
+      "epoch": 0.01174,
+      "grad_norm": 0.5646769911878592,
+      "learning_rate": 0.003,
+      "loss": 4.3891,
+      "step": 1174
+    },
+    {
+      "epoch": 0.01175,
+      "grad_norm": 0.5675605646992437,
+      "learning_rate": 0.003,
+      "loss": 4.3751,
+      "step": 1175
+    },
+    {
+      "epoch": 0.01176,
+      "grad_norm": 0.5874888204753396,
+      "learning_rate": 0.003,
+      "loss": 4.4094,
+      "step": 1176
+    },
+    {
+      "epoch": 0.01177,
+      "grad_norm": 0.4402531057426568,
+      "learning_rate": 0.003,
+      "loss": 4.3711,
+      "step": 1177
+    },
+    {
+      "epoch": 0.01178,
+      "grad_norm": 0.542910388266956,
+      "learning_rate": 0.003,
+      "loss": 4.3959,
+      "step": 1178
+    },
+    {
+      "epoch": 0.01179,
+      "grad_norm": 0.6077990883816204,
+      "learning_rate": 0.003,
+      "loss": 4.3751,
+      "step": 1179
+    },
+    {
+      "epoch": 0.0118,
+      "grad_norm": 0.5935696360755632,
+      "learning_rate": 0.003,
+      "loss": 4.4393,
+      "step": 1180
+    },
+    {
+      "epoch": 0.01181,
+      "grad_norm": 0.5077497618150401,
+      "learning_rate": 0.003,
+      "loss": 4.4227,
+      "step": 1181
+    },
+    {
+      "epoch": 0.01182,
+      "grad_norm": 0.4525800009003507,
+      "learning_rate": 0.003,
+      "loss": 4.3791,
+      "step": 1182
+    },
+    {
+      "epoch": 0.01183,
+      "grad_norm": 0.465730915302176,
+      "learning_rate": 0.003,
+      "loss": 4.3741,
+      "step": 1183
+    },
+    {
+      "epoch": 0.01184,
+      "grad_norm": 0.5691375544130988,
+      "learning_rate": 0.003,
+      "loss": 4.4172,
+      "step": 1184
+    },
+    {
+      "epoch": 0.01185,
+      "grad_norm": 0.6220094109188331,
+      "learning_rate": 0.003,
+      "loss": 4.3896,
+      "step": 1185
+    },
+    {
+      "epoch": 0.01186,
+      "grad_norm": 0.7180162404965751,
+      "learning_rate": 0.003,
+      "loss": 4.417,
+      "step": 1186
+    },
+    {
+      "epoch": 0.01187,
+      "grad_norm": 0.7918251317294935,
+      "learning_rate": 0.003,
+      "loss": 4.389,
+      "step": 1187
+    },
+    {
+      "epoch": 0.01188,
+      "grad_norm": 0.8521642061482538,
+      "learning_rate": 0.003,
+      "loss": 4.4081,
+      "step": 1188
+    },
+    {
+      "epoch": 0.01189,
+      "grad_norm": 0.884743262111852,
+      "learning_rate": 0.003,
+      "loss": 4.4173,
+      "step": 1189
+    },
+    {
+      "epoch": 0.0119,
+      "grad_norm": 0.809315268231435,
+      "learning_rate": 0.003,
+      "loss": 4.3961,
+      "step": 1190
+    },
+    {
+      "epoch": 0.01191,
+      "grad_norm": 0.8426026711506704,
+      "learning_rate": 0.003,
+      "loss": 4.3805,
+      "step": 1191
+    },
+    {
+      "epoch": 0.01192,
+      "grad_norm": 0.8039017391990436,
+      "learning_rate": 0.003,
+      "loss": 4.4134,
+      "step": 1192
+    },
+    {
+      "epoch": 0.01193,
+      "grad_norm": 1.0091767993853729,
+      "learning_rate": 0.003,
+      "loss": 4.4278,
+      "step": 1193
+    },
+    {
+      "epoch": 0.01194,
+      "grad_norm": 1.021446612913599,
+      "learning_rate": 0.003,
+      "loss": 4.4458,
+      "step": 1194
+    },
+    {
+      "epoch": 0.01195,
+      "grad_norm": 0.8197186317539255,
+      "learning_rate": 0.003,
+      "loss": 4.4136,
+      "step": 1195
+    },
+    {
+      "epoch": 0.01196,
+      "grad_norm": 1.0663075325190898,
+      "learning_rate": 0.003,
+      "loss": 4.4346,
+      "step": 1196
+    },
+    {
+      "epoch": 0.01197,
+      "grad_norm": 1.1163749964317249,
+      "learning_rate": 0.003,
+      "loss": 4.42,
+      "step": 1197
+    },
+    {
+      "epoch": 0.01198,
+      "grad_norm": 0.7911818503418244,
+      "learning_rate": 0.003,
+      "loss": 4.4756,
+      "step": 1198
+    },
+    {
+      "epoch": 0.01199,
+      "grad_norm": 0.7094954502714929,
+      "learning_rate": 0.003,
+      "loss": 4.4429,
+      "step": 1199
+    },
+    {
+      "epoch": 0.012,
+      "grad_norm": 0.6371349856097592,
+      "learning_rate": 0.003,
+      "loss": 4.4311,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01201,
+      "grad_norm": 0.6670059038710348,
+      "learning_rate": 0.003,
+      "loss": 4.4586,
+      "step": 1201
+    },
+    {
+      "epoch": 0.01202,
+      "grad_norm": 0.7057344621036731,
+      "learning_rate": 0.003,
+      "loss": 4.4253,
+      "step": 1202
+    },
+    {
+      "epoch": 0.01203,
+      "grad_norm": 0.6813638781778574,
+      "learning_rate": 0.003,
+      "loss": 4.4314,
+      "step": 1203
+    },
+    {
+      "epoch": 0.01204,
+      "grad_norm": 0.6487927748495244,
+      "learning_rate": 0.003,
+      "loss": 4.4252,
+      "step": 1204
+    },
+    {
+      "epoch": 0.01205,
+      "grad_norm": 0.5702088035511471,
+      "learning_rate": 0.003,
+      "loss": 4.4134,
+      "step": 1205
+    },
+    {
+      "epoch": 0.01206,
+      "grad_norm": 0.5185412999807271,
+      "learning_rate": 0.003,
+      "loss": 4.4527,
+      "step": 1206
+    },
+    {
+      "epoch": 0.01207,
+      "grad_norm": 0.5529206635637681,
+      "learning_rate": 0.003,
+      "loss": 4.3748,
+      "step": 1207
+    },
+    {
+      "epoch": 0.01208,
+      "grad_norm": 0.6027594396835911,
+      "learning_rate": 0.003,
+      "loss": 4.4355,
+      "step": 1208
+    },
+    {
+      "epoch": 0.01209,
+      "grad_norm": 0.6294683615242583,
+      "learning_rate": 0.003,
+      "loss": 4.4139,
+      "step": 1209
+    },
+    {
+      "epoch": 0.0121,
+      "grad_norm": 0.5634425369332556,
+      "learning_rate": 0.003,
+      "loss": 4.4142,
+      "step": 1210
+    },
+    {
+      "epoch": 0.01211,
+      "grad_norm": 0.489983773731107,
+      "learning_rate": 0.003,
+      "loss": 4.3953,
+      "step": 1211
+    },
+    {
+      "epoch": 0.01212,
+      "grad_norm": 0.48590408635878257,
+      "learning_rate": 0.003,
+      "loss": 4.4163,
+      "step": 1212
+    },
+    {
+      "epoch": 0.01213,
+      "grad_norm": 0.4654031475329597,
+      "learning_rate": 0.003,
+      "loss": 4.3921,
+      "step": 1213
+    },
+    {
+      "epoch": 0.01214,
+      "grad_norm": 0.4519325610344299,
+      "learning_rate": 0.003,
+      "loss": 4.3755,
+      "step": 1214
+    },
+    {
+      "epoch": 0.01215,
+      "grad_norm": 0.4986453835908023,
+      "learning_rate": 0.003,
+      "loss": 4.3833,
+      "step": 1215
+    },
+    {
+      "epoch": 0.01216,
+      "grad_norm": 0.6836671205807361,
+      "learning_rate": 0.003,
+      "loss": 4.3816,
+      "step": 1216
+    },
+    {
+      "epoch": 0.01217,
+      "grad_norm": 0.8634560994181636,
+      "learning_rate": 0.003,
+      "loss": 4.4035,
+      "step": 1217
+    },
+    {
+      "epoch": 0.01218,
+      "grad_norm": 0.7430166358012542,
+      "learning_rate": 0.003,
+      "loss": 4.3918,
+      "step": 1218
+    },
+    {
+      "epoch": 0.01219,
+      "grad_norm": 0.5693399114743081,
+      "learning_rate": 0.003,
+      "loss": 4.3834,
+      "step": 1219
+    },
+    {
+      "epoch": 0.0122,
+      "grad_norm": 0.6203041570130704,
+      "learning_rate": 0.003,
+      "loss": 4.3803,
+      "step": 1220
+    },
+    {
+      "epoch": 0.01221,
+      "grad_norm": 0.6221216866599081,
+      "learning_rate": 0.003,
+      "loss": 4.3932,
+      "step": 1221
+    },
+    {
+      "epoch": 0.01222,
+      "grad_norm": 0.5678017035835031,
+      "learning_rate": 0.003,
+      "loss": 4.3744,
+      "step": 1222
+    },
+    {
+      "epoch": 0.01223,
+      "grad_norm": 0.4544260730538049,
+      "learning_rate": 0.003,
+      "loss": 4.3669,
+      "step": 1223
+    },
+    {
+      "epoch": 0.01224,
+      "grad_norm": 0.4498657047747612,
+      "learning_rate": 0.003,
+      "loss": 4.3528,
+      "step": 1224
+    },
+    {
+      "epoch": 0.01225,
+      "grad_norm": 0.44954702929076595,
+      "learning_rate": 0.003,
+      "loss": 4.3751,
+      "step": 1225
+    },
+    {
+      "epoch": 0.01226,
+      "grad_norm": 0.45922987677597427,
+      "learning_rate": 0.003,
+      "loss": 4.3678,
+      "step": 1226
+    },
+    {
+      "epoch": 0.01227,
+      "grad_norm": 0.444736072897216,
+      "learning_rate": 0.003,
+      "loss": 4.3955,
+      "step": 1227
+    },
+    {
+      "epoch": 0.01228,
+      "grad_norm": 0.39964234776344315,
+      "learning_rate": 0.003,
+      "loss": 4.3749,
+      "step": 1228
+    },
+    {
+      "epoch": 0.01229,
+      "grad_norm": 0.4304709469798317,
+      "learning_rate": 0.003,
+      "loss": 4.3636,
+      "step": 1229
+    },
+    {
+      "epoch": 0.0123,
+      "grad_norm": 0.45096371519587236,
+      "learning_rate": 0.003,
+      "loss": 4.3946,
+      "step": 1230
+    },
+    {
+      "epoch": 0.01231,
+      "grad_norm": 0.45260016324469293,
+      "learning_rate": 0.003,
+      "loss": 4.3662,
+      "step": 1231
+    },
+    {
+      "epoch": 0.01232,
+      "grad_norm": 0.5056071407476121,
+      "learning_rate": 0.003,
+      "loss": 4.3416,
+      "step": 1232
+    },
+    {
+      "epoch": 0.01233,
+      "grad_norm": 0.6137961234604853,
+      "learning_rate": 0.003,
+      "loss": 4.3562,
+      "step": 1233
+    },
+    {
+      "epoch": 0.01234,
+      "grad_norm": 0.7580924108092045,
+      "learning_rate": 0.003,
+      "loss": 4.3771,
+      "step": 1234
+    },
+    {
+      "epoch": 0.01235,
+      "grad_norm": 0.7682739117958468,
+      "learning_rate": 0.003,
+      "loss": 4.377,
+      "step": 1235
+    },
+    {
+      "epoch": 0.01236,
+      "grad_norm": 0.7726095308103765,
+      "learning_rate": 0.003,
+      "loss": 4.379,
+      "step": 1236
+    },
+    {
+      "epoch": 0.01237,
+      "grad_norm": 0.6515520396776402,
+      "learning_rate": 0.003,
+      "loss": 4.3868,
+      "step": 1237
+    },
+    {
+      "epoch": 0.01238,
+      "grad_norm": 0.5911401797128396,
+      "learning_rate": 0.003,
+      "loss": 4.3811,
+      "step": 1238
+    },
+    {
+      "epoch": 0.01239,
+      "grad_norm": 0.6868743136698521,
+      "learning_rate": 0.003,
+      "loss": 4.3689,
+      "step": 1239
+    },
+    {
+      "epoch": 0.0124,
+      "grad_norm": 0.7235342138686123,
+      "learning_rate": 0.003,
+      "loss": 4.3578,
+      "step": 1240
+    },
+    {
+      "epoch": 0.01241,
+      "grad_norm": 0.7086323715527123,
+      "learning_rate": 0.003,
+      "loss": 4.3714,
+      "step": 1241
+    },
+    {
+      "epoch": 0.01242,
+      "grad_norm": 0.7273279516261576,
+      "learning_rate": 0.003,
+      "loss": 4.3772,
+      "step": 1242
+    },
+    {
+      "epoch": 0.01243,
+      "grad_norm": 0.7241329186524991,
+      "learning_rate": 0.003,
+      "loss": 4.358,
+      "step": 1243
+    },
+    {
+      "epoch": 0.01244,
+      "grad_norm": 0.659589685877625,
+      "learning_rate": 0.003,
+      "loss": 4.3697,
+      "step": 1244
+    },
+    {
+      "epoch": 0.01245,
+      "grad_norm": 0.5679488489092839,
+      "learning_rate": 0.003,
+      "loss": 4.3588,
+      "step": 1245
+    },
+    {
+      "epoch": 0.01246,
+      "grad_norm": 0.5843246373122641,
+      "learning_rate": 0.003,
+      "loss": 4.3356,
+      "step": 1246
+    },
+    {
+      "epoch": 0.01247,
+      "grad_norm": 0.5386207185407424,
+      "learning_rate": 0.003,
+      "loss": 4.3884,
+      "step": 1247
+    },
+    {
+      "epoch": 0.01248,
+      "grad_norm": 0.4947062329390487,
+      "learning_rate": 0.003,
+      "loss": 4.3718,
+      "step": 1248
+    },
+    {
+      "epoch": 0.01249,
+      "grad_norm": 0.47539230004779465,
+      "learning_rate": 0.003,
+      "loss": 4.3597,
+      "step": 1249
+    },
+    {
+      "epoch": 0.0125,
+      "grad_norm": 0.4876155686681362,
+      "learning_rate": 0.003,
+      "loss": 4.3353,
+      "step": 1250
+    },
+    {
+      "epoch": 0.01251,
+      "grad_norm": 0.587984128950104,
+      "learning_rate": 0.003,
+      "loss": 4.3725,
+      "step": 1251
+    },
+    {
+      "epoch": 0.01252,
+      "grad_norm": 0.74609447185642,
+      "learning_rate": 0.003,
+      "loss": 4.3585,
+      "step": 1252
+    },
+    {
+      "epoch": 0.01253,
+      "grad_norm": 0.8297104750282515,
+      "learning_rate": 0.003,
+      "loss": 4.3426,
+      "step": 1253
+    },
+    {
+      "epoch": 0.01254,
+      "grad_norm": 0.6714602053035584,
+      "learning_rate": 0.003,
+      "loss": 4.3879,
+      "step": 1254
+    },
+    {
+      "epoch": 0.01255,
+      "grad_norm": 0.5575284697102026,
+      "learning_rate": 0.003,
+      "loss": 4.3328,
+      "step": 1255
+    },
+    {
+      "epoch": 0.01256,
+      "grad_norm": 0.6531801875316168,
+      "learning_rate": 0.003,
+      "loss": 4.3617,
+      "step": 1256
+    },
+    {
+      "epoch": 0.01257,
+      "grad_norm": 0.6725833064810157,
+      "learning_rate": 0.003,
+      "loss": 4.3523,
+      "step": 1257
+    },
+    {
+      "epoch": 0.01258,
+      "grad_norm": 0.5415379584452896,
+      "learning_rate": 0.003,
+      "loss": 4.3533,
+      "step": 1258
+    },
+    {
+      "epoch": 0.01259,
+      "grad_norm": 0.6189096611706483,
+      "learning_rate": 0.003,
+      "loss": 4.3565,
+      "step": 1259
+    },
+    {
+      "epoch": 0.0126,
+      "grad_norm": 0.7719709473721972,
+      "learning_rate": 0.003,
+      "loss": 4.3943,
+      "step": 1260
+    },
+    {
+      "epoch": 0.01261,
+      "grad_norm": 0.7141309260632757,
+      "learning_rate": 0.003,
+      "loss": 4.3828,
+      "step": 1261
+    },
+    {
+      "epoch": 0.01262,
+      "grad_norm": 0.82201580215985,
+      "learning_rate": 0.003,
+      "loss": 4.3803,
+      "step": 1262
+    },
+    {
+      "epoch": 0.01263,
+      "grad_norm": 0.7648394097891381,
+      "learning_rate": 0.003,
+      "loss": 4.3764,
+      "step": 1263
+    },
+    {
+      "epoch": 0.01264,
+      "grad_norm": 0.6534936022081808,
+      "learning_rate": 0.003,
+      "loss": 4.3615,
+      "step": 1264
+    },
+    {
+      "epoch": 0.01265,
+      "grad_norm": 0.8083064566897021,
+      "learning_rate": 0.003,
+      "loss": 4.3795,
+      "step": 1265
+    },
+    {
+      "epoch": 0.01266,
+      "grad_norm": 0.7625627444580204,
+      "learning_rate": 0.003,
+      "loss": 4.3817,
+      "step": 1266
+    },
+    {
+      "epoch": 0.01267,
+      "grad_norm": 0.7144669763786441,
+      "learning_rate": 0.003,
+      "loss": 4.3961,
+      "step": 1267
+    },
+    {
+      "epoch": 0.01268,
+      "grad_norm": 0.6176763363932604,
+      "learning_rate": 0.003,
+      "loss": 4.3815,
+      "step": 1268
+    },
+    {
+      "epoch": 0.01269,
+      "grad_norm": 0.5844219741679575,
+      "learning_rate": 0.003,
+      "loss": 4.3442,
+      "step": 1269
+    },
+    {
+      "epoch": 0.0127,
+      "grad_norm": 0.5570920646541462,
+      "learning_rate": 0.003,
+      "loss": 4.3764,
+      "step": 1270
+    },
+    {
+      "epoch": 0.01271,
+      "grad_norm": 0.665679672510672,
+      "learning_rate": 0.003,
+      "loss": 4.3628,
+      "step": 1271
+    },
+    {
+      "epoch": 0.01272,
+      "grad_norm": 0.7700987517359916,
+      "learning_rate": 0.003,
+      "loss": 4.3638,
+      "step": 1272
+    },
+    {
+      "epoch": 0.01273,
+      "grad_norm": 0.7473118276250652,
+      "learning_rate": 0.003,
+      "loss": 4.3932,
+      "step": 1273
+    },
+    {
+      "epoch": 0.01274,
+      "grad_norm": 0.6491851354934152,
+      "learning_rate": 0.003,
+      "loss": 4.3789,
+      "step": 1274
+    },
+    {
+      "epoch": 0.01275,
+      "grad_norm": 0.6744635149292406,
+      "learning_rate": 0.003,
+      "loss": 4.3494,
+      "step": 1275
+    },
+    {
+      "epoch": 0.01276,
+      "grad_norm": 0.652401206080157,
+      "learning_rate": 0.003,
+      "loss": 4.3572,
+      "step": 1276
+    },
+    {
+      "epoch": 0.01277,
+      "grad_norm": 0.8226392169345714,
+      "learning_rate": 0.003,
+      "loss": 4.3706,
+      "step": 1277
+    },
+    {
+      "epoch": 0.01278,
+      "grad_norm": 0.8325255271008436,
+      "learning_rate": 0.003,
+      "loss": 4.3906,
+      "step": 1278
+    },
+    {
+      "epoch": 0.01279,
+      "grad_norm": 0.7029576010205191,
+      "learning_rate": 0.003,
+      "loss": 4.3612,
+      "step": 1279
+    },
+    {
+      "epoch": 0.0128,
+      "grad_norm": 0.6056600803694466,
+      "learning_rate": 0.003,
+      "loss": 4.3548,
+      "step": 1280
+    },
+    {
+      "epoch": 0.01281,
+      "grad_norm": 0.6068824045406035,
+      "learning_rate": 0.003,
+      "loss": 4.3433,
+      "step": 1281
+    },
+    {
+      "epoch": 0.01282,
+      "grad_norm": 0.5449477323585487,
+      "learning_rate": 0.003,
+      "loss": 4.3801,
+      "step": 1282
+    },
+    {
+      "epoch": 0.01283,
+      "grad_norm": 0.5086376593278678,
+      "learning_rate": 0.003,
+      "loss": 4.3465,
+      "step": 1283
+    },
+    {
+      "epoch": 0.01284,
+      "grad_norm": 0.49507526795006285,
+      "learning_rate": 0.003,
+      "loss": 4.3484,
+      "step": 1284
+    },
+    {
+      "epoch": 0.01285,
+      "grad_norm": 0.4794063886412805,
+      "learning_rate": 0.003,
+      "loss": 4.3499,
+      "step": 1285
+    },
+    {
+      "epoch": 0.01286,
+      "grad_norm": 0.4314216967223394,
+      "learning_rate": 0.003,
+      "loss": 4.37,
+      "step": 1286
+    },
+    {
+      "epoch": 0.01287,
+      "grad_norm": 0.43751633400555473,
+      "learning_rate": 0.003,
+      "loss": 4.3355,
+      "step": 1287
+    },
+    {
+      "epoch": 0.01288,
+      "grad_norm": 0.4402964980769675,
+      "learning_rate": 0.003,
+      "loss": 4.3477,
+      "step": 1288
+    },
+    {
+      "epoch": 0.01289,
+      "grad_norm": 0.4330370123851316,
+      "learning_rate": 0.003,
+      "loss": 4.34,
+      "step": 1289
+    },
+    {
+      "epoch": 0.0129,
+      "grad_norm": 0.4577360505134826,
+      "learning_rate": 0.003,
+      "loss": 4.3613,
+      "step": 1290
+    },
+    {
+      "epoch": 0.01291,
+      "grad_norm": 0.4908761222350223,
+      "learning_rate": 0.003,
+      "loss": 4.3434,
+      "step": 1291
+    },
+    {
+      "epoch": 0.01292,
+      "grad_norm": 0.5197962656056466,
+      "learning_rate": 0.003,
+      "loss": 4.3769,
+      "step": 1292
+    },
+    {
+      "epoch": 0.01293,
+      "grad_norm": 0.5249407732060195,
+      "learning_rate": 0.003,
+      "loss": 4.3636,
+      "step": 1293
+    },
+    {
+      "epoch": 0.01294,
+      "grad_norm": 0.611698986416238,
+      "learning_rate": 0.003,
+      "loss": 4.3337,
+      "step": 1294
+    },
+    {
+      "epoch": 0.01295,
+      "grad_norm": 0.7156402524759599,
+      "learning_rate": 0.003,
+      "loss": 4.369,
+      "step": 1295
+    },
+    {
+      "epoch": 0.01296,
+      "grad_norm": 0.8400871327514794,
+      "learning_rate": 0.003,
+      "loss": 4.3429,
+      "step": 1296
+    },
+    {
+      "epoch": 0.01297,
+      "grad_norm": 0.9437097734689512,
+      "learning_rate": 0.003,
+      "loss": 4.3707,
+      "step": 1297
+    },
+    {
+      "epoch": 0.01298,
+      "grad_norm": 0.9584374674542437,
+      "learning_rate": 0.003,
+      "loss": 4.3505,
+      "step": 1298
+    },
+    {
+      "epoch": 0.01299,
+      "grad_norm": 0.8671660761915297,
+      "learning_rate": 0.003,
+      "loss": 4.3625,
+      "step": 1299
+    },
+    {
+      "epoch": 0.013,
+      "grad_norm": 0.7502528073695865,
+      "learning_rate": 0.003,
+      "loss": 4.3495,
+      "step": 1300
+    },
+    {
+      "epoch": 0.01301,
+      "grad_norm": 0.7063515582852367,
+      "learning_rate": 0.003,
+      "loss": 4.3619,
+      "step": 1301
+    },
+    {
+      "epoch": 0.01302,
+      "grad_norm": 0.6149436229664508,
+      "learning_rate": 0.003,
+      "loss": 4.4034,
+      "step": 1302
+    },
+    {
+      "epoch": 0.01303,
+      "grad_norm": 0.6398205689943912,
+      "learning_rate": 0.003,
+      "loss": 4.3511,
+      "step": 1303
+    },
+    {
+      "epoch": 0.01304,
+      "grad_norm": 0.5125654671319883,
+      "learning_rate": 0.003,
+      "loss": 4.3743,
+      "step": 1304
+    },
+    {
+      "epoch": 0.01305,
+      "grad_norm": 0.49151327704885306,
+      "learning_rate": 0.003,
+      "loss": 4.357,
+      "step": 1305
+    },
+    {
+      "epoch": 0.01306,
+      "grad_norm": 0.449035259984806,
+      "learning_rate": 0.003,
+      "loss": 4.3496,
+      "step": 1306
+    },
+    {
+      "epoch": 0.01307,
+      "grad_norm": 0.3786564732477656,
+      "learning_rate": 0.003,
+      "loss": 4.365,
+      "step": 1307
+    },
+    {
+      "epoch": 0.01308,
+      "grad_norm": 0.40747339154306583,
+      "learning_rate": 0.003,
+      "loss": 4.3407,
+      "step": 1308
+    },
+    {
+      "epoch": 0.01309,
+      "grad_norm": 0.3477777170477395,
+      "learning_rate": 0.003,
+      "loss": 4.3771,
+      "step": 1309
+    },
+    {
+      "epoch": 0.0131,
+      "grad_norm": 0.33985557007437,
+      "learning_rate": 0.003,
+      "loss": 4.3221,
+      "step": 1310
+    },
+    {
+      "epoch": 0.01311,
+      "grad_norm": 0.3781744690142208,
+      "learning_rate": 0.003,
+      "loss": 4.3621,
+      "step": 1311
+    },
+    {
+      "epoch": 0.01312,
+      "grad_norm": 0.390430095045798,
+      "learning_rate": 0.003,
+      "loss": 4.3295,
+      "step": 1312
+    },
+    {
+      "epoch": 0.01313,
+      "grad_norm": 0.4411339748954226,
+      "learning_rate": 0.003,
+      "loss": 4.3219,
+      "step": 1313
+    },
+    {
+      "epoch": 0.01314,
+      "grad_norm": 0.4398621851217481,
+      "learning_rate": 0.003,
+      "loss": 4.3182,
+      "step": 1314
+    },
+    {
+      "epoch": 0.01315,
+      "grad_norm": 0.4543520117536255,
+      "learning_rate": 0.003,
+      "loss": 4.3454,
+      "step": 1315
+    },
+    {
+      "epoch": 0.01316,
+      "grad_norm": 0.4647119001484631,
+      "learning_rate": 0.003,
+      "loss": 4.3315,
+      "step": 1316
+    },
+    {
+      "epoch": 0.01317,
+      "grad_norm": 0.6634726948283797,
+      "learning_rate": 0.003,
+      "loss": 4.3607,
+      "step": 1317
+    },
+    {
+      "epoch": 0.01318,
+      "grad_norm": 1.0496539085022096,
+      "learning_rate": 0.003,
+      "loss": 4.3666,
+      "step": 1318
+    },
+    {
+      "epoch": 0.01319,
+      "grad_norm": 1.3165114729988252,
+      "learning_rate": 0.003,
+      "loss": 4.392,
+      "step": 1319
+    },
+    {
+      "epoch": 0.0132,
+      "grad_norm": 0.7650097283623507,
+      "learning_rate": 0.003,
+      "loss": 4.3723,
+      "step": 1320
+    },
+    {
+      "epoch": 0.01321,
+      "grad_norm": 0.7692901736546863,
+      "learning_rate": 0.003,
+      "loss": 4.3563,
+      "step": 1321
+    },
+    {
+      "epoch": 0.01322,
+      "grad_norm": 0.8430579356733313,
+      "learning_rate": 0.003,
+      "loss": 4.3811,
+      "step": 1322
+    },
+    {
+      "epoch": 0.01323,
+      "grad_norm": 0.7787983772827455,
+      "learning_rate": 0.003,
+      "loss": 4.3749,
+      "step": 1323
+    },
+    {
+      "epoch": 0.01324,
+      "grad_norm": 0.7585640630062677,
+      "learning_rate": 0.003,
+      "loss": 4.3595,
+      "step": 1324
+    },
+    {
+      "epoch": 0.01325,
+      "grad_norm": 0.662871102926864,
+      "learning_rate": 0.003,
+      "loss": 4.3604,
+      "step": 1325
+    },
+    {
+      "epoch": 0.01326,
+      "grad_norm": 0.6552542644296041,
+      "learning_rate": 0.003,
+      "loss": 4.3471,
+      "step": 1326
+    },
+    {
+      "epoch": 0.01327,
+      "grad_norm": 0.6597442221568894,
+      "learning_rate": 0.003,
+      "loss": 4.3636,
+      "step": 1327
+    },
+    {
+      "epoch": 0.01328,
+      "grad_norm": 0.6991774516177431,
+      "learning_rate": 0.003,
+      "loss": 4.3572,
+      "step": 1328
+    },
+    {
+      "epoch": 0.01329,
+      "grad_norm": 0.7004568547673504,
+      "learning_rate": 0.003,
+      "loss": 4.323,
+      "step": 1329
+    },
+    {
+      "epoch": 0.0133,
+      "grad_norm": 0.6451796491929548,
+      "learning_rate": 0.003,
+      "loss": 4.3619,
+      "step": 1330
+    },
+    {
+      "epoch": 0.01331,
+      "grad_norm": 0.5603586741834496,
+      "learning_rate": 0.003,
+      "loss": 4.3461,
+      "step": 1331
+    },
+    {
+      "epoch": 0.01332,
+      "grad_norm": 0.6118296137650349,
+      "learning_rate": 0.003,
+      "loss": 4.3674,
+      "step": 1332
+    },
+    {
+      "epoch": 0.01333,
+      "grad_norm": 0.6556746462798643,
+      "learning_rate": 0.003,
+      "loss": 4.3385,
+      "step": 1333
+    },
+    {
+      "epoch": 0.01334,
+      "grad_norm": 0.6569171248898434,
+      "learning_rate": 0.003,
+      "loss": 4.3292,
+      "step": 1334
+    },
+    {
+      "epoch": 0.01335,
+      "grad_norm": 0.7308490257197903,
+      "learning_rate": 0.003,
+      "loss": 4.355,
+      "step": 1335
+    },
+    {
+      "epoch": 0.01336,
+      "grad_norm": 0.7694263760249205,
+      "learning_rate": 0.003,
+      "loss": 4.3562,
+      "step": 1336
+    },
+    {
+      "epoch": 0.01337,
+      "grad_norm": 0.829484027358136,
+      "learning_rate": 0.003,
+      "loss": 4.3739,
+      "step": 1337
+    },
+    {
+      "epoch": 0.01338,
+      "grad_norm": 0.7394720058707402,
+      "learning_rate": 0.003,
+      "loss": 4.3375,
+      "step": 1338
+    },
+    {
+      "epoch": 0.01339,
+      "grad_norm": 0.7663877331658486,
+      "learning_rate": 0.003,
+      "loss": 4.3614,
+      "step": 1339
+    },
+    {
+      "epoch": 0.0134,
+      "grad_norm": 0.6814802672525545,
+      "learning_rate": 0.003,
+      "loss": 4.3431,
+      "step": 1340
+    },
+    {
+      "epoch": 0.01341,
+      "grad_norm": 0.5804799698607723,
+      "learning_rate": 0.003,
+      "loss": 4.3566,
+      "step": 1341
+    },
+    {
+      "epoch": 0.01342,
+      "grad_norm": 0.5489998904660267,
+      "learning_rate": 0.003,
+      "loss": 4.3242,
+      "step": 1342
+    },
+    {
+      "epoch": 0.01343,
+      "grad_norm": 0.5528070757921824,
+      "learning_rate": 0.003,
+      "loss": 4.3308,
+      "step": 1343
+    },
+    {
+      "epoch": 0.01344,
+      "grad_norm": 0.5249427916401641,
+      "learning_rate": 0.003,
+      "loss": 4.3286,
+      "step": 1344
+    },
+    {
+      "epoch": 0.01345,
+      "grad_norm": 0.5796492875823627,
+      "learning_rate": 0.003,
+      "loss": 4.3698,
+      "step": 1345
+    },
+    {
+      "epoch": 0.01346,
+      "grad_norm": 0.7791779206443923,
+      "learning_rate": 0.003,
+      "loss": 4.349,
+      "step": 1346
+    },
+    {
+      "epoch": 0.01347,
+      "grad_norm": 0.9787862373774305,
+      "learning_rate": 0.003,
+      "loss": 4.3601,
+      "step": 1347
+    },
+    {
+      "epoch": 0.01348,
+      "grad_norm": 1.124691751963475,
+      "learning_rate": 0.003,
+      "loss": 4.3695,
+      "step": 1348
+    },
+    {
+      "epoch": 0.01349,
+      "grad_norm": 0.7239534112442599,
+      "learning_rate": 0.003,
+      "loss": 4.3703,
+      "step": 1349
+    },
+    {
+      "epoch": 0.0135,
+      "grad_norm": 0.7058824966707771,
+      "learning_rate": 0.003,
+      "loss": 4.3665,
+      "step": 1350
+    },
+    {
+      "epoch": 0.01351,
+      "grad_norm": 0.7269767228946674,
+      "learning_rate": 0.003,
+      "loss": 4.3913,
+      "step": 1351
+    },
+    {
+      "epoch": 0.01352,
+      "grad_norm": 0.629762174967035,
+      "learning_rate": 0.003,
+      "loss": 4.3652,
+      "step": 1352
+    },
+    {
+      "epoch": 0.01353,
+      "grad_norm": 0.46709847900974316,
+      "learning_rate": 0.003,
+      "loss": 4.3426,
+      "step": 1353
+    },
+    {
+      "epoch": 0.01354,
+      "grad_norm": 0.4348570974243878,
+      "learning_rate": 0.003,
+      "loss": 4.33,
+      "step": 1354
+    },
+    {
+      "epoch": 0.01355,
+      "grad_norm": 0.4058626528377844,
+      "learning_rate": 0.003,
+      "loss": 4.3397,
+      "step": 1355
+    },
+    {
+      "epoch": 0.01356,
+      "grad_norm": 0.4077904728688183,
+      "learning_rate": 0.003,
+      "loss": 4.3329,
+      "step": 1356
+    },
+    {
+      "epoch": 0.01357,
+      "grad_norm": 0.3881492908104237,
+      "learning_rate": 0.003,
+      "loss": 4.3196,
+      "step": 1357
+    },
+    {
+      "epoch": 0.01358,
+      "grad_norm": 0.37883341535998916,
+      "learning_rate": 0.003,
+      "loss": 4.3135,
+      "step": 1358
+    },
+    {
+      "epoch": 0.01359,
+      "grad_norm": 0.3673079033329598,
+      "learning_rate": 0.003,
+      "loss": 4.3515,
+      "step": 1359
+    },
+    {
+      "epoch": 0.0136,
+      "grad_norm": 0.37976733712994504,
+      "learning_rate": 0.003,
+      "loss": 4.3177,
+      "step": 1360
+    },
+    {
+      "epoch": 0.01361,
+      "grad_norm": 0.4118348378412375,
+      "learning_rate": 0.003,
+      "loss": 4.3313,
+      "step": 1361
+    },
+    {
+      "epoch": 0.01362,
+      "grad_norm": 0.4407593683397167,
+      "learning_rate": 0.003,
+      "loss": 4.345,
+      "step": 1362
+    },
+    {
+      "epoch": 0.01363,
+      "grad_norm": 0.5058671106167254,
+      "learning_rate": 0.003,
+      "loss": 4.3464,
+      "step": 1363
+    },
+    {
+      "epoch": 0.01364,
+      "grad_norm": 0.6807238620694973,
+      "learning_rate": 0.003,
+      "loss": 4.3446,
+      "step": 1364
+    },
+    {
+      "epoch": 0.01365,
+      "grad_norm": 0.8032111226209311,
+      "learning_rate": 0.003,
+      "loss": 4.3702,
+      "step": 1365
+    },
+    {
+      "epoch": 0.01366,
+      "grad_norm": 0.7976336994683142,
+      "learning_rate": 0.003,
+      "loss": 4.3233,
+      "step": 1366
+    },
+    {
+      "epoch": 0.01367,
+      "grad_norm": 0.7881501972797743,
+      "learning_rate": 0.003,
+      "loss": 4.3766,
+      "step": 1367
+    },
+    {
+      "epoch": 0.01368,
+      "grad_norm": 0.6437176795201012,
+      "learning_rate": 0.003,
+      "loss": 4.346,
+      "step": 1368
+    },
+    {
+      "epoch": 0.01369,
+      "grad_norm": 0.7544973412872852,
+      "learning_rate": 0.003,
+      "loss": 4.3393,
+      "step": 1369
+    },
+    {
+      "epoch": 0.0137,
+      "grad_norm": 0.9164166166999201,
+      "learning_rate": 0.003,
+      "loss": 4.3794,
+      "step": 1370
+    },
+    {
+      "epoch": 0.01371,
+      "grad_norm": 0.9303263432444765,
+      "learning_rate": 0.003,
+      "loss": 4.3692,
+      "step": 1371
+    },
+    {
+      "epoch": 0.01372,
+      "grad_norm": 0.7939151912920627,
+      "learning_rate": 0.003,
+      "loss": 4.3176,
+      "step": 1372
+    },
+    {
+      "epoch": 0.01373,
+      "grad_norm": 0.8088750323088586,
+      "learning_rate": 0.003,
+      "loss": 4.3626,
+      "step": 1373
+    },
+    {
+      "epoch": 0.01374,
+      "grad_norm": 0.6721224672262793,
+      "learning_rate": 0.003,
+      "loss": 4.3585,
+      "step": 1374
+    },
+    {
+      "epoch": 0.01375,
+      "grad_norm": 0.6771097427365759,
+      "learning_rate": 0.003,
+      "loss": 4.3249,
+      "step": 1375
+    },
+    {
+      "epoch": 0.01376,
+      "grad_norm": 0.6762752470113543,
+      "learning_rate": 0.003,
+      "loss": 4.347,
+      "step": 1376
+    },
+    {
+      "epoch": 0.01377,
+      "grad_norm": 0.6604690846076395,
+      "learning_rate": 0.003,
+      "loss": 4.3466,
+      "step": 1377
+    },
+    {
+      "epoch": 0.01378,
+      "grad_norm": 0.8027631418948692,
+      "learning_rate": 0.003,
+      "loss": 4.3812,
+      "step": 1378
+    },
+    {
+      "epoch": 0.01379,
+      "grad_norm": 0.8058991494927565,
+      "learning_rate": 0.003,
+      "loss": 4.3543,
+      "step": 1379
+    },
+    {
+      "epoch": 0.0138,
+      "grad_norm": 0.6522516144796072,
+      "learning_rate": 0.003,
+      "loss": 4.3459,
+      "step": 1380
+    },
+    {
+      "epoch": 0.01381,
+      "grad_norm": 0.6569912385058241,
+      "learning_rate": 0.003,
+      "loss": 4.3524,
+      "step": 1381
+    },
+    {
+      "epoch": 0.01382,
+      "grad_norm": 0.5996275495095534,
+      "learning_rate": 0.003,
+      "loss": 4.3471,
+      "step": 1382
+    },
+    {
+      "epoch": 0.01383,
+      "grad_norm": 0.7099037071072233,
+      "learning_rate": 0.003,
+      "loss": 4.3442,
+      "step": 1383
+    },
+    {
+      "epoch": 0.01384,
+      "grad_norm": 0.9143332820449893,
+      "learning_rate": 0.003,
+      "loss": 4.3854,
+      "step": 1384
+    },
+    {
+      "epoch": 0.01385,
+      "grad_norm": 1.0346171765908472,
+      "learning_rate": 0.003,
+      "loss": 4.3815,
+      "step": 1385
+    },
+    {
+      "epoch": 0.01386,
+      "grad_norm": 0.8985740978326894,
+      "learning_rate": 0.003,
+      "loss": 4.3558,
+      "step": 1386
+    },
+    {
+      "epoch": 0.01387,
+      "grad_norm": 0.6958974730086266,
+      "learning_rate": 0.003,
+      "loss": 4.335,
+      "step": 1387
+    },
+    {
+      "epoch": 0.01388,
+      "grad_norm": 0.7497022428211844,
+      "learning_rate": 0.003,
+      "loss": 4.3735,
+      "step": 1388
+    },
+    {
+      "epoch": 0.01389,
+      "grad_norm": 0.7334888360507559,
+      "learning_rate": 0.003,
+      "loss": 4.3541,
+      "step": 1389
+    },
+    {
+      "epoch": 0.0139,
+      "grad_norm": 0.6634716599193375,
+      "learning_rate": 0.003,
+      "loss": 4.3533,
+      "step": 1390
+    },
+    {
+      "epoch": 0.01391,
+      "grad_norm": 0.5884420459817105,
+      "learning_rate": 0.003,
+      "loss": 4.3483,
+      "step": 1391
+    },
+    {
+      "epoch": 0.01392,
+      "grad_norm": 0.536202889592441,
+      "learning_rate": 0.003,
+      "loss": 4.3599,
+      "step": 1392
+    },
+    {
+      "epoch": 0.01393,
+      "grad_norm": 0.5041805750347179,
+      "learning_rate": 0.003,
+      "loss": 4.327,
+      "step": 1393
+    },
+    {
+      "epoch": 0.01394,
+      "grad_norm": 0.46187226586984687,
+      "learning_rate": 0.003,
+      "loss": 4.3118,
+      "step": 1394
+    },
+    {
+      "epoch": 0.01395,
+      "grad_norm": 0.4642311356997612,
+      "learning_rate": 0.003,
+      "loss": 4.343,
+      "step": 1395
+    },
+    {
+      "epoch": 0.01396,
+      "grad_norm": 0.49219234547480073,
+      "learning_rate": 0.003,
+      "loss": 4.3333,
+      "step": 1396
+    },
+    {
+      "epoch": 0.01397,
+      "grad_norm": 0.535253115362672,
+      "learning_rate": 0.003,
+      "loss": 4.3471,
+      "step": 1397
+    },
+    {
+      "epoch": 0.01398,
+      "grad_norm": 0.6052714387106423,
+      "learning_rate": 0.003,
+      "loss": 4.3267,
+      "step": 1398
+    },
+    {
+      "epoch": 0.01399,
+      "grad_norm": 0.6277477924110956,
+      "learning_rate": 0.003,
+      "loss": 4.3504,
+      "step": 1399
+    },
+    {
+      "epoch": 0.014,
+      "grad_norm": 0.6846592906181149,
+      "learning_rate": 0.003,
+      "loss": 4.3103,
+      "step": 1400
+    },
+    {
+      "epoch": 0.01401,
+      "grad_norm": 0.8200984983857136,
+      "learning_rate": 0.003,
+      "loss": 4.3387,
+      "step": 1401
+    },
+    {
+      "epoch": 0.01402,
+      "grad_norm": 0.8742108124137586,
+      "learning_rate": 0.003,
+      "loss": 4.3557,
+      "step": 1402
+    },
+    {
+      "epoch": 0.01403,
+      "grad_norm": 0.7096844776193044,
+      "learning_rate": 0.003,
+      "loss": 4.3233,
+      "step": 1403
+    },
+    {
+      "epoch": 0.01404,
+      "grad_norm": 0.5388679123577896,
+      "learning_rate": 0.003,
+      "loss": 4.3401,
+      "step": 1404
+    },
+    {
+      "epoch": 0.01405,
+      "grad_norm": 0.4910540359683292,
+      "learning_rate": 0.003,
+      "loss": 4.3559,
+      "step": 1405
+    },
+    {
+      "epoch": 0.01406,
+      "grad_norm": 0.5213447941957257,
+      "learning_rate": 0.003,
+      "loss": 4.3366,
+      "step": 1406
+    },
+    {
+      "epoch": 0.01407,
+      "grad_norm": 0.555926299222398,
+      "learning_rate": 0.003,
+      "loss": 4.3521,
+      "step": 1407
+    },
+    {
+      "epoch": 0.01408,
+      "grad_norm": 0.692327807092523,
+      "learning_rate": 0.003,
+      "loss": 4.3137,
+      "step": 1408
+    },
+    {
+      "epoch": 0.01409,
+      "grad_norm": 0.8697281888439876,
+      "learning_rate": 0.003,
+      "loss": 4.3588,
+      "step": 1409
+    },
+    {
+      "epoch": 0.0141,
+      "grad_norm": 0.9566955718451174,
+      "learning_rate": 0.003,
+      "loss": 4.3446,
+      "step": 1410
+    },
+    {
+      "epoch": 0.01411,
+      "grad_norm": 0.7215244847418736,
+      "learning_rate": 0.003,
+      "loss": 4.3131,
+      "step": 1411
+    },
+    {
+      "epoch": 0.01412,
+      "grad_norm": 0.6315141391735041,
+      "learning_rate": 0.003,
+      "loss": 4.3535,
+      "step": 1412
+    },
+    {
+      "epoch": 0.01413,
+      "grad_norm": 0.6841381857718712,
+      "learning_rate": 0.003,
+      "loss": 4.342,
+      "step": 1413
+    },
+    {
+      "epoch": 0.01414,
+      "grad_norm": 0.5798611042820683,
+      "learning_rate": 0.003,
+      "loss": 4.3235,
+      "step": 1414
+    },
+    {
+      "epoch": 0.01415,
+      "grad_norm": 0.547398429036414,
+      "learning_rate": 0.003,
+      "loss": 4.3157,
+      "step": 1415
+    },
+    {
+      "epoch": 0.01416,
+      "grad_norm": 0.532688568072748,
+      "learning_rate": 0.003,
+      "loss": 4.3388,
+      "step": 1416
+    },
+    {
+      "epoch": 0.01417,
+      "grad_norm": 0.5474366342351658,
+      "learning_rate": 0.003,
+      "loss": 4.3317,
+      "step": 1417
+    },
+    {
+      "epoch": 0.01418,
+      "grad_norm": 0.5499547169322332,
+      "learning_rate": 0.003,
+      "loss": 4.3477,
+      "step": 1418
+    },
+    {
+      "epoch": 0.01419,
+      "grad_norm": 0.58527195779541,
+      "learning_rate": 0.003,
+      "loss": 4.3403,
+      "step": 1419
+    },
+    {
+      "epoch": 0.0142,
+      "grad_norm": 0.4513568933907034,
+      "learning_rate": 0.003,
+      "loss": 4.3401,
+      "step": 1420
+    },
+    {
+      "epoch": 0.01421,
+      "grad_norm": 0.433334289258551,
+      "learning_rate": 0.003,
+      "loss": 4.3072,
+      "step": 1421
+    },
+    {
+      "epoch": 0.01422,
+      "grad_norm": 0.39636752625840327,
+      "learning_rate": 0.003,
+      "loss": 4.3095,
+      "step": 1422
+    },
+    {
+      "epoch": 0.01423,
+      "grad_norm": 0.38308187149456013,
+      "learning_rate": 0.003,
+      "loss": 4.3141,
+      "step": 1423
+    },
+    {
+      "epoch": 0.01424,
+      "grad_norm": 0.3793175762281477,
+      "learning_rate": 0.003,
+      "loss": 4.3274,
+      "step": 1424
+    },
+    {
+      "epoch": 0.01425,
+      "grad_norm": 0.4426704955835019,
+      "learning_rate": 0.003,
+      "loss": 4.334,
+      "step": 1425
+    },
+    {
+      "epoch": 0.01426,
+      "grad_norm": 0.5006373442027768,
+      "learning_rate": 0.003,
+      "loss": 4.3106,
+      "step": 1426
+    },
+    {
+      "epoch": 0.01427,
+      "grad_norm": 0.654459732986379,
+      "learning_rate": 0.003,
+      "loss": 4.3121,
+      "step": 1427
+    },
+    {
+      "epoch": 0.01428,
+      "grad_norm": 0.8067469794918197,
+      "learning_rate": 0.003,
+      "loss": 4.3172,
+      "step": 1428
+    },
+    {
+      "epoch": 0.01429,
+      "grad_norm": 0.8729241550485854,
+      "learning_rate": 0.003,
+      "loss": 4.3306,
+      "step": 1429
+    },
+    {
+      "epoch": 0.0143,
+      "grad_norm": 0.7278485959649278,
+      "learning_rate": 0.003,
+      "loss": 4.3432,
+      "step": 1430
+    },
+    {
+      "epoch": 0.01431,
+      "grad_norm": 0.7004237859708793,
+      "learning_rate": 0.003,
+      "loss": 4.3301,
+      "step": 1431
+    },
+    {
+      "epoch": 0.01432,
+      "grad_norm": 0.8171598559675952,
+      "learning_rate": 0.003,
+      "loss": 4.3267,
+      "step": 1432
+    },
+    {
+      "epoch": 0.01433,
+      "grad_norm": 0.7196947939958487,
+      "learning_rate": 0.003,
+      "loss": 4.323,
+      "step": 1433
+    },
+    {
+      "epoch": 0.01434,
+      "grad_norm": 0.6846858190137799,
+      "learning_rate": 0.003,
+      "loss": 4.3048,
+      "step": 1434
+    },
+    {
+      "epoch": 0.01435,
+      "grad_norm": 0.8455490886566032,
+      "learning_rate": 0.003,
+      "loss": 4.3242,
+      "step": 1435
+    },
+    {
+      "epoch": 0.01436,
+      "grad_norm": 0.83940902394985,
+      "learning_rate": 0.003,
+      "loss": 4.3453,
+      "step": 1436
+    },
+    {
+      "epoch": 0.01437,
+      "grad_norm": 0.7328785180307785,
+      "learning_rate": 0.003,
+      "loss": 4.3547,
+      "step": 1437
+    },
+    {
+      "epoch": 0.01438,
+      "grad_norm": 0.7369569320140708,
+      "learning_rate": 0.003,
+      "loss": 4.349,
+      "step": 1438
+    },
+    {
+      "epoch": 0.01439,
+      "grad_norm": 0.7139149757788127,
+      "learning_rate": 0.003,
+      "loss": 4.3129,
+      "step": 1439
+    },
+    {
+      "epoch": 0.0144,
+      "grad_norm": 0.7299011882512658,
+      "learning_rate": 0.003,
+      "loss": 4.3275,
+      "step": 1440
+    },
+    {
+      "epoch": 0.01441,
+      "grad_norm": 0.670750314419178,
+      "learning_rate": 0.003,
+      "loss": 4.3309,
+      "step": 1441
+    },
+    {
+      "epoch": 0.01442,
+      "grad_norm": 0.7691195459017623,
+      "learning_rate": 0.003,
+      "loss": 4.3219,
+      "step": 1442
+    },
+    {
+      "epoch": 0.01443,
+      "grad_norm": 0.7471914299871487,
+      "learning_rate": 0.003,
+      "loss": 4.3829,
+      "step": 1443
+    },
+    {
+      "epoch": 0.01444,
+      "grad_norm": 0.7410199276775529,
+      "learning_rate": 0.003,
+      "loss": 4.3445,
+      "step": 1444
+    },
+    {
+      "epoch": 0.01445,
+      "grad_norm": 0.6515002318773166,
+      "learning_rate": 0.003,
+      "loss": 4.3041,
+      "step": 1445
+    },
+    {
+      "epoch": 0.01446,
+      "grad_norm": 0.6781090959776701,
+      "learning_rate": 0.003,
+      "loss": 4.3207,
+      "step": 1446
+    },
+    {
+      "epoch": 0.01447,
+      "grad_norm": 0.6643065072159584,
+      "learning_rate": 0.003,
+      "loss": 4.3293,
+      "step": 1447
+    },
+    {
+      "epoch": 0.01448,
+      "grad_norm": 0.6532713033801402,
+      "learning_rate": 0.003,
+      "loss": 4.3375,
+      "step": 1448
+    },
+    {
+      "epoch": 0.01449,
+      "grad_norm": 0.6344501403492658,
+      "learning_rate": 0.003,
+      "loss": 4.3302,
+      "step": 1449
+    },
+    {
+      "epoch": 0.0145,
+      "grad_norm": 0.5404283744793761,
+      "learning_rate": 0.003,
+      "loss": 4.3401,
+      "step": 1450
+    },
+    {
+      "epoch": 0.01451,
+      "grad_norm": 0.5381895515345636,
+      "learning_rate": 0.003,
+      "loss": 4.3181,
+      "step": 1451
+    },
+    {
+      "epoch": 0.01452,
+      "grad_norm": 0.5085114651371503,
+      "learning_rate": 0.003,
+      "loss": 4.3176,
+      "step": 1452
+    },
+    {
+      "epoch": 0.01453,
+      "grad_norm": 0.5297212710147206,
+      "learning_rate": 0.003,
+      "loss": 4.3302,
+      "step": 1453
+    },
+    {
+      "epoch": 0.01454,
+      "grad_norm": 0.6082304279549751,
+      "learning_rate": 0.003,
+      "loss": 4.299,
+      "step": 1454
+    },
+    {
+      "epoch": 0.01455,
+      "grad_norm": 0.799761378996593,
+      "learning_rate": 0.003,
+      "loss": 4.3379,
+      "step": 1455
+    },
+    {
+      "epoch": 0.01456,
+      "grad_norm": 0.8200199061255905,
+      "learning_rate": 0.003,
+      "loss": 4.3248,
+      "step": 1456
+    },
+    {
+      "epoch": 0.01457,
+      "grad_norm": 0.6729328173983412,
+      "learning_rate": 0.003,
+      "loss": 4.3054,
+      "step": 1457
+    },
+    {
+      "epoch": 0.01458,
+      "grad_norm": 0.5761787071333488,
+      "learning_rate": 0.003,
+      "loss": 4.2955,
+      "step": 1458
+    },
+    {
+      "epoch": 0.01459,
+      "grad_norm": 0.6468019060138231,
+      "learning_rate": 0.003,
+      "loss": 4.3289,
+      "step": 1459
+    },
+    {
+      "epoch": 0.0146,
+      "grad_norm": 0.6012651007343053,
+      "learning_rate": 0.003,
+      "loss": 4.3283,
+      "step": 1460
+    },
+    {
+      "epoch": 0.01461,
+      "grad_norm": 0.4832305210154182,
+      "learning_rate": 0.003,
+      "loss": 4.3037,
+      "step": 1461
+    },
+    {
+      "epoch": 0.01462,
+      "grad_norm": 0.4933402571331549,
+      "learning_rate": 0.003,
+      "loss": 4.3088,
+      "step": 1462
+    },
+    {
+      "epoch": 0.01463,
+      "grad_norm": 0.4774020866387011,
+      "learning_rate": 0.003,
+      "loss": 4.3072,
+      "step": 1463
+    },
+    {
+      "epoch": 0.01464,
+      "grad_norm": 0.4604171826341042,
+      "learning_rate": 0.003,
+      "loss": 4.2821,
+      "step": 1464
+    },
+    {
+      "epoch": 0.01465,
+      "grad_norm": 0.4294089797917102,
+      "learning_rate": 0.003,
+      "loss": 4.3245,
+      "step": 1465
+    },
+    {
+      "epoch": 0.01466,
+      "grad_norm": 0.41929661999999984,
+      "learning_rate": 0.003,
+      "loss": 4.2919,
+      "step": 1466
+    },
+    {
+      "epoch": 0.01467,
+      "grad_norm": 0.4391785760355703,
+      "learning_rate": 0.003,
+      "loss": 4.3283,
+      "step": 1467
+    },
+    {
+      "epoch": 0.01468,
+      "grad_norm": 0.4538846759969807,
+      "learning_rate": 0.003,
+      "loss": 4.3037,
+      "step": 1468
+    },
+    {
+      "epoch": 0.01469,
+      "grad_norm": 0.5119886030394242,
+      "learning_rate": 0.003,
+      "loss": 4.3053,
+      "step": 1469
+    },
+    {
+      "epoch": 0.0147,
+      "grad_norm": 0.716327286811815,
+      "learning_rate": 0.003,
+      "loss": 4.3041,
+      "step": 1470
+    },
+    {
+      "epoch": 0.01471,
+      "grad_norm": 0.9150052503540761,
+      "learning_rate": 0.003,
+      "loss": 4.3274,
+      "step": 1471
+    },
+    {
+      "epoch": 0.01472,
+      "grad_norm": 0.7809983181295278,
+      "learning_rate": 0.003,
+      "loss": 4.3143,
+      "step": 1472
+    },
+    {
+      "epoch": 0.01473,
+      "grad_norm": 0.8086105929105535,
+      "learning_rate": 0.003,
+      "loss": 4.3289,
+      "step": 1473
+    },
+    {
+      "epoch": 0.01474,
+      "grad_norm": 0.8656063557310713,
+      "learning_rate": 0.003,
+      "loss": 4.3258,
+      "step": 1474
+    },
+    {
+      "epoch": 0.01475,
+      "grad_norm": 0.8637306499434225,
+      "learning_rate": 0.003,
+      "loss": 4.3398,
+      "step": 1475
+    },
+    {
+      "epoch": 0.01476,
+      "grad_norm": 0.6743755524690542,
+      "learning_rate": 0.003,
+      "loss": 4.3227,
+      "step": 1476
+    },
+    {
+      "epoch": 0.01477,
+      "grad_norm": 0.6414161546100536,
+      "learning_rate": 0.003,
+      "loss": 4.3256,
+      "step": 1477
+    },
+    {
+      "epoch": 0.01478,
+      "grad_norm": 0.7067557912668699,
+      "learning_rate": 0.003,
+      "loss": 4.3113,
+      "step": 1478
+    },
+    {
+      "epoch": 0.01479,
+      "grad_norm": 0.7427717638657159,
+      "learning_rate": 0.003,
+      "loss": 4.3499,
+      "step": 1479
+    },
+    {
+      "epoch": 0.0148,
+      "grad_norm": 0.7972486147312002,
+      "learning_rate": 0.003,
+      "loss": 4.3212,
+      "step": 1480
+    },
+    {
+      "epoch": 0.01481,
+      "grad_norm": 0.8238319142257903,
+      "learning_rate": 0.003,
+      "loss": 4.3261,
+      "step": 1481
+    },
+    {
+      "epoch": 0.01482,
+      "grad_norm": 0.7010334962030825,
+      "learning_rate": 0.003,
+      "loss": 4.3089,
+      "step": 1482
+    },
+    {
+      "epoch": 0.01483,
+      "grad_norm": 0.6287754589698554,
+      "learning_rate": 0.003,
+      "loss": 4.303,
+      "step": 1483
+    },
+    {
+      "epoch": 0.01484,
+      "grad_norm": 0.5953574922658559,
+      "learning_rate": 0.003,
+      "loss": 4.3534,
+      "step": 1484
+    },
+    {
+      "epoch": 0.01485,
+      "grad_norm": 0.6859671225358986,
+      "learning_rate": 0.003,
+      "loss": 4.3172,
+      "step": 1485
+    },
+    {
+      "epoch": 0.01486,
+      "grad_norm": 0.7393843625917039,
+      "learning_rate": 0.003,
+      "loss": 4.3181,
+      "step": 1486
+    },
+    {
+      "epoch": 0.01487,
+      "grad_norm": 0.6948333501030851,
+      "learning_rate": 0.003,
+      "loss": 4.3225,
+      "step": 1487
+    },
+    {
+      "epoch": 0.01488,
+      "grad_norm": 0.5806619385051136,
+      "learning_rate": 0.003,
+      "loss": 4.3259,
+      "step": 1488
+    },
+    {
+      "epoch": 0.01489,
+      "grad_norm": 0.6342989775695094,
+      "learning_rate": 0.003,
+      "loss": 4.3221,
+      "step": 1489
+    },
+    {
+      "epoch": 0.0149,
+      "grad_norm": 0.6704129451979153,
+      "learning_rate": 0.003,
+      "loss": 4.3147,
+      "step": 1490
+    },
+    {
+      "epoch": 0.01491,
+      "grad_norm": 0.7017206728292057,
+      "learning_rate": 0.003,
+      "loss": 4.3305,
+      "step": 1491
+    },
+    {
+      "epoch": 0.01492,
+      "grad_norm": 0.6813188063148864,
+      "learning_rate": 0.003,
+      "loss": 4.2961,
+      "step": 1492
+    },
+    {
+      "epoch": 0.01493,
+      "grad_norm": 0.8525559699799811,
+      "learning_rate": 0.003,
+      "loss": 4.3324,
+      "step": 1493
+    },
+    {
+      "epoch": 0.01494,
+      "grad_norm": 0.8687834663259932,
+      "learning_rate": 0.003,
+      "loss": 4.3213,
+      "step": 1494
+    },
+    {
+      "epoch": 0.01495,
+      "grad_norm": 0.7274136348295218,
+      "learning_rate": 0.003,
+      "loss": 4.3243,
+      "step": 1495
+    },
+    {
+      "epoch": 0.01496,
+      "grad_norm": 0.5434712910501038,
+      "learning_rate": 0.003,
+      "loss": 4.2923,
+      "step": 1496
+    },
+    {
+      "epoch": 0.01497,
+      "grad_norm": 0.6148760530691155,
+      "learning_rate": 0.003,
+      "loss": 4.2701,
+      "step": 1497
+    },
+    {
+      "epoch": 0.01498,
+      "grad_norm": 0.5665585769325068,
+      "learning_rate": 0.003,
+      "loss": 4.3224,
+      "step": 1498
+    },
+    {
+      "epoch": 0.01499,
+      "grad_norm": 0.543044486095261,
+      "learning_rate": 0.003,
+      "loss": 4.3408,
+      "step": 1499
+    },
+    {
+      "epoch": 0.015,
+      "grad_norm": 0.5868245294183819,
+      "learning_rate": 0.003,
+      "loss": 4.2952,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01501,
+      "grad_norm": 0.6488345831013783,
+      "learning_rate": 0.003,
+      "loss": 4.2943,
+      "step": 1501
+    },
+    {
+      "epoch": 0.01502,
+      "grad_norm": 0.5560720951051904,
+      "learning_rate": 0.003,
+      "loss": 4.3322,
+      "step": 1502
+    },
+    {
+      "epoch": 0.01503,
+      "grad_norm": 0.6170651368212409,
+      "learning_rate": 0.003,
+      "loss": 4.3142,
+      "step": 1503
+    },
+    {
+      "epoch": 0.01504,
+      "grad_norm": 0.7253173696408437,
+      "learning_rate": 0.003,
+      "loss": 4.3199,
+      "step": 1504
+    },
+    {
+      "epoch": 0.01505,
+      "grad_norm": 0.7573152828389973,
+      "learning_rate": 0.003,
+      "loss": 4.3067,
+      "step": 1505
+    },
+    {
+      "epoch": 0.01506,
+      "grad_norm": 0.7119425101662294,
+      "learning_rate": 0.003,
+      "loss": 4.3233,
+      "step": 1506
+    },
+    {
+      "epoch": 0.01507,
+      "grad_norm": 0.7408719494100054,
+      "learning_rate": 0.003,
+      "loss": 4.3394,
+      "step": 1507
+    },
+    {
+      "epoch": 0.01508,
+      "grad_norm": 0.6023053184100382,
+      "learning_rate": 0.003,
+      "loss": 4.3108,
+      "step": 1508
+    },
+    {
+      "epoch": 0.01509,
+      "grad_norm": 0.5054062616818538,
+      "learning_rate": 0.003,
+      "loss": 4.3257,
+      "step": 1509
+    },
+    {
+      "epoch": 0.0151,
+      "grad_norm": 0.4440427822667444,
+      "learning_rate": 0.003,
+      "loss": 4.2952,
+      "step": 1510
+    },
+    {
+      "epoch": 0.01511,
+      "grad_norm": 0.503043960784533,
+      "learning_rate": 0.003,
+      "loss": 4.2838,
+      "step": 1511
+    },
+    {
+      "epoch": 0.01512,
+      "grad_norm": 0.4478910644887885,
+      "learning_rate": 0.003,
+      "loss": 4.2937,
+      "step": 1512
+    },
+    {
+      "epoch": 0.01513,
+      "grad_norm": 0.46313433916934627,
+      "learning_rate": 0.003,
+      "loss": 4.2989,
+      "step": 1513
+    },
+    {
+      "epoch": 0.01514,
+      "grad_norm": 0.5237062692888148,
+      "learning_rate": 0.003,
+      "loss": 4.3066,
+      "step": 1514
+    },
+    {
+      "epoch": 0.01515,
+      "grad_norm": 0.6686824710181497,
+      "learning_rate": 0.003,
+      "loss": 4.3056,
+      "step": 1515
+    },
+    {
+      "epoch": 0.01516,
+      "grad_norm": 0.8162150436928566,
+      "learning_rate": 0.003,
+      "loss": 4.3198,
+      "step": 1516
+    },
+    {
+      "epoch": 0.01517,
+      "grad_norm": 0.8250386268904906,
+      "learning_rate": 0.003,
+      "loss": 4.274,
+      "step": 1517
+    },
+    {
+      "epoch": 0.01518,
+      "grad_norm": 0.8551846044396628,
+      "learning_rate": 0.003,
+      "loss": 4.3102,
+      "step": 1518
+    },
+    {
+      "epoch": 0.01519,
+      "grad_norm": 0.7778994522735176,
+      "learning_rate": 0.003,
+      "loss": 4.3471,
+      "step": 1519
+    },
+    {
+      "epoch": 0.0152,
+      "grad_norm": 0.6407333772962789,
+      "learning_rate": 0.003,
+      "loss": 4.2901,
+      "step": 1520
+    },
+    {
+      "epoch": 0.01521,
+      "grad_norm": 0.590099483112598,
+      "learning_rate": 0.003,
+      "loss": 4.3053,
+      "step": 1521
+    },
+    {
+      "epoch": 0.01522,
+      "grad_norm": 0.6220605799864671,
+      "learning_rate": 0.003,
+      "loss": 4.3014,
+      "step": 1522
+    },
+    {
+      "epoch": 0.01523,
+      "grad_norm": 0.6378640000910417,
+      "learning_rate": 0.003,
+      "loss": 4.2957,
+      "step": 1523
+    },
+    {
+      "epoch": 0.01524,
+      "grad_norm": 0.6800153761279158,
+      "learning_rate": 0.003,
+      "loss": 4.3332,
+      "step": 1524
+    },
+    {
+      "epoch": 0.01525,
+      "grad_norm": 0.6552309843606878,
+      "learning_rate": 0.003,
+      "loss": 4.3306,
+      "step": 1525
+    },
+    {
+      "epoch": 0.01526,
+      "grad_norm": 0.5832236683536917,
+      "learning_rate": 0.003,
+      "loss": 4.3154,
+      "step": 1526
+    },
+    {
+      "epoch": 0.01527,
+      "grad_norm": 0.5923964968536319,
+      "learning_rate": 0.003,
+      "loss": 4.2868,
+      "step": 1527
+    },
+    {
+      "epoch": 0.01528,
+      "grad_norm": 0.6438689697714796,
+      "learning_rate": 0.003,
+      "loss": 4.3067,
+      "step": 1528
+    },
+    {
+      "epoch": 0.01529,
+      "grad_norm": 0.7863523458278316,
+      "learning_rate": 0.003,
+      "loss": 4.3071,
+      "step": 1529
+    },
+    {
+      "epoch": 0.0153,
+      "grad_norm": 0.8716684688082496,
+      "learning_rate": 0.003,
+      "loss": 4.3236,
+      "step": 1530
+    },
+    {
+      "epoch": 0.01531,
+      "grad_norm": 0.7632243352103186,
+      "learning_rate": 0.003,
+      "loss": 4.314,
+      "step": 1531
+    },
+    {
+      "epoch": 0.01532,
+      "grad_norm": 0.5830764209700208,
+      "learning_rate": 0.003,
+      "loss": 4.2863,
+      "step": 1532
+    },
+    {
+      "epoch": 0.01533,
+      "grad_norm": 0.6925578088506626,
+      "learning_rate": 0.003,
+      "loss": 4.3193,
+      "step": 1533
+    },
+    {
+      "epoch": 0.01534,
+      "grad_norm": 0.6884106279584329,
+      "learning_rate": 0.003,
+      "loss": 4.3131,
+      "step": 1534
+    },
+    {
+      "epoch": 0.01535,
+      "grad_norm": 0.612086877528724,
+      "learning_rate": 0.003,
+      "loss": 4.3216,
+      "step": 1535
+    },
+    {
+      "epoch": 0.01536,
+      "grad_norm": 0.6277711721176384,
+      "learning_rate": 0.003,
+      "loss": 4.3039,
+      "step": 1536
+    },
+    {
+      "epoch": 0.01537,
+      "grad_norm": 0.7930101311629847,
+      "learning_rate": 0.003,
+      "loss": 4.3078,
+      "step": 1537
+    },
+    {
+      "epoch": 0.01538,
+      "grad_norm": 0.8987557100700295,
+      "learning_rate": 0.003,
+      "loss": 4.2971,
+      "step": 1538
+    },
+    {
+      "epoch": 0.01539,
+      "grad_norm": 0.9391352557198052,
+      "learning_rate": 0.003,
+      "loss": 4.334,
+      "step": 1539
+    },
+    {
+      "epoch": 0.0154,
+      "grad_norm": 0.9440902063787158,
+      "learning_rate": 0.003,
+      "loss": 4.3487,
+      "step": 1540
+    },
+    {
+      "epoch": 0.01541,
+      "grad_norm": 0.9127476278073733,
+      "learning_rate": 0.003,
+      "loss": 4.328,
+      "step": 1541
+    },
+    {
+      "epoch": 0.01542,
+      "grad_norm": 0.7753546344208377,
+      "learning_rate": 0.003,
+      "loss": 4.3528,
+      "step": 1542
+    },
+    {
+      "epoch": 0.01543,
+      "grad_norm": 0.7317390533493494,
+      "learning_rate": 0.003,
+      "loss": 4.3318,
+      "step": 1543
+    },
+    {
+      "epoch": 0.01544,
+      "grad_norm": 0.5946634053593371,
+      "learning_rate": 0.003,
+      "loss": 4.3162,
+      "step": 1544
+    },
+    {
+      "epoch": 0.01545,
+      "grad_norm": 0.49808121797063876,
+      "learning_rate": 0.003,
+      "loss": 4.375,
+      "step": 1545
+    },
+    {
+      "epoch": 0.01546,
+      "grad_norm": 0.5189277397467036,
+      "learning_rate": 0.003,
+      "loss": 4.3435,
+      "step": 1546
+    },
+    {
+      "epoch": 0.01547,
+      "grad_norm": 0.5075409907094708,
+      "learning_rate": 0.003,
+      "loss": 4.3431,
+      "step": 1547
+    },
+    {
+      "epoch": 0.01548,
+      "grad_norm": 0.4269782583856954,
+      "learning_rate": 0.003,
+      "loss": 4.3016,
+      "step": 1548
+    },
+    {
+      "epoch": 0.01549,
+      "grad_norm": 0.44259739876123444,
+      "learning_rate": 0.003,
+      "loss": 4.2898,
+      "step": 1549
+    },
+    {
+      "epoch": 0.0155,
+      "grad_norm": 0.39476104723697186,
+      "learning_rate": 0.003,
+      "loss": 4.3072,
+      "step": 1550
+    },
+    {
+      "epoch": 0.01551,
+      "grad_norm": 0.39719568498427116,
+      "learning_rate": 0.003,
+      "loss": 4.2836,
+      "step": 1551
+    },
+    {
+      "epoch": 0.01552,
+      "grad_norm": 0.5112856286097273,
+      "learning_rate": 0.003,
+      "loss": 4.3059,
+      "step": 1552
+    },
+    {
+      "epoch": 0.01553,
+      "grad_norm": 0.6994822828905559,
+      "learning_rate": 0.003,
+      "loss": 4.3366,
+      "step": 1553
+    },
+    {
+      "epoch": 0.01554,
+      "grad_norm": 0.9918562900778292,
+      "learning_rate": 0.003,
+      "loss": 4.3337,
+      "step": 1554
+    },
+    {
+      "epoch": 0.01555,
+      "grad_norm": 0.8597749628867734,
+      "learning_rate": 0.003,
+      "loss": 4.3154,
+      "step": 1555
+    },
+    {
+      "epoch": 0.01556,
+      "grad_norm": 0.4859845337872918,
+      "learning_rate": 0.003,
+      "loss": 4.3299,
+      "step": 1556
+    },
+    {
+      "epoch": 0.01557,
+      "grad_norm": 0.8105266239108289,
+      "learning_rate": 0.003,
+      "loss": 4.317,
+      "step": 1557
+    },
+    {
+      "epoch": 0.01558,
+      "grad_norm": 0.6797261746777549,
+      "learning_rate": 0.003,
+      "loss": 4.3032,
+      "step": 1558
+    },
+    {
+      "epoch": 0.01559,
+      "grad_norm": 0.5540991472031954,
+      "learning_rate": 0.003,
+      "loss": 4.3349,
+      "step": 1559
+    },
+    {
+      "epoch": 0.0156,
+      "grad_norm": 0.5222478848499021,
+      "learning_rate": 0.003,
+      "loss": 4.3207,
+      "step": 1560
+    },
+    {
+      "epoch": 0.01561,
+      "grad_norm": 0.4547690370956997,
+      "learning_rate": 0.003,
+      "loss": 4.3006,
+      "step": 1561
+    },
+    {
+      "epoch": 0.01562,
+      "grad_norm": 0.3960981969234329,
+      "learning_rate": 0.003,
+      "loss": 4.2916,
+      "step": 1562
+    },
+    {
+      "epoch": 0.01563,
+      "grad_norm": 0.4365008440994641,
+      "learning_rate": 0.003,
+      "loss": 4.2998,
+      "step": 1563
+    },
+    {
+      "epoch": 0.01564,
+      "grad_norm": 0.4153235041333595,
+      "learning_rate": 0.003,
+      "loss": 4.3237,
+      "step": 1564
+    },
+    {
+      "epoch": 0.01565,
+      "grad_norm": 0.4172430370227481,
+      "learning_rate": 0.003,
+      "loss": 4.2984,
+      "step": 1565
+    },
+    {
+      "epoch": 0.01566,
+      "grad_norm": 0.4030631404816366,
+      "learning_rate": 0.003,
+      "loss": 4.3153,
+      "step": 1566
+    },
+    {
+      "epoch": 0.01567,
+      "grad_norm": 0.4006086575381243,
+      "learning_rate": 0.003,
+      "loss": 4.3082,
+      "step": 1567
+    },
+    {
+      "epoch": 0.01568,
+      "grad_norm": 0.4692595605101829,
+      "learning_rate": 0.003,
+      "loss": 4.2921,
+      "step": 1568
+    },
+    {
+      "epoch": 0.01569,
+      "grad_norm": 0.5156092105425303,
+      "learning_rate": 0.003,
+      "loss": 4.2975,
+      "step": 1569
+    },
+    {
+      "epoch": 0.0157,
+      "grad_norm": 0.5574730520335043,
+      "learning_rate": 0.003,
+      "loss": 4.3166,
+      "step": 1570
+    },
+    {
+      "epoch": 0.01571,
+      "grad_norm": 0.7147942584489586,
+      "learning_rate": 0.003,
+      "loss": 4.306,
+      "step": 1571
+    },
+    {
+      "epoch": 0.01572,
+      "grad_norm": 0.8176596681794123,
+      "learning_rate": 0.003,
+      "loss": 4.3202,
+      "step": 1572
+    },
+    {
+      "epoch": 0.01573,
+      "grad_norm": 0.8859299700656846,
+      "learning_rate": 0.003,
+      "loss": 4.311,
+      "step": 1573
+    },
+    {
+      "epoch": 0.01574,
+      "grad_norm": 1.0167563301248608,
+      "learning_rate": 0.003,
+      "loss": 4.3365,
+      "step": 1574
+    },
+    {
+      "epoch": 0.01575,
+      "grad_norm": 0.9276915846212054,
+      "learning_rate": 0.003,
+      "loss": 4.3012,
+      "step": 1575
+    },
+    {
+      "epoch": 0.01576,
+      "grad_norm": 0.7519027259067711,
+      "learning_rate": 0.003,
+      "loss": 4.317,
+      "step": 1576
+    },
+    {
+      "epoch": 0.01577,
+      "grad_norm": 0.5990656751852772,
+      "learning_rate": 0.003,
+      "loss": 4.2743,
+      "step": 1577
+    },
+    {
+      "epoch": 0.01578,
+      "grad_norm": 0.652141379980981,
+      "learning_rate": 0.003,
+      "loss": 4.3047,
+      "step": 1578
+    },
+    {
+      "epoch": 0.01579,
+      "grad_norm": 0.6130941880481341,
+      "learning_rate": 0.003,
+      "loss": 4.3001,
+      "step": 1579
+    },
+    {
+      "epoch": 0.0158,
+      "grad_norm": 0.6566738181349618,
+      "learning_rate": 0.003,
+      "loss": 4.3101,
+      "step": 1580
+    },
+    {
+      "epoch": 0.01581,
+      "grad_norm": 0.5602300919728743,
+      "learning_rate": 0.003,
+      "loss": 4.3154,
+      "step": 1581
+    },
+    {
+      "epoch": 0.01582,
+      "grad_norm": 0.6614515024819504,
+      "learning_rate": 0.003,
+      "loss": 4.3018,
+      "step": 1582
+    },
+    {
+      "epoch": 0.01583,
+      "grad_norm": 0.6685116238668469,
+      "learning_rate": 0.003,
+      "loss": 4.2884,
+      "step": 1583
+    },
+    {
+      "epoch": 0.01584,
+      "grad_norm": 0.7663983884039552,
+      "learning_rate": 0.003,
+      "loss": 4.3227,
+      "step": 1584
+    },
+    {
+      "epoch": 0.01585,
+      "grad_norm": 0.8471607724061854,
+      "learning_rate": 0.003,
+      "loss": 4.3183,
+      "step": 1585
+    },
+    {
+      "epoch": 0.01586,
+      "grad_norm": 0.8029205674223586,
+      "learning_rate": 0.003,
+      "loss": 4.2999,
+      "step": 1586
+    },
+    {
+      "epoch": 0.01587,
+      "grad_norm": 0.7335558361580696,
+      "learning_rate": 0.003,
+      "loss": 4.3092,
+      "step": 1587
+    },
+    {
+      "epoch": 0.01588,
+      "grad_norm": 0.6576426322376321,
+      "learning_rate": 0.003,
+      "loss": 4.3341,
+      "step": 1588
+    },
+    {
+      "epoch": 0.01589,
+      "grad_norm": 0.7314604361721292,
+      "learning_rate": 0.003,
+      "loss": 4.328,
+      "step": 1589
+    },
+    {
+      "epoch": 0.0159,
+      "grad_norm": 0.7484813941023133,
+      "learning_rate": 0.003,
+      "loss": 4.323,
+      "step": 1590
+    },
+    {
+      "epoch": 0.01591,
+      "grad_norm": 0.6836450786052791,
+      "learning_rate": 0.003,
+      "loss": 4.3264,
+      "step": 1591
+    },
+    {
+      "epoch": 0.01592,
+      "grad_norm": 0.6522043667699231,
+      "learning_rate": 0.003,
+      "loss": 4.3127,
+      "step": 1592
+    },
+    {
+      "epoch": 0.01593,
+      "grad_norm": 0.6311012555499182,
+      "learning_rate": 0.003,
+      "loss": 4.3174,
+      "step": 1593
+    },
+    {
+      "epoch": 0.01594,
+      "grad_norm": 0.5510623120073839,
+      "learning_rate": 0.003,
+      "loss": 4.3009,
+      "step": 1594
+    },
+    {
+      "epoch": 0.01595,
+      "grad_norm": 0.48793948155833994,
+      "learning_rate": 0.003,
+      "loss": 4.3248,
+      "step": 1595
+    },
+    {
+      "epoch": 0.01596,
+      "grad_norm": 0.42717068818596066,
+      "learning_rate": 0.003,
+      "loss": 4.3142,
+      "step": 1596
+    },
+    {
+      "epoch": 0.01597,
+      "grad_norm": 0.43677376148637953,
+      "learning_rate": 0.003,
+      "loss": 4.2773,
+      "step": 1597
+    },
+    {
+      "epoch": 0.01598,
+      "grad_norm": 0.47981748858846524,
+      "learning_rate": 0.003,
+      "loss": 4.2706,
+      "step": 1598
+    },
+    {
+      "epoch": 0.01599,
+      "grad_norm": 0.5068183957178387,
+      "learning_rate": 0.003,
+      "loss": 4.2944,
+      "step": 1599
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 0.570411039839509,
+      "learning_rate": 0.003,
+      "loss": 4.261,
+      "step": 1600
+    },
+    {
+      "epoch": 0.01601,
+      "grad_norm": 0.7428825863420937,
+      "learning_rate": 0.003,
+      "loss": 4.3164,
+      "step": 1601
+    },
+    {
+      "epoch": 0.01602,
+      "grad_norm": 1.0316614845943484,
+      "learning_rate": 0.003,
+      "loss": 4.3056,
+      "step": 1602
+    },
+    {
+      "epoch": 0.01603,
+      "grad_norm": 0.9265092099146023,
+      "learning_rate": 0.003,
+      "loss": 4.3273,
+      "step": 1603
+    },
+    {
+      "epoch": 0.01604,
+      "grad_norm": 0.6714138671323193,
+      "learning_rate": 0.003,
+      "loss": 4.3099,
+      "step": 1604
+    },
+    {
+      "epoch": 0.01605,
+      "grad_norm": 0.7449186454711473,
+      "learning_rate": 0.003,
+      "loss": 4.2937,
+      "step": 1605
+    },
+    {
+      "epoch": 0.01606,
+      "grad_norm": 0.6551508425933098,
+      "learning_rate": 0.003,
+      "loss": 4.29,
+      "step": 1606
+    },
+    {
+      "epoch": 0.01607,
+      "grad_norm": 0.6368386814210635,
+      "learning_rate": 0.003,
+      "loss": 4.3196,
+      "step": 1607
+    },
+    {
+      "epoch": 0.01608,
+      "grad_norm": 0.6007545339051799,
+      "learning_rate": 0.003,
+      "loss": 4.287,
+      "step": 1608
+    },
+    {
+      "epoch": 0.01609,
+      "grad_norm": 0.5685481620694516,
+      "learning_rate": 0.003,
+      "loss": 4.2973,
+      "step": 1609
+    },
+    {
+      "epoch": 0.0161,
+      "grad_norm": 0.5084184586189073,
+      "learning_rate": 0.003,
+      "loss": 4.2592,
+      "step": 1610
+    },
+    {
+      "epoch": 0.01611,
+      "grad_norm": 0.4840833495127669,
+      "learning_rate": 0.003,
+      "loss": 4.2802,
+      "step": 1611
+    },
+    {
+      "epoch": 0.01612,
+      "grad_norm": 0.48942028074279065,
+      "learning_rate": 0.003,
+      "loss": 4.2702,
+      "step": 1612
+    },
+    {
+      "epoch": 0.01613,
+      "grad_norm": 0.5011285237133881,
+      "learning_rate": 0.003,
+      "loss": 4.315,
+      "step": 1613
+    },
+    {
+      "epoch": 0.01614,
+      "grad_norm": 0.43995362094592044,
+      "learning_rate": 0.003,
+      "loss": 4.3364,
+      "step": 1614
+    },
+    {
+      "epoch": 0.01615,
+      "grad_norm": 0.42434808753845177,
+      "learning_rate": 0.003,
+      "loss": 4.3023,
+      "step": 1615
+    },
+    {
+      "epoch": 0.01616,
+      "grad_norm": 0.40673555464346545,
+      "learning_rate": 0.003,
+      "loss": 4.2826,
+      "step": 1616
+    },
+    {
+      "epoch": 0.01617,
+      "grad_norm": 0.3578337742407294,
+      "learning_rate": 0.003,
+      "loss": 4.2706,
+      "step": 1617
+    },
+    {
+      "epoch": 0.01618,
+      "grad_norm": 0.41081917922167766,
+      "learning_rate": 0.003,
+      "loss": 4.2753,
+      "step": 1618
+    },
+    {
+      "epoch": 0.01619,
+      "grad_norm": 0.47886760935645833,
+      "learning_rate": 0.003,
+      "loss": 4.2659,
+      "step": 1619
+    },
+    {
+      "epoch": 0.0162,
+      "grad_norm": 0.6222321239813,
+      "learning_rate": 0.003,
+      "loss": 4.2767,
+      "step": 1620
+    },
+    {
+      "epoch": 0.01621,
+      "grad_norm": 0.6903491585345826,
+      "learning_rate": 0.003,
+      "loss": 4.2786,
+      "step": 1621
+    },
+    {
+      "epoch": 0.01622,
+      "grad_norm": 0.7573622136296652,
+      "learning_rate": 0.003,
+      "loss": 4.3118,
+      "step": 1622
+    },
+    {
+      "epoch": 0.01623,
+      "grad_norm": 0.9256995453444036,
+      "learning_rate": 0.003,
+      "loss": 4.3026,
+      "step": 1623
+    },
+    {
+      "epoch": 0.01624,
+      "grad_norm": 0.9558305805696413,
+      "learning_rate": 0.003,
+      "loss": 4.3216,
+      "step": 1624
+    },
+    {
+      "epoch": 0.01625,
+      "grad_norm": 0.8114385487348762,
+      "learning_rate": 0.003,
+      "loss": 4.2898,
+      "step": 1625
+    },
+    {
+      "epoch": 0.01626,
+      "grad_norm": 0.8322826717160557,
+      "learning_rate": 0.003,
+      "loss": 4.3177,
+      "step": 1626
+    },
+    {
+      "epoch": 0.01627,
+      "grad_norm": 0.7727713367753223,
+      "learning_rate": 0.003,
+      "loss": 4.2816,
+      "step": 1627
+    },
+    {
+      "epoch": 0.01628,
+      "grad_norm": 0.7109224896248458,
+      "learning_rate": 0.003,
+      "loss": 4.2929,
+      "step": 1628
+    },
+    {
+      "epoch": 0.01629,
+      "grad_norm": 0.560897124973334,
+      "learning_rate": 0.003,
+      "loss": 4.2858,
+      "step": 1629
+    },
+    {
+      "epoch": 0.0163,
+      "grad_norm": 0.6779380557060044,
+      "learning_rate": 0.003,
+      "loss": 4.2826,
+      "step": 1630
+    },
+    {
+      "epoch": 0.01631,
+      "grad_norm": 0.6886638086492523,
+      "learning_rate": 0.003,
+      "loss": 4.3011,
+      "step": 1631
+    },
+    {
+      "epoch": 0.01632,
+      "grad_norm": 0.6589349584439033,
+      "learning_rate": 0.003,
+      "loss": 4.3124,
+      "step": 1632
+    },
+    {
+      "epoch": 0.01633,
+      "grad_norm": 0.7155276302347654,
+      "learning_rate": 0.003,
+      "loss": 4.2893,
+      "step": 1633
+    },
+    {
+      "epoch": 0.01634,
+      "grad_norm": 0.7783909364325312,
+      "learning_rate": 0.003,
+      "loss": 4.3094,
+      "step": 1634
+    },
+    {
+      "epoch": 0.01635,
+      "grad_norm": 0.8288192042951168,
+      "learning_rate": 0.003,
+      "loss": 4.2952,
+      "step": 1635
+    },
+    {
+      "epoch": 0.01636,
+      "grad_norm": 0.8309682765627266,
+      "learning_rate": 0.003,
+      "loss": 4.3157,
+      "step": 1636
+    },
+    {
+      "epoch": 0.01637,
+      "grad_norm": 0.7885672766922769,
+      "learning_rate": 0.003,
+      "loss": 4.2985,
+      "step": 1637
+    },
+    {
+      "epoch": 0.01638,
+      "grad_norm": 0.6820787883112478,
+      "learning_rate": 0.003,
+      "loss": 4.3026,
+      "step": 1638
+    },
+    {
+      "epoch": 0.01639,
+      "grad_norm": 0.5912531628997083,
+      "learning_rate": 0.003,
+      "loss": 4.3221,
+      "step": 1639
+    },
+    {
+      "epoch": 0.0164,
+      "grad_norm": 0.7081315099543188,
+      "learning_rate": 0.003,
+      "loss": 4.3062,
+      "step": 1640
+    },
+    {
+      "epoch": 0.01641,
+      "grad_norm": 0.7390837101488571,
+      "learning_rate": 0.003,
+      "loss": 4.279,
+      "step": 1641
+    },
+    {
+      "epoch": 0.01642,
+      "grad_norm": 0.8019499352702314,
+      "learning_rate": 0.003,
+      "loss": 4.3244,
+      "step": 1642
+    },
+    {
+      "epoch": 0.01643,
+      "grad_norm": 0.7922382978586373,
+      "learning_rate": 0.003,
+      "loss": 4.3126,
+      "step": 1643
+    },
+    {
+      "epoch": 0.01644,
+      "grad_norm": 0.7260441686104165,
+      "learning_rate": 0.003,
+      "loss": 4.3282,
+      "step": 1644
+    },
+    {
+      "epoch": 0.01645,
+      "grad_norm": 0.5765166219346322,
+      "learning_rate": 0.003,
+      "loss": 4.2686,
+      "step": 1645
+    },
+    {
+      "epoch": 0.01646,
+      "grad_norm": 0.4846869130781467,
+      "learning_rate": 0.003,
+      "loss": 4.298,
+      "step": 1646
+    },
+    {
+      "epoch": 0.01647,
+      "grad_norm": 0.4324516045425632,
+      "learning_rate": 0.003,
+      "loss": 4.294,
+      "step": 1647
+    },
+    {
+      "epoch": 0.01648,
+      "grad_norm": 0.45801960619516824,
+      "learning_rate": 0.003,
+      "loss": 4.2725,
+      "step": 1648
+    },
+    {
+      "epoch": 0.01649,
+      "grad_norm": 0.4519054523297515,
+      "learning_rate": 0.003,
+      "loss": 4.3034,
+      "step": 1649
+    },
+    {
+      "epoch": 0.0165,
+      "grad_norm": 0.5030069407293299,
+      "learning_rate": 0.003,
+      "loss": 4.2846,
+      "step": 1650
+    },
+    {
+      "epoch": 0.01651,
+      "grad_norm": 0.7338143580585036,
+      "learning_rate": 0.003,
+      "loss": 4.2988,
+      "step": 1651
+    },
+    {
+      "epoch": 0.01652,
+      "grad_norm": 0.9179472565177694,
+      "learning_rate": 0.003,
+      "loss": 4.2688,
+      "step": 1652
+    },
+    {
+      "epoch": 0.01653,
+      "grad_norm": 0.7974402781604697,
+      "learning_rate": 0.003,
+      "loss": 4.307,
+      "step": 1653
+    },
+    {
+      "epoch": 0.01654,
+      "grad_norm": 0.8094858670947136,
+      "learning_rate": 0.003,
+      "loss": 4.2972,
+      "step": 1654
+    },
+    {
+      "epoch": 0.01655,
+      "grad_norm": 0.8204525750124729,
+      "learning_rate": 0.003,
+      "loss": 4.313,
+      "step": 1655
+    },
+    {
+      "epoch": 0.01656,
+      "grad_norm": 0.6293302149986666,
+      "learning_rate": 0.003,
+      "loss": 4.271,
+      "step": 1656
+    },
+    {
+      "epoch": 0.01657,
+      "grad_norm": 0.5706278835378605,
+      "learning_rate": 0.003,
+      "loss": 4.293,
+      "step": 1657
+    },
+    {
+      "epoch": 0.01658,
+      "grad_norm": 0.6110254736121277,
+      "learning_rate": 0.003,
+      "loss": 4.2809,
+      "step": 1658
+    },
+    {
+      "epoch": 0.01659,
+      "grad_norm": 0.4984603935459828,
+      "learning_rate": 0.003,
+      "loss": 4.2824,
+      "step": 1659
+    },
+    {
+      "epoch": 0.0166,
+      "grad_norm": 0.46804303085612786,
+      "learning_rate": 0.003,
+      "loss": 4.2649,
+      "step": 1660
+    },
+    {
+      "epoch": 0.01661,
+      "grad_norm": 0.48879681023532395,
+      "learning_rate": 0.003,
+      "loss": 4.2671,
+      "step": 1661
+    },
+    {
+      "epoch": 0.01662,
+      "grad_norm": 0.5571765680768687,
+      "learning_rate": 0.003,
+      "loss": 4.2719,
+      "step": 1662
+    },
+    {
+      "epoch": 0.01663,
+      "grad_norm": 0.532882366154707,
+      "learning_rate": 0.003,
+      "loss": 4.294,
+      "step": 1663
+    },
+    {
+      "epoch": 0.01664,
+      "grad_norm": 0.5088338427561463,
+      "learning_rate": 0.003,
+      "loss": 4.3094,
+      "step": 1664
+    },
+    {
+      "epoch": 0.01665,
+      "grad_norm": 0.4963301596331138,
+      "learning_rate": 0.003,
+      "loss": 4.2712,
+      "step": 1665
+    },
+    {
+      "epoch": 0.01666,
+      "grad_norm": 0.5472162068898109,
+      "learning_rate": 0.003,
+      "loss": 4.2732,
+      "step": 1666
+    },
+    {
+      "epoch": 0.01667,
+      "grad_norm": 0.6863444215859392,
+      "learning_rate": 0.003,
+      "loss": 4.2951,
+      "step": 1667
+    },
+    {
+      "epoch": 0.01668,
+      "grad_norm": 0.7945747889757943,
+      "learning_rate": 0.003,
+      "loss": 4.3019,
+      "step": 1668
+    },
+    {
+      "epoch": 0.01669,
+      "grad_norm": 0.7029756676891371,
+      "learning_rate": 0.003,
+      "loss": 4.2868,
+      "step": 1669
+    },
+    {
+      "epoch": 0.0167,
+      "grad_norm": 0.564779057565769,
+      "learning_rate": 0.003,
+      "loss": 4.2701,
+      "step": 1670
+    },
+    {
+      "epoch": 0.01671,
+      "grad_norm": 0.5946383928046679,
+      "learning_rate": 0.003,
+      "loss": 4.2602,
+      "step": 1671
+    },
+    {
+      "epoch": 0.01672,
+      "grad_norm": 0.7328231067257597,
+      "learning_rate": 0.003,
+      "loss": 4.2779,
+      "step": 1672
+    },
+    {
+      "epoch": 0.01673,
+      "grad_norm": 0.7730808965686631,
+      "learning_rate": 0.003,
+      "loss": 4.2759,
+      "step": 1673
+    },
+    {
+      "epoch": 0.01674,
+      "grad_norm": 0.8250670120229329,
+      "learning_rate": 0.003,
+      "loss": 4.3013,
+      "step": 1674
+    },
+    {
+      "epoch": 0.01675,
+      "grad_norm": 0.7145691985141881,
+      "learning_rate": 0.003,
+      "loss": 4.2806,
+      "step": 1675
+    },
+    {
+      "epoch": 0.01676,
+      "grad_norm": 0.6268084931910631,
+      "learning_rate": 0.003,
+      "loss": 4.2872,
+      "step": 1676
+    },
+    {
+      "epoch": 0.01677,
+      "grad_norm": 0.6170941899124056,
+      "learning_rate": 0.003,
+      "loss": 4.2772,
+      "step": 1677
+    },
+    {
+      "epoch": 0.01678,
+      "grad_norm": 0.5951034196601164,
+      "learning_rate": 0.003,
+      "loss": 4.2879,
+      "step": 1678
+    },
+    {
+      "epoch": 0.01679,
+      "grad_norm": 0.5254502211849061,
+      "learning_rate": 0.003,
+      "loss": 4.2679,
+      "step": 1679
+    },
+    {
+      "epoch": 0.0168,
+      "grad_norm": 0.6096842133031645,
+      "learning_rate": 0.003,
+      "loss": 4.2892,
+      "step": 1680
+    },
+    {
+      "epoch": 0.01681,
+      "grad_norm": 0.6218177000062469,
+      "learning_rate": 0.003,
+      "loss": 4.2892,
+      "step": 1681
+    },
+    {
+      "epoch": 0.01682,
+      "grad_norm": 0.5832829705689185,
+      "learning_rate": 0.003,
+      "loss": 4.2817,
+      "step": 1682
+    },
+    {
+      "epoch": 0.01683,
+      "grad_norm": 0.5910768318994952,
+      "learning_rate": 0.003,
+      "loss": 4.2959,
+      "step": 1683
+    },
+    {
+      "epoch": 0.01684,
+      "grad_norm": 0.6050344929193974,
+      "learning_rate": 0.003,
+      "loss": 4.2707,
+      "step": 1684
+    },
+    {
+      "epoch": 0.01685,
+      "grad_norm": 0.6382986956510986,
+      "learning_rate": 0.003,
+      "loss": 4.2898,
+      "step": 1685
+    },
+    {
+      "epoch": 0.01686,
+      "grad_norm": 0.8077022270811443,
+      "learning_rate": 0.003,
+      "loss": 4.2942,
+      "step": 1686
+    },
+    {
+      "epoch": 0.01687,
+      "grad_norm": 0.8169553039156104,
+      "learning_rate": 0.003,
+      "loss": 4.2925,
+      "step": 1687
+    },
+    {
+      "epoch": 0.01688,
+      "grad_norm": 0.8368087154638648,
+      "learning_rate": 0.003,
+      "loss": 4.2843,
+      "step": 1688
+    },
+    {
+      "epoch": 0.01689,
+      "grad_norm": 0.869445348031783,
+      "learning_rate": 0.003,
+      "loss": 4.3002,
+      "step": 1689
+    },
+    {
+      "epoch": 0.0169,
+      "grad_norm": 0.9062293741017381,
+      "learning_rate": 0.003,
+      "loss": 4.2906,
+      "step": 1690
+    },
+    {
+      "epoch": 0.01691,
+      "grad_norm": 0.8487759964259872,
+      "learning_rate": 0.003,
+      "loss": 4.3075,
+      "step": 1691
+    },
+    {
+      "epoch": 0.01692,
+      "grad_norm": 0.8483641489848167,
+      "learning_rate": 0.003,
+      "loss": 4.3088,
+      "step": 1692
+    },
+    {
+      "epoch": 0.01693,
+      "grad_norm": 0.8279189520542338,
+      "learning_rate": 0.003,
+      "loss": 4.3369,
+      "step": 1693
+    },
+    {
+      "epoch": 0.01694,
+      "grad_norm": 0.8089873138093994,
+      "learning_rate": 0.003,
+      "loss": 4.2887,
+      "step": 1694
+    },
+    {
+      "epoch": 0.01695,
+      "grad_norm": 0.9455670720071184,
+      "learning_rate": 0.003,
+      "loss": 4.3248,
+      "step": 1695
+    },
+    {
+      "epoch": 0.01696,
+      "grad_norm": 1.064016692244757,
+      "learning_rate": 0.003,
+      "loss": 4.3315,
+      "step": 1696
+    },
+    {
+      "epoch": 0.01697,
+      "grad_norm": 0.9264587036694925,
+      "learning_rate": 0.003,
+      "loss": 4.302,
+      "step": 1697
+    },
+    {
+      "epoch": 0.01698,
+      "grad_norm": 0.6997503897611733,
+      "learning_rate": 0.003,
+      "loss": 4.289,
+      "step": 1698
+    },
+    {
+      "epoch": 0.01699,
+      "grad_norm": 0.6394457308265625,
+      "learning_rate": 0.003,
+      "loss": 4.2969,
+      "step": 1699
+    },
+    {
+      "epoch": 0.017,
+      "grad_norm": 0.5891910529153762,
+      "learning_rate": 0.003,
+      "loss": 4.3094,
+      "step": 1700
+    },
+    {
+      "epoch": 0.01701,
+      "grad_norm": 0.6435550929707158,
+      "learning_rate": 0.003,
+      "loss": 4.2939,
+      "step": 1701
+    },
+    {
+      "epoch": 0.01702,
+      "grad_norm": 0.6810306253220882,
+      "learning_rate": 0.003,
+      "loss": 4.3163,
+      "step": 1702
+    },
+    {
+      "epoch": 0.01703,
+      "grad_norm": 0.6849363850272787,
+      "learning_rate": 0.003,
+      "loss": 4.3034,
+      "step": 1703
+    },
+    {
+      "epoch": 0.01704,
+      "grad_norm": 0.6708561417744424,
+      "learning_rate": 0.003,
+      "loss": 4.2869,
+      "step": 1704
+    },
+    {
+      "epoch": 0.01705,
+      "grad_norm": 0.5968699215981799,
+      "learning_rate": 0.003,
+      "loss": 4.2879,
+      "step": 1705
+    },
+    {
+      "epoch": 0.01706,
+      "grad_norm": 0.505702133399309,
+      "learning_rate": 0.003,
+      "loss": 4.2792,
+      "step": 1706
+    },
+    {
+      "epoch": 0.01707,
+      "grad_norm": 0.5302615087387699,
+      "learning_rate": 0.003,
+      "loss": 4.3011,
+      "step": 1707
+    },
+    {
+      "epoch": 0.01708,
+      "grad_norm": 0.5210003345198112,
+      "learning_rate": 0.003,
+      "loss": 4.2762,
+      "step": 1708
+    },
+    {
+      "epoch": 0.01709,
+      "grad_norm": 0.46584995061994,
+      "learning_rate": 0.003,
+      "loss": 4.2953,
+      "step": 1709
+    },
+    {
+      "epoch": 0.0171,
+      "grad_norm": 0.42345394468243425,
+      "learning_rate": 0.003,
+      "loss": 4.2414,
+      "step": 1710
+    },
+    {
+      "epoch": 0.01711,
+      "grad_norm": 0.43754438467909607,
+      "learning_rate": 0.003,
+      "loss": 4.2717,
+      "step": 1711
+    },
+    {
+      "epoch": 0.01712,
+      "grad_norm": 0.41267034274673536,
+      "learning_rate": 0.003,
+      "loss": 4.2801,
+      "step": 1712
+    },
+    {
+      "epoch": 0.01713,
+      "grad_norm": 0.3568885732041511,
+      "learning_rate": 0.003,
+      "loss": 4.2523,
+      "step": 1713
+    },
+    {
+      "epoch": 0.01714,
+      "grad_norm": 0.3457637645495226,
+      "learning_rate": 0.003,
+      "loss": 4.2876,
+      "step": 1714
+    },
+    {
+      "epoch": 0.01715,
+      "grad_norm": 0.3722023563902025,
+      "learning_rate": 0.003,
+      "loss": 4.2958,
+      "step": 1715
+    },
+    {
+      "epoch": 0.01716,
+      "grad_norm": 0.3390050605620623,
+      "learning_rate": 0.003,
+      "loss": 4.2697,
+      "step": 1716
+    },
+    {
+      "epoch": 0.01717,
+      "grad_norm": 0.4093688383515523,
+      "learning_rate": 0.003,
+      "loss": 4.3028,
+      "step": 1717
+    },
+    {
+      "epoch": 0.01718,
+      "grad_norm": 0.5445666194239754,
+      "learning_rate": 0.003,
+      "loss": 4.2735,
+      "step": 1718
+    },
+    {
+      "epoch": 0.01719,
+      "grad_norm": 0.8818554274614839,
+      "learning_rate": 0.003,
+      "loss": 4.3037,
+      "step": 1719
+    },
+    {
+      "epoch": 0.0172,
+      "grad_norm": 1.123998698583038,
+      "learning_rate": 0.003,
+      "loss": 4.2879,
+      "step": 1720
+    },
+    {
+      "epoch": 0.01721,
+      "grad_norm": 0.6518188394707584,
+      "learning_rate": 0.003,
+      "loss": 4.2815,
+      "step": 1721
+    },
+    {
+      "epoch": 0.01722,
+      "grad_norm": 0.7473961890284897,
+      "learning_rate": 0.003,
+      "loss": 4.281,
+      "step": 1722
+    },
+    {
+      "epoch": 0.01723,
+      "grad_norm": 0.7241082541176558,
+      "learning_rate": 0.003,
+      "loss": 4.2942,
+      "step": 1723
+    },
+    {
+      "epoch": 0.01724,
+      "grad_norm": 0.5406243852034243,
+      "learning_rate": 0.003,
+      "loss": 4.2839,
+      "step": 1724
+    },
+    {
+      "epoch": 0.01725,
+      "grad_norm": 0.7661047717663526,
+      "learning_rate": 0.003,
+      "loss": 4.2758,
+      "step": 1725
+    },
+    {
+      "epoch": 0.01726,
+      "grad_norm": 0.6519733535814134,
+      "learning_rate": 0.003,
+      "loss": 4.2678,
+      "step": 1726
+    },
+    {
+      "epoch": 0.01727,
+      "grad_norm": 0.5095447102684948,
+      "learning_rate": 0.003,
+      "loss": 4.2681,
+      "step": 1727
+    },
+    {
+      "epoch": 0.01728,
+      "grad_norm": 0.5716779713985857,
+      "learning_rate": 0.003,
+      "loss": 4.2703,
+      "step": 1728
+    },
+    {
+      "epoch": 0.01729,
+      "grad_norm": 0.5506488662519434,
+      "learning_rate": 0.003,
+      "loss": 4.2777,
+      "step": 1729
+    },
+    {
+      "epoch": 0.0173,
+      "grad_norm": 0.4864830783539529,
+      "learning_rate": 0.003,
+      "loss": 4.2498,
+      "step": 1730
+    },
+    {
+      "epoch": 0.01731,
+      "grad_norm": 0.49289174915320866,
+      "learning_rate": 0.003,
+      "loss": 4.2691,
+      "step": 1731
+    },
+    {
+      "epoch": 0.01732,
+      "grad_norm": 0.5258408981010323,
+      "learning_rate": 0.003,
+      "loss": 4.2782,
+      "step": 1732
+    },
+    {
+      "epoch": 0.01733,
+      "grad_norm": 0.5689089752248919,
+      "learning_rate": 0.003,
+      "loss": 4.2578,
+      "step": 1733
+    },
+    {
+      "epoch": 0.01734,
+      "grad_norm": 0.5515499021599949,
+      "learning_rate": 0.003,
+      "loss": 4.2704,
+      "step": 1734
+    },
+    {
+      "epoch": 0.01735,
+      "grad_norm": 0.5101191368780025,
+      "learning_rate": 0.003,
+      "loss": 4.2394,
+      "step": 1735
+    },
+    {
+      "epoch": 0.01736,
+      "grad_norm": 0.5538259418110674,
+      "learning_rate": 0.003,
+      "loss": 4.2405,
+      "step": 1736
+    },
+    {
+      "epoch": 0.01737,
+      "grad_norm": 0.6368154932756357,
+      "learning_rate": 0.003,
+      "loss": 4.273,
+      "step": 1737
+    },
+    {
+      "epoch": 0.01738,
+      "grad_norm": 0.6302527957263243,
+      "learning_rate": 0.003,
+      "loss": 4.2633,
+      "step": 1738
+    },
+    {
+      "epoch": 0.01739,
+      "grad_norm": 0.5659864529447296,
+      "learning_rate": 0.003,
+      "loss": 4.2757,
+      "step": 1739
+    },
+    {
+      "epoch": 0.0174,
+      "grad_norm": 0.6034857544005698,
+      "learning_rate": 0.003,
+      "loss": 4.278,
+      "step": 1740
+    },
+    {
+      "epoch": 0.01741,
+      "grad_norm": 0.7023281656557057,
+      "learning_rate": 0.003,
+      "loss": 4.2883,
+      "step": 1741
+    },
+    {
+      "epoch": 0.01742,
+      "grad_norm": 0.7127714842932543,
+      "learning_rate": 0.003,
+      "loss": 4.2766,
+      "step": 1742
+    },
+    {
+      "epoch": 0.01743,
+      "grad_norm": 0.7178940660267726,
+      "learning_rate": 0.003,
+      "loss": 4.2889,
+      "step": 1743
+    },
+    {
+      "epoch": 0.01744,
+      "grad_norm": 0.770440209657255,
+      "learning_rate": 0.003,
+      "loss": 4.2773,
+      "step": 1744
+    },
+    {
+      "epoch": 0.01745,
+      "grad_norm": 0.7362175602563142,
+      "learning_rate": 0.003,
+      "loss": 4.2484,
+      "step": 1745
+    },
+    {
+      "epoch": 0.01746,
+      "grad_norm": 0.7588375597390518,
+      "learning_rate": 0.003,
+      "loss": 4.2719,
+      "step": 1746
+    },
+    {
+      "epoch": 0.01747,
+      "grad_norm": 0.6839956521205998,
+      "learning_rate": 0.003,
+      "loss": 4.267,
+      "step": 1747
+    },
+    {
+      "epoch": 0.01748,
+      "grad_norm": 0.6088836411369196,
+      "learning_rate": 0.003,
+      "loss": 4.2555,
+      "step": 1748
+    },
+    {
+      "epoch": 0.01749,
+      "grad_norm": 0.548971540706334,
+      "learning_rate": 0.003,
+      "loss": 4.2516,
+      "step": 1749
+    },
+    {
+      "epoch": 0.0175,
+      "grad_norm": 0.5440778124239496,
+      "learning_rate": 0.003,
+      "loss": 4.2665,
+      "step": 1750
+    },
+    {
+      "epoch": 0.01751,
+      "grad_norm": 0.5918941393242653,
+      "learning_rate": 0.003,
+      "loss": 4.2593,
+      "step": 1751
+    },
+    {
+      "epoch": 0.01752,
+      "grad_norm": 0.6455620934648258,
+      "learning_rate": 0.003,
+      "loss": 4.3014,
+      "step": 1752
+    },
+    {
+      "epoch": 0.01753,
+      "grad_norm": 0.6781800073296501,
+      "learning_rate": 0.003,
+      "loss": 4.2535,
+      "step": 1753
+    },
+    {
+      "epoch": 0.01754,
+      "grad_norm": 0.6882980076107725,
+      "learning_rate": 0.003,
+      "loss": 4.2853,
+      "step": 1754
+    },
+    {
+      "epoch": 0.01755,
+      "grad_norm": 0.7961632146248975,
+      "learning_rate": 0.003,
+      "loss": 4.233,
+      "step": 1755
+    },
+    {
+      "epoch": 0.01756,
+      "grad_norm": 0.8494780429534167,
+      "learning_rate": 0.003,
+      "loss": 4.314,
+      "step": 1756
+    },
+    {
+      "epoch": 0.01757,
+      "grad_norm": 0.9294824362906122,
+      "learning_rate": 0.003,
+      "loss": 4.2987,
+      "step": 1757
+    },
+    {
+      "epoch": 0.01758,
+      "grad_norm": 0.852175947314296,
+      "learning_rate": 0.003,
+      "loss": 4.3126,
+      "step": 1758
+    },
+    {
+      "epoch": 0.01759,
+      "grad_norm": 0.7347749740307682,
+      "learning_rate": 0.003,
+      "loss": 4.2719,
+      "step": 1759
+    },
+    {
+      "epoch": 0.0176,
+      "grad_norm": 0.832142747460363,
+      "learning_rate": 0.003,
+      "loss": 4.2994,
+      "step": 1760
+    },
+    {
+      "epoch": 0.01761,
+      "grad_norm": 0.8686282951443787,
+      "learning_rate": 0.003,
+      "loss": 4.2965,
+      "step": 1761
+    },
+    {
+      "epoch": 0.01762,
+      "grad_norm": 0.9408046252281581,
+      "learning_rate": 0.003,
+      "loss": 4.3172,
+      "step": 1762
+    },
+    {
+      "epoch": 0.01763,
+      "grad_norm": 0.8594428065536988,
+      "learning_rate": 0.003,
+      "loss": 4.3092,
+      "step": 1763
+    },
+    {
+      "epoch": 0.01764,
+      "grad_norm": 0.7749341487974934,
+      "learning_rate": 0.003,
+      "loss": 4.2754,
+      "step": 1764
+    },
+    {
+      "epoch": 0.01765,
+      "grad_norm": 0.7897730382355919,
+      "learning_rate": 0.003,
+      "loss": 4.3076,
+      "step": 1765
+    },
+    {
+      "epoch": 0.01766,
+      "grad_norm": 0.7174878978412247,
+      "learning_rate": 0.003,
+      "loss": 4.317,
+      "step": 1766
+    },
+    {
+      "epoch": 0.01767,
+      "grad_norm": 0.7171956437495152,
+      "learning_rate": 0.003,
+      "loss": 4.2757,
+      "step": 1767
+    },
+    {
+      "epoch": 0.01768,
+      "grad_norm": 0.8077487259776668,
+      "learning_rate": 0.003,
+      "loss": 4.2866,
+      "step": 1768
+    },
+    {
+      "epoch": 0.01769,
+      "grad_norm": 0.7100605985326003,
+      "learning_rate": 0.003,
+      "loss": 4.2811,
+      "step": 1769
+    },
+    {
+      "epoch": 0.0177,
+      "grad_norm": 0.578964917508603,
+      "learning_rate": 0.003,
+      "loss": 4.2635,
+      "step": 1770
+    },
+    {
+      "epoch": 0.01771,
+      "grad_norm": 0.6086271828330341,
+      "learning_rate": 0.003,
+      "loss": 4.2929,
+      "step": 1771
+    },
+    {
+      "epoch": 0.01772,
+      "grad_norm": 0.497546957954685,
+      "learning_rate": 0.003,
+      "loss": 4.262,
+      "step": 1772
+    },
+    {
+      "epoch": 0.01773,
+      "grad_norm": 0.490787842661778,
+      "learning_rate": 0.003,
+      "loss": 4.2539,
+      "step": 1773
+    },
+    {
+      "epoch": 0.01774,
+      "grad_norm": 0.40528808868462435,
+      "learning_rate": 0.003,
+      "loss": 4.272,
+      "step": 1774
+    },
+    {
+      "epoch": 0.01775,
+      "grad_norm": 0.3996997018074339,
+      "learning_rate": 0.003,
+      "loss": 4.2574,
+      "step": 1775
+    },
+    {
+      "epoch": 0.01776,
+      "grad_norm": 0.40093115740510726,
+      "learning_rate": 0.003,
+      "loss": 4.2491,
+      "step": 1776
+    },
+    {
+      "epoch": 0.01777,
+      "grad_norm": 0.4577551703461772,
+      "learning_rate": 0.003,
+      "loss": 4.27,
+      "step": 1777
+    },
+    {
+      "epoch": 0.01778,
+      "grad_norm": 0.4894775985846586,
+      "learning_rate": 0.003,
+      "loss": 4.262,
+      "step": 1778
+    },
+    {
+      "epoch": 0.01779,
+      "grad_norm": 0.619715296261353,
+      "learning_rate": 0.003,
+      "loss": 4.2821,
+      "step": 1779
+    },
+    {
+      "epoch": 0.0178,
+      "grad_norm": 0.8092821046637155,
+      "learning_rate": 0.003,
+      "loss": 4.2576,
+      "step": 1780
+    },
+    {
+      "epoch": 0.01781,
+      "grad_norm": 1.007368746628607,
+      "learning_rate": 0.003,
+      "loss": 4.2833,
+      "step": 1781
+    },
+    {
+      "epoch": 0.01782,
+      "grad_norm": 1.0605028910459124,
+      "learning_rate": 0.003,
+      "loss": 4.314,
+      "step": 1782
+    },
+    {
+      "epoch": 0.01783,
+      "grad_norm": 0.8361916169047299,
+      "learning_rate": 0.003,
+      "loss": 4.2912,
+      "step": 1783
+    },
+    {
+      "epoch": 0.01784,
+      "grad_norm": 0.9950086095977337,
+      "learning_rate": 0.003,
+      "loss": 4.2963,
+      "step": 1784
+    },
+    {
+      "epoch": 0.01785,
+      "grad_norm": 1.0228877834258285,
+      "learning_rate": 0.003,
+      "loss": 4.2673,
+      "step": 1785
+    },
+    {
+      "epoch": 0.01786,
+      "grad_norm": 0.8766088407747356,
+      "learning_rate": 0.003,
+      "loss": 4.2899,
+      "step": 1786
+    },
+    {
+      "epoch": 0.01787,
+      "grad_norm": 0.9346024274483619,
+      "learning_rate": 0.003,
+      "loss": 4.2991,
+      "step": 1787
+    },
+    {
+      "epoch": 0.01788,
+      "grad_norm": 0.8655394124739175,
+      "learning_rate": 0.003,
+      "loss": 4.3097,
+      "step": 1788
+    },
+    {
+      "epoch": 0.01789,
+      "grad_norm": 0.7320722132470949,
+      "learning_rate": 0.003,
+      "loss": 4.2742,
+      "step": 1789
+    },
+    {
+      "epoch": 0.0179,
+      "grad_norm": 0.6043898276387406,
+      "learning_rate": 0.003,
+      "loss": 4.288,
+      "step": 1790
+    },
+    {
+      "epoch": 0.01791,
+      "grad_norm": 0.551731430114501,
+      "learning_rate": 0.003,
+      "loss": 4.2984,
+      "step": 1791
+    },
+    {
+      "epoch": 0.01792,
+      "grad_norm": 0.4116411374695503,
+      "learning_rate": 0.003,
+      "loss": 4.2944,
+      "step": 1792
+    },
+    {
+      "epoch": 0.01793,
+      "grad_norm": 0.4368353784765224,
+      "learning_rate": 0.003,
+      "loss": 4.2574,
+      "step": 1793
+    },
+    {
+      "epoch": 0.01794,
+      "grad_norm": 0.4134683249639472,
+      "learning_rate": 0.003,
+      "loss": 4.2696,
+      "step": 1794
+    },
+    {
+      "epoch": 0.01795,
+      "grad_norm": 0.3898886643379656,
+      "learning_rate": 0.003,
+      "loss": 4.2661,
+      "step": 1795
+    },
+    {
+      "epoch": 0.01796,
+      "grad_norm": 0.3690232703108766,
+      "learning_rate": 0.003,
+      "loss": 4.2797,
+      "step": 1796
+    },
+    {
+      "epoch": 0.01797,
+      "grad_norm": 0.3352292075992397,
+      "learning_rate": 0.003,
+      "loss": 4.2421,
+      "step": 1797
+    },
+    {
+      "epoch": 0.01798,
+      "grad_norm": 0.3577790623133374,
+      "learning_rate": 0.003,
+      "loss": 4.2547,
+      "step": 1798
+    },
+    {
+      "epoch": 0.01799,
+      "grad_norm": 0.40125449243674877,
+      "learning_rate": 0.003,
+      "loss": 4.269,
+      "step": 1799
+    },
+    {
+      "epoch": 0.018,
+      "grad_norm": 0.4677938601833102,
+      "learning_rate": 0.003,
+      "loss": 4.26,
+      "step": 1800
+    },
+    {
+      "epoch": 0.01801,
+      "grad_norm": 0.5610552208097703,
+      "learning_rate": 0.003,
+      "loss": 4.2554,
+      "step": 1801
+    },
+    {
+      "epoch": 0.01802,
+      "grad_norm": 0.5864236478232465,
+      "learning_rate": 0.003,
+      "loss": 4.2736,
+      "step": 1802
+    },
+    {
+      "epoch": 0.01803,
+      "grad_norm": 0.540429672022444,
+      "learning_rate": 0.003,
+      "loss": 4.2776,
+      "step": 1803
+    },
+    {
+      "epoch": 0.01804,
+      "grad_norm": 0.60459180760597,
+      "learning_rate": 0.003,
+      "loss": 4.2503,
+      "step": 1804
+    },
+    {
+      "epoch": 0.01805,
+      "grad_norm": 0.6945199909530504,
+      "learning_rate": 0.003,
+      "loss": 4.2727,
+      "step": 1805
+    },
+    {
+      "epoch": 0.01806,
+      "grad_norm": 0.6888751318415884,
+      "learning_rate": 0.003,
+      "loss": 4.2338,
+      "step": 1806
+    },
+    {
+      "epoch": 0.01807,
+      "grad_norm": 0.652641222784165,
+      "learning_rate": 0.003,
+      "loss": 4.2674,
+      "step": 1807
+    },
+    {
+      "epoch": 0.01808,
+      "grad_norm": 0.7163920878946057,
+      "learning_rate": 0.003,
+      "loss": 4.2613,
+      "step": 1808
+    },
+    {
+      "epoch": 0.01809,
+      "grad_norm": 0.7231527059402414,
+      "learning_rate": 0.003,
+      "loss": 4.284,
+      "step": 1809
+    },
+    {
+      "epoch": 0.0181,
+      "grad_norm": 0.7979758654578836,
+      "learning_rate": 0.003,
+      "loss": 4.2839,
+      "step": 1810
+    },
+    {
+      "epoch": 0.01811,
+      "grad_norm": 0.7630566976684018,
+      "learning_rate": 0.003,
+      "loss": 4.2692,
+      "step": 1811
+    },
+    {
+      "epoch": 0.01812,
+      "grad_norm": 0.68601985737994,
+      "learning_rate": 0.003,
+      "loss": 4.2545,
+      "step": 1812
+    },
+    {
+      "epoch": 0.01813,
+      "grad_norm": 0.601926611394952,
+      "learning_rate": 0.003,
+      "loss": 4.2949,
+      "step": 1813
+    },
+    {
+      "epoch": 0.01814,
+      "grad_norm": 0.66852657571521,
+      "learning_rate": 0.003,
+      "loss": 4.2647,
+      "step": 1814
+    },
+    {
+      "epoch": 0.01815,
+      "grad_norm": 0.6050905896281803,
+      "learning_rate": 0.003,
+      "loss": 4.2474,
+      "step": 1815
+    },
+    {
+      "epoch": 0.01816,
+      "grad_norm": 0.531953848915138,
+      "learning_rate": 0.003,
+      "loss": 4.2731,
+      "step": 1816
+    },
+    {
+      "epoch": 0.01817,
+      "grad_norm": 0.6652005043351588,
+      "learning_rate": 0.003,
+      "loss": 4.2676,
+      "step": 1817
+    },
+    {
+      "epoch": 0.01818,
+      "grad_norm": 0.6653492631522442,
+      "learning_rate": 0.003,
+      "loss": 4.2669,
+      "step": 1818
+    },
+    {
+      "epoch": 0.01819,
+      "grad_norm": 0.6430139858317634,
+      "learning_rate": 0.003,
+      "loss": 4.2522,
+      "step": 1819
+    },
+    {
+      "epoch": 0.0182,
+      "grad_norm": 0.7307512465670702,
+      "learning_rate": 0.003,
+      "loss": 4.2731,
+      "step": 1820
+    },
+    {
+      "epoch": 0.01821,
+      "grad_norm": 0.7117919312146421,
+      "learning_rate": 0.003,
+      "loss": 4.2608,
+      "step": 1821
+    },
+    {
+      "epoch": 0.01822,
+      "grad_norm": 0.6849992334976018,
+      "learning_rate": 0.003,
+      "loss": 4.2737,
+      "step": 1822
+    },
+    {
+      "epoch": 0.01823,
+      "grad_norm": 0.6437329965546797,
+      "learning_rate": 0.003,
+      "loss": 4.2861,
+      "step": 1823
+    },
+    {
+      "epoch": 0.01824,
+      "grad_norm": 0.6298055587951418,
+      "learning_rate": 0.003,
+      "loss": 4.2663,
+      "step": 1824
+    },
+    {
+      "epoch": 0.01825,
+      "grad_norm": 0.6105471946808345,
+      "learning_rate": 0.003,
+      "loss": 4.2706,
+      "step": 1825
+    },
+    {
+      "epoch": 0.01826,
+      "grad_norm": 0.5680736127986394,
+      "learning_rate": 0.003,
+      "loss": 4.2798,
+      "step": 1826
+    },
+    {
+      "epoch": 0.01827,
+      "grad_norm": 0.6203032072106797,
+      "learning_rate": 0.003,
+      "loss": 4.2493,
+      "step": 1827
+    },
+    {
+      "epoch": 0.01828,
+      "grad_norm": 0.5646483418870301,
+      "learning_rate": 0.003,
+      "loss": 4.2595,
+      "step": 1828
+    },
+    {
+      "epoch": 0.01829,
+      "grad_norm": 0.43449345506164244,
+      "learning_rate": 0.003,
+      "loss": 4.2724,
+      "step": 1829
+    },
+    {
+      "epoch": 0.0183,
+      "grad_norm": 0.4995474978675237,
+      "learning_rate": 0.003,
+      "loss": 4.268,
+      "step": 1830
+    },
+    {
+      "epoch": 0.01831,
+      "grad_norm": 0.6731917230470822,
+      "learning_rate": 0.003,
+      "loss": 4.2625,
+      "step": 1831
+    },
+    {
+      "epoch": 0.01832,
+      "grad_norm": 1.0735909763241713,
+      "learning_rate": 0.003,
+      "loss": 4.2749,
+      "step": 1832
+    },
+    {
+      "epoch": 0.01833,
+      "grad_norm": 0.9454987640162065,
+      "learning_rate": 0.003,
+      "loss": 4.3126,
+      "step": 1833
+    },
+    {
+      "epoch": 0.01834,
+      "grad_norm": 0.6279857981331142,
+      "learning_rate": 0.003,
+      "loss": 4.265,
+      "step": 1834
+    },
+    {
+      "epoch": 0.01835,
+      "grad_norm": 0.7795333868843384,
+      "learning_rate": 0.003,
+      "loss": 4.2708,
+      "step": 1835
+    },
+    {
+      "epoch": 0.01836,
+      "grad_norm": 0.7540135177559686,
+      "learning_rate": 0.003,
+      "loss": 4.269,
+      "step": 1836
+    },
+    {
+      "epoch": 0.01837,
+      "grad_norm": 0.4873694552023737,
+      "learning_rate": 0.003,
+      "loss": 4.2594,
+      "step": 1837
+    },
+    {
+      "epoch": 0.01838,
+      "grad_norm": 0.6475671331829288,
+      "learning_rate": 0.003,
+      "loss": 4.2679,
+      "step": 1838
+    },
+    {
+      "epoch": 0.01839,
+      "grad_norm": 0.5413345448794652,
+      "learning_rate": 0.003,
+      "loss": 4.2849,
+      "step": 1839
+    },
+    {
+      "epoch": 0.0184,
+      "grad_norm": 0.505876271981352,
+      "learning_rate": 0.003,
+      "loss": 4.2815,
+      "step": 1840
+    },
+    {
+      "epoch": 0.01841,
+      "grad_norm": 0.5559518639320817,
+      "learning_rate": 0.003,
+      "loss": 4.2211,
+      "step": 1841
+    },
+    {
+      "epoch": 0.01842,
+      "grad_norm": 0.565402658078461,
+      "learning_rate": 0.003,
+      "loss": 4.2563,
+      "step": 1842
+    },
+    {
+      "epoch": 0.01843,
+      "grad_norm": 0.5176031889881334,
+      "learning_rate": 0.003,
+      "loss": 4.2347,
+      "step": 1843
+    },
+    {
+      "epoch": 0.01844,
+      "grad_norm": 0.515476660965344,
+      "learning_rate": 0.003,
+      "loss": 4.2336,
+      "step": 1844
+    },
+    {
+      "epoch": 0.01845,
+      "grad_norm": 0.537738729997812,
+      "learning_rate": 0.003,
+      "loss": 4.2366,
+      "step": 1845
+    },
+    {
+      "epoch": 0.01846,
+      "grad_norm": 0.5825872331164712,
+      "learning_rate": 0.003,
+      "loss": 4.2629,
+      "step": 1846
+    },
+    {
+      "epoch": 0.01847,
+      "grad_norm": 0.6172426924095343,
+      "learning_rate": 0.003,
+      "loss": 4.261,
+      "step": 1847
+    },
+    {
+      "epoch": 0.01848,
+      "grad_norm": 0.5675520394460739,
+      "learning_rate": 0.003,
+      "loss": 4.2733,
+      "step": 1848
+    },
+    {
+      "epoch": 0.01849,
+      "grad_norm": 0.5804129655862207,
+      "learning_rate": 0.003,
+      "loss": 4.2733,
+      "step": 1849
+    },
+    {
+      "epoch": 0.0185,
+      "grad_norm": 0.5903509862738283,
+      "learning_rate": 0.003,
+      "loss": 4.2637,
+      "step": 1850
+    },
+    {
+      "epoch": 0.01851,
+      "grad_norm": 0.5789415751735032,
+      "learning_rate": 0.003,
+      "loss": 4.2453,
+      "step": 1851
+    },
+    {
+      "epoch": 0.01852,
+      "grad_norm": 0.6073114122094332,
+      "learning_rate": 0.003,
+      "loss": 4.249,
+      "step": 1852
+    },
+    {
+      "epoch": 0.01853,
+      "grad_norm": 0.6172059367670675,
+      "learning_rate": 0.003,
+      "loss": 4.2514,
+      "step": 1853
+    },
+    {
+      "epoch": 0.01854,
+      "grad_norm": 0.5952391168819953,
+      "learning_rate": 0.003,
+      "loss": 4.282,
+      "step": 1854
+    },
+    {
+      "epoch": 0.01855,
+      "grad_norm": 0.599929499071102,
+      "learning_rate": 0.003,
+      "loss": 4.2645,
+      "step": 1855
+    },
+    {
+      "epoch": 0.01856,
+      "grad_norm": 0.7854590983227943,
+      "learning_rate": 0.003,
+      "loss": 4.2582,
+      "step": 1856
+    },
+    {
+      "epoch": 0.01857,
+      "grad_norm": 0.9479235992175757,
+      "learning_rate": 0.003,
+      "loss": 4.2521,
+      "step": 1857
+    },
+    {
+      "epoch": 0.01858,
+      "grad_norm": 1.0547185212118042,
+      "learning_rate": 0.003,
+      "loss": 4.2782,
+      "step": 1858
+    },
+    {
+      "epoch": 0.01859,
+      "grad_norm": 0.7498301751007965,
+      "learning_rate": 0.003,
+      "loss": 4.259,
+      "step": 1859
+    },
+    {
+      "epoch": 0.0186,
+      "grad_norm": 0.6772314759733408,
+      "learning_rate": 0.003,
+      "loss": 4.238,
+      "step": 1860
+    },
+    {
+      "epoch": 0.01861,
+      "grad_norm": 0.7554457171283483,
+      "learning_rate": 0.003,
+      "loss": 4.2858,
+      "step": 1861
+    },
+    {
+      "epoch": 0.01862,
+      "grad_norm": 0.6601398192741809,
+      "learning_rate": 0.003,
+      "loss": 4.2482,
+      "step": 1862
+    },
+    {
+      "epoch": 0.01863,
+      "grad_norm": 0.5632517970928785,
+      "learning_rate": 0.003,
+      "loss": 4.2548,
+      "step": 1863
+    },
+    {
+      "epoch": 0.01864,
+      "grad_norm": 0.5518520472207163,
+      "learning_rate": 0.003,
+      "loss": 4.2848,
+      "step": 1864
+    },
+    {
+      "epoch": 0.01865,
+      "grad_norm": 0.5191605026436766,
+      "learning_rate": 0.003,
+      "loss": 4.2571,
+      "step": 1865
+    },
+    {
+      "epoch": 0.01866,
+      "grad_norm": 0.487977336215745,
+      "learning_rate": 0.003,
+      "loss": 4.2583,
+      "step": 1866
+    },
+    {
+      "epoch": 0.01867,
+      "grad_norm": 0.38111750510593084,
+      "learning_rate": 0.003,
+      "loss": 4.2288,
+      "step": 1867
+    },
+    {
+      "epoch": 0.01868,
+      "grad_norm": 0.36979378128361096,
+      "learning_rate": 0.003,
+      "loss": 4.2458,
+      "step": 1868
+    },
+    {
+      "epoch": 0.01869,
+      "grad_norm": 0.4056719991155462,
+      "learning_rate": 0.003,
+      "loss": 4.259,
+      "step": 1869
+    },
+    {
+      "epoch": 0.0187,
+      "grad_norm": 0.511550197815793,
+      "learning_rate": 0.003,
+      "loss": 4.2534,
+      "step": 1870
+    },
+    {
+      "epoch": 0.01871,
+      "grad_norm": 0.6908775527772675,
+      "learning_rate": 0.003,
+      "loss": 4.2322,
+      "step": 1871
+    },
+    {
+      "epoch": 0.01872,
+      "grad_norm": 0.8655201254185445,
+      "learning_rate": 0.003,
+      "loss": 4.2892,
+      "step": 1872
+    },
+    {
+      "epoch": 0.01873,
+      "grad_norm": 0.903028038452321,
+      "learning_rate": 0.003,
+      "loss": 4.2527,
+      "step": 1873
+    },
+    {
+      "epoch": 0.01874,
+      "grad_norm": 0.7655189256325052,
+      "learning_rate": 0.003,
+      "loss": 4.2534,
+      "step": 1874
+    },
+    {
+      "epoch": 0.01875,
+      "grad_norm": 0.6845228657861865,
+      "learning_rate": 0.003,
+      "loss": 4.2697,
+      "step": 1875
+    },
+    {
+      "epoch": 0.01876,
+      "grad_norm": 0.7967028641532641,
+      "learning_rate": 0.003,
+      "loss": 4.2796,
+      "step": 1876
+    },
+    {
+      "epoch": 0.01877,
+      "grad_norm": 0.7276633971950639,
+      "learning_rate": 0.003,
+      "loss": 4.2792,
+      "step": 1877
+    },
+    {
+      "epoch": 0.01878,
+      "grad_norm": 0.6745755254962531,
+      "learning_rate": 0.003,
+      "loss": 4.2647,
+      "step": 1878
+    },
+    {
+      "epoch": 0.01879,
+      "grad_norm": 0.6794948462625433,
+      "learning_rate": 0.003,
+      "loss": 4.2421,
+      "step": 1879
+    },
+    {
+      "epoch": 0.0188,
+      "grad_norm": 0.7496570961239313,
+      "learning_rate": 0.003,
+      "loss": 4.2581,
+      "step": 1880
+    },
+    {
+      "epoch": 0.01881,
+      "grad_norm": 0.7216177452798419,
+      "learning_rate": 0.003,
+      "loss": 4.2527,
+      "step": 1881
+    },
+    {
+      "epoch": 0.01882,
+      "grad_norm": 0.6411166720639763,
+      "learning_rate": 0.003,
+      "loss": 4.2764,
+      "step": 1882
+    },
+    {
+      "epoch": 0.01883,
+      "grad_norm": 0.4976595430995637,
+      "learning_rate": 0.003,
+      "loss": 4.2463,
+      "step": 1883
+    },
+    {
+      "epoch": 0.01884,
+      "grad_norm": 0.5806395175516162,
+      "learning_rate": 0.003,
+      "loss": 4.2463,
+      "step": 1884
+    },
+    {
+      "epoch": 0.01885,
+      "grad_norm": 0.5932999040186268,
+      "learning_rate": 0.003,
+      "loss": 4.2267,
+      "step": 1885
+    },
+    {
+      "epoch": 0.01886,
+      "grad_norm": 0.6243294916603833,
+      "learning_rate": 0.003,
+      "loss": 4.2637,
+      "step": 1886
+    },
+    {
+      "epoch": 0.01887,
+      "grad_norm": 0.6594473434297762,
+      "learning_rate": 0.003,
+      "loss": 4.2583,
+      "step": 1887
+    },
+    {
+      "epoch": 0.01888,
+      "grad_norm": 0.5833874852548948,
+      "learning_rate": 0.003,
+      "loss": 4.2402,
+      "step": 1888
+    },
+    {
+      "epoch": 0.01889,
+      "grad_norm": 0.5554783687186964,
+      "learning_rate": 0.003,
+      "loss": 4.2547,
+      "step": 1889
+    },
+    {
+      "epoch": 0.0189,
+      "grad_norm": 0.5267765258339018,
+      "learning_rate": 0.003,
+      "loss": 4.2329,
+      "step": 1890
+    },
+    {
+      "epoch": 0.01891,
+      "grad_norm": 0.5533738473996538,
+      "learning_rate": 0.003,
+      "loss": 4.2819,
+      "step": 1891
+    },
+    {
+      "epoch": 0.01892,
+      "grad_norm": 0.5233002886731548,
+      "learning_rate": 0.003,
+      "loss": 4.2568,
+      "step": 1892
+    },
+    {
+      "epoch": 0.01893,
+      "grad_norm": 0.598656523235247,
+      "learning_rate": 0.003,
+      "loss": 4.2415,
+      "step": 1893
+    },
+    {
+      "epoch": 0.01894,
+      "grad_norm": 0.6329032192123055,
+      "learning_rate": 0.003,
+      "loss": 4.2498,
+      "step": 1894
+    },
+    {
+      "epoch": 0.01895,
+      "grad_norm": 0.7448049826558554,
+      "learning_rate": 0.003,
+      "loss": 4.2404,
+      "step": 1895
+    },
+    {
+      "epoch": 0.01896,
+      "grad_norm": 0.7896808634621466,
+      "learning_rate": 0.003,
+      "loss": 4.2666,
+      "step": 1896
+    },
+    {
+      "epoch": 0.01897,
+      "grad_norm": 0.8670691506268475,
+      "learning_rate": 0.003,
+      "loss": 4.267,
+      "step": 1897
+    },
+    {
+      "epoch": 0.01898,
+      "grad_norm": 0.7564766735493696,
+      "learning_rate": 0.003,
+      "loss": 4.2371,
+      "step": 1898
+    },
+    {
+      "epoch": 0.01899,
+      "grad_norm": 0.6014591021089313,
+      "learning_rate": 0.003,
+      "loss": 4.246,
+      "step": 1899
+    },
+    {
+      "epoch": 0.019,
+      "grad_norm": 0.6487238549786433,
+      "learning_rate": 0.003,
+      "loss": 4.2427,
+      "step": 1900
+    },
+    {
+      "epoch": 0.01901,
+      "grad_norm": 0.7335622998657767,
+      "learning_rate": 0.003,
+      "loss": 4.2939,
+      "step": 1901
+    },
+    {
+      "epoch": 0.01902,
+      "grad_norm": 0.727489215463111,
+      "learning_rate": 0.003,
+      "loss": 4.2706,
+      "step": 1902
+    },
+    {
+      "epoch": 0.01903,
+      "grad_norm": 0.6444511274857245,
+      "learning_rate": 0.003,
+      "loss": 4.2217,
+      "step": 1903
+    },
+    {
+      "epoch": 0.01904,
+      "grad_norm": 0.5669107570524424,
+      "learning_rate": 0.003,
+      "loss": 4.2738,
+      "step": 1904
+    },
+    {
+      "epoch": 0.01905,
+      "grad_norm": 0.47567496733321213,
+      "learning_rate": 0.003,
+      "loss": 4.2471,
+      "step": 1905
+    },
+    {
+      "epoch": 0.01906,
+      "grad_norm": 0.568871434515019,
+      "learning_rate": 0.003,
+      "loss": 4.2665,
+      "step": 1906
+    },
+    {
+      "epoch": 0.01907,
+      "grad_norm": 0.6437840413516579,
+      "learning_rate": 0.003,
+      "loss": 4.2605,
+      "step": 1907
+    },
+    {
+      "epoch": 0.01908,
+      "grad_norm": 0.8102395542085408,
+      "learning_rate": 0.003,
+      "loss": 4.2806,
+      "step": 1908
+    },
+    {
+      "epoch": 0.01909,
+      "grad_norm": 0.9019612301377007,
+      "learning_rate": 0.003,
+      "loss": 4.2522,
+      "step": 1909
+    },
+    {
+      "epoch": 0.0191,
+      "grad_norm": 0.8060528165222425,
+      "learning_rate": 0.003,
+      "loss": 4.2682,
+      "step": 1910
+    },
+    {
+      "epoch": 0.01911,
+      "grad_norm": 0.6514562643306322,
+      "learning_rate": 0.003,
+      "loss": 4.2341,
+      "step": 1911
+    },
+    {
+      "epoch": 0.01912,
+      "grad_norm": 0.7512556786074466,
+      "learning_rate": 0.003,
+      "loss": 4.2481,
+      "step": 1912
+    },
+    {
+      "epoch": 0.01913,
+      "grad_norm": 0.7206930768558686,
+      "learning_rate": 0.003,
+      "loss": 4.2688,
+      "step": 1913
+    },
+    {
+      "epoch": 0.01914,
+      "grad_norm": 0.6802559754696879,
+      "learning_rate": 0.003,
+      "loss": 4.2815,
+      "step": 1914
+    },
+    {
+      "epoch": 0.01915,
+      "grad_norm": 0.6329709283705148,
+      "learning_rate": 0.003,
+      "loss": 4.2674,
+      "step": 1915
+    },
+    {
+      "epoch": 0.01916,
+      "grad_norm": 0.6539184714683666,
+      "learning_rate": 0.003,
+      "loss": 4.2462,
+      "step": 1916
+    },
+    {
+      "epoch": 0.01917,
+      "grad_norm": 0.710899415708157,
+      "learning_rate": 0.003,
+      "loss": 4.2674,
+      "step": 1917
+    },
+    {
+      "epoch": 0.01918,
+      "grad_norm": 0.7167946113405885,
+      "learning_rate": 0.003,
+      "loss": 4.2562,
+      "step": 1918
+    },
+    {
+      "epoch": 0.01919,
+      "grad_norm": 0.7533071012807427,
+      "learning_rate": 0.003,
+      "loss": 4.2594,
+      "step": 1919
+    },
+    {
+      "epoch": 0.0192,
+      "grad_norm": 0.6703396884694851,
+      "learning_rate": 0.003,
+      "loss": 4.2612,
+      "step": 1920
+    },
+    {
+      "epoch": 0.01921,
+      "grad_norm": 0.5834644623189628,
+      "learning_rate": 0.003,
+      "loss": 4.2441,
+      "step": 1921
+    },
+    {
+      "epoch": 0.01922,
+      "grad_norm": 0.6481277683352221,
+      "learning_rate": 0.003,
+      "loss": 4.2625,
+      "step": 1922
+    },
+    {
+      "epoch": 0.01923,
+      "grad_norm": 0.6044219599617415,
+      "learning_rate": 0.003,
+      "loss": 4.2661,
+      "step": 1923
+    },
+    {
+      "epoch": 0.01924,
+      "grad_norm": 0.5337323934986105,
+      "learning_rate": 0.003,
+      "loss": 4.2643,
+      "step": 1924
+    },
+    {
+      "epoch": 0.01925,
+      "grad_norm": 0.5129044588459396,
+      "learning_rate": 0.003,
+      "loss": 4.251,
+      "step": 1925
+    },
+    {
+      "epoch": 0.01926,
+      "grad_norm": 0.48201171188935904,
+      "learning_rate": 0.003,
+      "loss": 4.2563,
+      "step": 1926
+    },
+    {
+      "epoch": 0.01927,
+      "grad_norm": 0.4295757083815179,
+      "learning_rate": 0.003,
+      "loss": 4.2456,
+      "step": 1927
+    },
+    {
+      "epoch": 0.01928,
+      "grad_norm": 0.5221994128637466,
+      "learning_rate": 0.003,
+      "loss": 4.2241,
+      "step": 1928
+    },
+    {
+      "epoch": 0.01929,
+      "grad_norm": 0.6151786259894082,
+      "learning_rate": 0.003,
+      "loss": 4.246,
+      "step": 1929
+    },
+    {
+      "epoch": 0.0193,
+      "grad_norm": 0.717185888834206,
+      "learning_rate": 0.003,
+      "loss": 4.2399,
+      "step": 1930
+    },
+    {
+      "epoch": 0.01931,
+      "grad_norm": 0.865419932280181,
+      "learning_rate": 0.003,
+      "loss": 4.2485,
+      "step": 1931
+    },
+    {
+      "epoch": 0.01932,
+      "grad_norm": 1.015142018105296,
+      "learning_rate": 0.003,
+      "loss": 4.2679,
+      "step": 1932
+    },
+    {
+      "epoch": 0.01933,
+      "grad_norm": 1.0909531598947217,
+      "learning_rate": 0.003,
+      "loss": 4.2582,
+      "step": 1933
+    },
+    {
+      "epoch": 0.01934,
+      "grad_norm": 0.7995257446738059,
+      "learning_rate": 0.003,
+      "loss": 4.2554,
+      "step": 1934
+    },
+    {
+      "epoch": 0.01935,
+      "grad_norm": 0.6434815333532291,
+      "learning_rate": 0.003,
+      "loss": 4.2669,
+      "step": 1935
+    },
+    {
+      "epoch": 0.01936,
+      "grad_norm": 0.6569925784458591,
+      "learning_rate": 0.003,
+      "loss": 4.2489,
+      "step": 1936
+    },
+    {
+      "epoch": 0.01937,
+      "grad_norm": 0.5996128457789688,
+      "learning_rate": 0.003,
+      "loss": 4.2577,
+      "step": 1937
+    },
+    {
+      "epoch": 0.01938,
+      "grad_norm": 0.6329170311784792,
+      "learning_rate": 0.003,
+      "loss": 4.2615,
+      "step": 1938
+    },
+    {
+      "epoch": 0.01939,
+      "grad_norm": 0.6354665323417178,
+      "learning_rate": 0.003,
+      "loss": 4.2808,
+      "step": 1939
+    },
+    {
+      "epoch": 0.0194,
+      "grad_norm": 0.6008328437037136,
+      "learning_rate": 0.003,
+      "loss": 4.2581,
+      "step": 1940
+    },
+    {
+      "epoch": 0.01941,
+      "grad_norm": 0.5343827591468221,
+      "learning_rate": 0.003,
+      "loss": 4.2378,
+      "step": 1941
+    },
+    {
+      "epoch": 0.01942,
+      "grad_norm": 0.5512416106954112,
+      "learning_rate": 0.003,
+      "loss": 4.2611,
+      "step": 1942
+    },
+    {
+      "epoch": 0.01943,
+      "grad_norm": 0.5866960320410021,
+      "learning_rate": 0.003,
+      "loss": 4.2341,
+      "step": 1943
+    },
+    {
+      "epoch": 0.01944,
+      "grad_norm": 0.6240489288808161,
+      "learning_rate": 0.003,
+      "loss": 4.2751,
+      "step": 1944
+    },
+    {
+      "epoch": 0.01945,
+      "grad_norm": 0.6243606216006328,
+      "learning_rate": 0.003,
+      "loss": 4.2437,
+      "step": 1945
+    },
+    {
+      "epoch": 0.01946,
+      "grad_norm": 0.5880226151273782,
+      "learning_rate": 0.003,
+      "loss": 4.2379,
+      "step": 1946
+    },
+    {
+      "epoch": 0.01947,
+      "grad_norm": 0.5934159043950513,
+      "learning_rate": 0.003,
+      "loss": 4.2471,
+      "step": 1947
+    },
+    {
+      "epoch": 0.01948,
+      "grad_norm": 0.5378813319639308,
+      "learning_rate": 0.003,
+      "loss": 4.2659,
+      "step": 1948
+    },
+    {
+      "epoch": 0.01949,
+      "grad_norm": 0.5777955568924656,
+      "learning_rate": 0.003,
+      "loss": 4.2595,
+      "step": 1949
+    },
+    {
+      "epoch": 0.0195,
+      "grad_norm": 0.6588236649280669,
+      "learning_rate": 0.003,
+      "loss": 4.2493,
+      "step": 1950
+    },
+    {
+      "epoch": 0.01951,
+      "grad_norm": 0.6662863362190039,
+      "learning_rate": 0.003,
+      "loss": 4.2457,
+      "step": 1951
+    },
+    {
+      "epoch": 0.01952,
+      "grad_norm": 0.6956758803553106,
+      "learning_rate": 0.003,
+      "loss": 4.2563,
+      "step": 1952
+    },
+    {
+      "epoch": 0.01953,
+      "grad_norm": 0.7309670323591311,
+      "learning_rate": 0.003,
+      "loss": 4.2663,
+      "step": 1953
+    },
+    {
+      "epoch": 0.01954,
+      "grad_norm": 0.7102958436956044,
+      "learning_rate": 0.003,
+      "loss": 4.2656,
+      "step": 1954
+    },
+    {
+      "epoch": 0.01955,
+      "grad_norm": 0.748927101392194,
+      "learning_rate": 0.003,
+      "loss": 4.2596,
+      "step": 1955
+    },
+    {
+      "epoch": 0.01956,
+      "grad_norm": 0.6357959051502906,
+      "learning_rate": 0.003,
+      "loss": 4.2507,
+      "step": 1956
+    },
+    {
+      "epoch": 0.01957,
+      "grad_norm": 0.5776415679220375,
+      "learning_rate": 0.003,
+      "loss": 4.2172,
+      "step": 1957
+    },
+    {
+      "epoch": 0.01958,
+      "grad_norm": 0.5010308008271186,
+      "learning_rate": 0.003,
+      "loss": 4.2108,
+      "step": 1958
+    },
+    {
+      "epoch": 0.01959,
+      "grad_norm": 0.500419502823399,
+      "learning_rate": 0.003,
+      "loss": 4.2342,
+      "step": 1959
+    },
+    {
+      "epoch": 0.0196,
+      "grad_norm": 0.5017850552651191,
+      "learning_rate": 0.003,
+      "loss": 4.229,
+      "step": 1960
+    },
+    {
+      "epoch": 0.01961,
+      "grad_norm": 0.49043553705552145,
+      "learning_rate": 0.003,
+      "loss": 4.2267,
+      "step": 1961
+    },
+    {
+      "epoch": 0.01962,
+      "grad_norm": 0.49806058358446914,
+      "learning_rate": 0.003,
+      "loss": 4.2301,
+      "step": 1962
+    },
+    {
+      "epoch": 0.01963,
+      "grad_norm": 0.5386284819858854,
+      "learning_rate": 0.003,
+      "loss": 4.2201,
+      "step": 1963
+    },
+    {
+      "epoch": 0.01964,
+      "grad_norm": 0.6462166130085853,
+      "learning_rate": 0.003,
+      "loss": 4.2492,
+      "step": 1964
+    },
+    {
+      "epoch": 0.01965,
+      "grad_norm": 0.8617584950578717,
+      "learning_rate": 0.003,
+      "loss": 4.2512,
+      "step": 1965
+    },
+    {
+      "epoch": 0.01966,
+      "grad_norm": 1.064420589029421,
+      "learning_rate": 0.003,
+      "loss": 4.2624,
+      "step": 1966
+    },
+    {
+      "epoch": 0.01967,
+      "grad_norm": 0.8064456905773513,
+      "learning_rate": 0.003,
+      "loss": 4.2378,
+      "step": 1967
+    },
+    {
+      "epoch": 0.01968,
+      "grad_norm": 0.6120027249682956,
+      "learning_rate": 0.003,
+      "loss": 4.2516,
+      "step": 1968
+    },
+    {
+      "epoch": 0.01969,
+      "grad_norm": 0.7508122971760963,
+      "learning_rate": 0.003,
+      "loss": 4.2526,
+      "step": 1969
+    },
+    {
+      "epoch": 0.0197,
+      "grad_norm": 0.7405449826366749,
+      "learning_rate": 0.003,
+      "loss": 4.2514,
+      "step": 1970
+    },
+    {
+      "epoch": 0.01971,
+      "grad_norm": 0.7056278637839583,
+      "learning_rate": 0.003,
+      "loss": 4.2377,
+      "step": 1971
+    },
+    {
+      "epoch": 0.01972,
+      "grad_norm": 0.7105683738164339,
+      "learning_rate": 0.003,
+      "loss": 4.2407,
+      "step": 1972
+    },
+    {
+      "epoch": 0.01973,
+      "grad_norm": 0.7508003521914575,
+      "learning_rate": 0.003,
+      "loss": 4.2264,
+      "step": 1973
+    },
+    {
+      "epoch": 0.01974,
+      "grad_norm": 0.7061088132196116,
+      "learning_rate": 0.003,
+      "loss": 4.2641,
+      "step": 1974
+    },
+    {
+      "epoch": 0.01975,
+      "grad_norm": 0.6140071978685933,
+      "learning_rate": 0.003,
+      "loss": 4.2539,
+      "step": 1975
+    },
+    {
+      "epoch": 0.01976,
+      "grad_norm": 0.6214170582947839,
+      "learning_rate": 0.003,
+      "loss": 4.243,
+      "step": 1976
+    },
+    {
+      "epoch": 0.01977,
+      "grad_norm": 0.6032091701282252,
+      "learning_rate": 0.003,
+      "loss": 4.2759,
+      "step": 1977
+    },
+    {
+      "epoch": 0.01978,
+      "grad_norm": 0.6890579485446906,
+      "learning_rate": 0.003,
+      "loss": 4.2782,
+      "step": 1978
+    },
+    {
+      "epoch": 0.01979,
+      "grad_norm": 0.7078287139411981,
+      "learning_rate": 0.003,
+      "loss": 4.269,
+      "step": 1979
+    },
+    {
+      "epoch": 0.0198,
+      "grad_norm": 0.726915367409026,
+      "learning_rate": 0.003,
+      "loss": 4.2503,
+      "step": 1980
+    },
+    {
+      "epoch": 0.01981,
+      "grad_norm": 0.7277727772925232,
+      "learning_rate": 0.003,
+      "loss": 4.2394,
+      "step": 1981
+    },
+    {
+      "epoch": 0.01982,
+      "grad_norm": 0.6618110470758236,
+      "learning_rate": 0.003,
+      "loss": 4.262,
+      "step": 1982
+    },
+    {
+      "epoch": 0.01983,
+      "grad_norm": 0.6282900232413579,
+      "learning_rate": 0.003,
+      "loss": 4.2497,
+      "step": 1983
+    },
+    {
+      "epoch": 0.01984,
+      "grad_norm": 0.6651788332059134,
+      "learning_rate": 0.003,
+      "loss": 4.2635,
+      "step": 1984
+    },
+    {
+      "epoch": 0.01985,
+      "grad_norm": 0.6195263920647736,
+      "learning_rate": 0.003,
+      "loss": 4.2484,
+      "step": 1985
+    },
+    {
+      "epoch": 0.01986,
+      "grad_norm": 0.6562857171469267,
+      "learning_rate": 0.003,
+      "loss": 4.2792,
+      "step": 1986
+    },
+    {
+      "epoch": 0.01987,
+      "grad_norm": 0.6274806569391715,
+      "learning_rate": 0.003,
+      "loss": 4.2331,
+      "step": 1987
+    },
+    {
+      "epoch": 0.01988,
+      "grad_norm": 0.6627977107268054,
+      "learning_rate": 0.003,
+      "loss": 4.2466,
+      "step": 1988
+    },
+    {
+      "epoch": 0.01989,
+      "grad_norm": 0.6286439849240877,
+      "learning_rate": 0.003,
+      "loss": 4.2344,
+      "step": 1989
+    },
+    {
+      "epoch": 0.0199,
+      "grad_norm": 0.5786457654364954,
+      "learning_rate": 0.003,
+      "loss": 4.246,
+      "step": 1990
+    },
+    {
+      "epoch": 0.01991,
+      "grad_norm": 0.5167049375276446,
+      "learning_rate": 0.003,
+      "loss": 4.2238,
+      "step": 1991
+    },
+    {
+      "epoch": 0.01992,
+      "grad_norm": 0.50421095170271,
+      "learning_rate": 0.003,
+      "loss": 4.261,
+      "step": 1992
+    },
+    {
+      "epoch": 0.01993,
+      "grad_norm": 0.4699320041008658,
+      "learning_rate": 0.003,
+      "loss": 4.2447,
+      "step": 1993
+    },
+    {
+      "epoch": 0.01994,
+      "grad_norm": 0.5186369446769569,
+      "learning_rate": 0.003,
+      "loss": 4.2629,
+      "step": 1994
+    },
+    {
+      "epoch": 0.01995,
+      "grad_norm": 0.5867398105885127,
+      "learning_rate": 0.003,
+      "loss": 4.2547,
+      "step": 1995
+    },
+    {
+      "epoch": 0.01996,
+      "grad_norm": 0.8031924386202848,
+      "learning_rate": 0.003,
+      "loss": 4.2326,
+      "step": 1996
+    },
+    {
+      "epoch": 0.01997,
+      "grad_norm": 0.9756949043315064,
+      "learning_rate": 0.003,
+      "loss": 4.2746,
+      "step": 1997
+    },
+    {
+      "epoch": 0.01998,
+      "grad_norm": 0.8340147112264455,
+      "learning_rate": 0.003,
+      "loss": 4.2546,
+      "step": 1998
+    },
+    {
+      "epoch": 0.01999,
+      "grad_norm": 0.6652469206537822,
+      "learning_rate": 0.003,
+      "loss": 4.2393,
+      "step": 1999
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7695645348180246,
+      "learning_rate": 0.003,
+      "loss": 4.2318,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 100000,
+  "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": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.9287285710848e+16,
+  "train_batch_size": 1024,
+  "trial_name": null,
+  "trial_params": null
+}