diff --git "a/checkpoint-1925/trainer_state.json" "b/checkpoint-1925/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1925/trainer_state.json"
@@ -0,0 +1,13508 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.01925,
+  "eval_steps": 500,
+  "global_step": 1925,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 1e-05,
+      "grad_norm": 1.7129806011276105,
+      "learning_rate": 3e-06,
+      "loss": 10.8348,
+      "step": 1
+    },
+    {
+      "epoch": 2e-05,
+      "grad_norm": 1.6872753303603527,
+      "learning_rate": 6e-06,
+      "loss": 10.8356,
+      "step": 2
+    },
+    {
+      "epoch": 3e-05,
+      "grad_norm": 1.6970020903903387,
+      "learning_rate": 9e-06,
+      "loss": 10.834,
+      "step": 3
+    },
+    {
+      "epoch": 4e-05,
+      "grad_norm": 1.690199421583159,
+      "learning_rate": 1.2e-05,
+      "loss": 10.8334,
+      "step": 4
+    },
+    {
+      "epoch": 5e-05,
+      "grad_norm": 1.6936208883930068,
+      "learning_rate": 1.5e-05,
+      "loss": 10.8294,
+      "step": 5
+    },
+    {
+      "epoch": 6e-05,
+      "grad_norm": 1.6935154610161474,
+      "learning_rate": 1.8e-05,
+      "loss": 10.8281,
+      "step": 6
+    },
+    {
+      "epoch": 7e-05,
+      "grad_norm": 1.6333694173725648,
+      "learning_rate": 2.1000000000000002e-05,
+      "loss": 10.8133,
+      "step": 7
+    },
+    {
+      "epoch": 8e-05,
+      "grad_norm": 1.4463755011186001,
+      "learning_rate": 2.4e-05,
+      "loss": 10.783,
+      "step": 8
+    },
+    {
+      "epoch": 9e-05,
+      "grad_norm": 1.3815123169360315,
+      "learning_rate": 2.7e-05,
+      "loss": 10.7779,
+      "step": 9
+    },
+    {
+      "epoch": 0.0001,
+      "grad_norm": 1.3507621465484316,
+      "learning_rate": 3e-05,
+      "loss": 10.7629,
+      "step": 10
+    },
+    {
+      "epoch": 0.00011,
+      "grad_norm": 1.257508561634155,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 10.7454,
+      "step": 11
+    },
+    {
+      "epoch": 0.00012,
+      "grad_norm": 1.224298046820689,
+      "learning_rate": 3.6e-05,
+      "loss": 10.7321,
+      "step": 12
+    },
+    {
+      "epoch": 0.00013,
+      "grad_norm": 1.1609107458726389,
+      "learning_rate": 3.9e-05,
+      "loss": 10.7098,
+      "step": 13
+    },
+    {
+      "epoch": 0.00014,
+      "grad_norm": 1.1251765756585856,
+      "learning_rate": 4.2000000000000004e-05,
+      "loss": 10.6986,
+      "step": 14
+    },
+    {
+      "epoch": 0.00015,
+      "grad_norm": 1.1021031797679595,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 10.6882,
+      "step": 15
+    },
+    {
+      "epoch": 0.00016,
+      "grad_norm": 1.05231707077907,
+      "learning_rate": 4.8e-05,
+      "loss": 10.6681,
+      "step": 16
+    },
+    {
+      "epoch": 0.00017,
+      "grad_norm": 1.0082613504118885,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 10.6513,
+      "step": 17
+    },
+    {
+      "epoch": 0.00018,
+      "grad_norm": 0.9840324393168475,
+      "learning_rate": 5.4e-05,
+      "loss": 10.6344,
+      "step": 18
+    },
+    {
+      "epoch": 0.00019,
+      "grad_norm": 0.953923239589578,
+      "learning_rate": 5.7e-05,
+      "loss": 10.6196,
+      "step": 19
+    },
+    {
+      "epoch": 0.0002,
+      "grad_norm": 0.9458057853193742,
+      "learning_rate": 6e-05,
+      "loss": 10.6069,
+      "step": 20
+    },
+    {
+      "epoch": 0.00021,
+      "grad_norm": 0.9177457999897578,
+      "learning_rate": 6.3e-05,
+      "loss": 10.5922,
+      "step": 21
+    },
+    {
+      "epoch": 0.00022,
+      "grad_norm": 0.9134874433162486,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 10.576,
+      "step": 22
+    },
+    {
+      "epoch": 0.00023,
+      "grad_norm": 0.908696989628468,
+      "learning_rate": 6.9e-05,
+      "loss": 10.5635,
+      "step": 23
+    },
+    {
+      "epoch": 0.00024,
+      "grad_norm": 0.9051143007426985,
+      "learning_rate": 7.2e-05,
+      "loss": 10.5499,
+      "step": 24
+    },
+    {
+      "epoch": 0.00025,
+      "grad_norm": 0.9082451576693834,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 10.5361,
+      "step": 25
+    },
+    {
+      "epoch": 0.00026,
+      "grad_norm": 0.9099344486243927,
+      "learning_rate": 7.8e-05,
+      "loss": 10.521,
+      "step": 26
+    },
+    {
+      "epoch": 0.00027,
+      "grad_norm": 0.9053293550746107,
+      "learning_rate": 8.1e-05,
+      "loss": 10.5103,
+      "step": 27
+    },
+    {
+      "epoch": 0.00028,
+      "grad_norm": 0.9002471153364864,
+      "learning_rate": 8.400000000000001e-05,
+      "loss": 10.4955,
+      "step": 28
+    },
+    {
+      "epoch": 0.00029,
+      "grad_norm": 0.9068699186733776,
+      "learning_rate": 8.7e-05,
+      "loss": 10.4811,
+      "step": 29
+    },
+    {
+      "epoch": 0.0003,
+      "grad_norm": 0.9095271030063902,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 10.4648,
+      "step": 30
+    },
+    {
+      "epoch": 0.00031,
+      "grad_norm": 0.9097010936405139,
+      "learning_rate": 9.3e-05,
+      "loss": 10.4503,
+      "step": 31
+    },
+    {
+      "epoch": 0.00032,
+      "grad_norm": 0.9047462396891427,
+      "learning_rate": 9.6e-05,
+      "loss": 10.4348,
+      "step": 32
+    },
+    {
+      "epoch": 0.00033,
+      "grad_norm": 0.9068703333942145,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 10.418,
+      "step": 33
+    },
+    {
+      "epoch": 0.00034,
+      "grad_norm": 0.9072965837486595,
+      "learning_rate": 0.00010200000000000001,
+      "loss": 10.4,
+      "step": 34
+    },
+    {
+      "epoch": 0.00035,
+      "grad_norm": 0.9110003633890357,
+      "learning_rate": 0.00010500000000000002,
+      "loss": 10.3835,
+      "step": 35
+    },
+    {
+      "epoch": 0.00036,
+      "grad_norm": 0.9049119959927198,
+      "learning_rate": 0.000108,
+      "loss": 10.3652,
+      "step": 36
+    },
+    {
+      "epoch": 0.00037,
+      "grad_norm": 0.8970709544624084,
+      "learning_rate": 0.000111,
+      "loss": 10.3479,
+      "step": 37
+    },
+    {
+      "epoch": 0.00038,
+      "grad_norm": 0.8959068278842482,
+      "learning_rate": 0.000114,
+      "loss": 10.3275,
+      "step": 38
+    },
+    {
+      "epoch": 0.00039,
+      "grad_norm": 0.9005947927478184,
+      "learning_rate": 0.000117,
+      "loss": 10.3069,
+      "step": 39
+    },
+    {
+      "epoch": 0.0004,
+      "grad_norm": 0.9014442598894896,
+      "learning_rate": 0.00012,
+      "loss": 10.2842,
+      "step": 40
+    },
+    {
+      "epoch": 0.00041,
+      "grad_norm": 0.8992939718171602,
+      "learning_rate": 0.000123,
+      "loss": 10.2657,
+      "step": 41
+    },
+    {
+      "epoch": 0.00042,
+      "grad_norm": 0.8994818536906172,
+      "learning_rate": 0.000126,
+      "loss": 10.2444,
+      "step": 42
+    },
+    {
+      "epoch": 0.00043,
+      "grad_norm": 0.9062946670458473,
+      "learning_rate": 0.000129,
+      "loss": 10.2208,
+      "step": 43
+    },
+    {
+      "epoch": 0.00044,
+      "grad_norm": 0.9072550424345267,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 10.1985,
+      "step": 44
+    },
+    {
+      "epoch": 0.00045,
+      "grad_norm": 0.908308760029939,
+      "learning_rate": 0.000135,
+      "loss": 10.1758,
+      "step": 45
+    },
+    {
+      "epoch": 0.00046,
+      "grad_norm": 0.8994605508976834,
+      "learning_rate": 0.000138,
+      "loss": 10.1528,
+      "step": 46
+    },
+    {
+      "epoch": 0.00047,
+      "grad_norm": 0.904955141258698,
+      "learning_rate": 0.000141,
+      "loss": 10.1274,
+      "step": 47
+    },
+    {
+      "epoch": 0.00048,
+      "grad_norm": 0.9044693581157806,
+      "learning_rate": 0.000144,
+      "loss": 10.1031,
+      "step": 48
+    },
+    {
+      "epoch": 0.00049,
+      "grad_norm": 0.8992120995192336,
+      "learning_rate": 0.000147,
+      "loss": 10.0777,
+      "step": 49
+    },
+    {
+      "epoch": 0.0005,
+      "grad_norm": 0.905676588399281,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 10.0519,
+      "step": 50
+    },
+    {
+      "epoch": 0.00051,
+      "grad_norm": 0.9066841497261428,
+      "learning_rate": 0.000153,
+      "loss": 10.0251,
+      "step": 51
+    },
+    {
+      "epoch": 0.00052,
+      "grad_norm": 0.9046656683417261,
+      "learning_rate": 0.000156,
+      "loss": 9.9981,
+      "step": 52
+    },
+    {
+      "epoch": 0.00053,
+      "grad_norm": 0.8943714853313668,
+      "learning_rate": 0.000159,
+      "loss": 9.974,
+      "step": 53
+    },
+    {
+      "epoch": 0.00054,
+      "grad_norm": 0.9141658233846578,
+      "learning_rate": 0.000162,
+      "loss": 9.9419,
+      "step": 54
+    },
+    {
+      "epoch": 0.00055,
+      "grad_norm": 0.9035944774643171,
+      "learning_rate": 0.000165,
+      "loss": 9.9169,
+      "step": 55
+    },
+    {
+      "epoch": 0.00056,
+      "grad_norm": 0.895407870582166,
+      "learning_rate": 0.00016800000000000002,
+      "loss": 9.8872,
+      "step": 56
+    },
+    {
+      "epoch": 0.00057,
+      "grad_norm": 0.9021731997760362,
+      "learning_rate": 0.000171,
+      "loss": 9.8601,
+      "step": 57
+    },
+    {
+      "epoch": 0.00058,
+      "grad_norm": 0.8980871554912008,
+      "learning_rate": 0.000174,
+      "loss": 9.8343,
+      "step": 58
+    },
+    {
+      "epoch": 0.00059,
+      "grad_norm": 0.9073832283363998,
+      "learning_rate": 0.000177,
+      "loss": 9.8028,
+      "step": 59
+    },
+    {
+      "epoch": 0.0006,
+      "grad_norm": 0.8921071838486323,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 9.777,
+      "step": 60
+    },
+    {
+      "epoch": 0.00061,
+      "grad_norm": 0.8918001641348363,
+      "learning_rate": 0.000183,
+      "loss": 9.7484,
+      "step": 61
+    },
+    {
+      "epoch": 0.00062,
+      "grad_norm": 0.897401330332219,
+      "learning_rate": 0.000186,
+      "loss": 9.717,
+      "step": 62
+    },
+    {
+      "epoch": 0.00063,
+      "grad_norm": 0.8914516241190131,
+      "learning_rate": 0.000189,
+      "loss": 9.6894,
+      "step": 63
+    },
+    {
+      "epoch": 0.00064,
+      "grad_norm": 0.8896652156254973,
+      "learning_rate": 0.000192,
+      "loss": 9.6587,
+      "step": 64
+    },
+    {
+      "epoch": 0.00065,
+      "grad_norm": 0.8995447585153489,
+      "learning_rate": 0.00019500000000000002,
+      "loss": 9.6261,
+      "step": 65
+    },
+    {
+      "epoch": 0.00066,
+      "grad_norm": 0.8896939041293862,
+      "learning_rate": 0.00019800000000000002,
+      "loss": 9.6026,
+      "step": 66
+    },
+    {
+      "epoch": 0.00067,
+      "grad_norm": 0.8935314234316469,
+      "learning_rate": 0.000201,
+      "loss": 9.5723,
+      "step": 67
+    },
+    {
+      "epoch": 0.00068,
+      "grad_norm": 0.8971584182008717,
+      "learning_rate": 0.00020400000000000003,
+      "loss": 9.5393,
+      "step": 68
+    },
+    {
+      "epoch": 0.00069,
+      "grad_norm": 0.8806455604370673,
+      "learning_rate": 0.00020700000000000002,
+      "loss": 9.5119,
+      "step": 69
+    },
+    {
+      "epoch": 0.0007,
+      "grad_norm": 0.892956094531968,
+      "learning_rate": 0.00021000000000000004,
+      "loss": 9.4751,
+      "step": 70
+    },
+    {
+      "epoch": 0.00071,
+      "grad_norm": 0.8848452972865632,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 9.4495,
+      "step": 71
+    },
+    {
+      "epoch": 0.00072,
+      "grad_norm": 0.8831148731992822,
+      "learning_rate": 0.000216,
+      "loss": 9.4223,
+      "step": 72
+    },
+    {
+      "epoch": 0.00073,
+      "grad_norm": 0.887150899449638,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 9.3879,
+      "step": 73
+    },
+    {
+      "epoch": 0.00074,
+      "grad_norm": 0.8878619769377328,
+      "learning_rate": 0.000222,
+      "loss": 9.3616,
+      "step": 74
+    },
+    {
+      "epoch": 0.00075,
+      "grad_norm": 0.8808154408936898,
+      "learning_rate": 0.000225,
+      "loss": 9.3275,
+      "step": 75
+    },
+    {
+      "epoch": 0.00076,
+      "grad_norm": 0.8908035269749474,
+      "learning_rate": 0.000228,
+      "loss": 9.3009,
+      "step": 76
+    },
+    {
+      "epoch": 0.00077,
+      "grad_norm": 0.884531047332737,
+      "learning_rate": 0.000231,
+      "loss": 9.2727,
+      "step": 77
+    },
+    {
+      "epoch": 0.00078,
+      "grad_norm": 0.8838664917591654,
+      "learning_rate": 0.000234,
+      "loss": 9.2422,
+      "step": 78
+    },
+    {
+      "epoch": 0.00079,
+      "grad_norm": 0.8858668201182466,
+      "learning_rate": 0.00023700000000000001,
+      "loss": 9.2056,
+      "step": 79
+    },
+    {
+      "epoch": 0.0008,
+      "grad_norm": 0.8856967305037482,
+      "learning_rate": 0.00024,
+      "loss": 9.1711,
+      "step": 80
+    },
+    {
+      "epoch": 0.00081,
+      "grad_norm": 0.8942846826675519,
+      "learning_rate": 0.00024300000000000002,
+      "loss": 9.1382,
+      "step": 81
+    },
+    {
+      "epoch": 0.00082,
+      "grad_norm": 0.897767651472895,
+      "learning_rate": 0.000246,
+      "loss": 9.1142,
+      "step": 82
+    },
+    {
+      "epoch": 0.00083,
+      "grad_norm": 0.8951752702012633,
+      "learning_rate": 0.00024900000000000004,
+      "loss": 9.0778,
+      "step": 83
+    },
+    {
+      "epoch": 0.00084,
+      "grad_norm": 0.8980395289622467,
+      "learning_rate": 0.000252,
+      "loss": 9.0469,
+      "step": 84
+    },
+    {
+      "epoch": 0.00085,
+      "grad_norm": 0.8894006576183595,
+      "learning_rate": 0.000255,
+      "loss": 9.0242,
+      "step": 85
+    },
+    {
+      "epoch": 0.00086,
+      "grad_norm": 0.8907945566480024,
+      "learning_rate": 0.000258,
+      "loss": 8.9886,
+      "step": 86
+    },
+    {
+      "epoch": 0.00087,
+      "grad_norm": 0.8869170795764568,
+      "learning_rate": 0.000261,
+      "loss": 8.9664,
+      "step": 87
+    },
+    {
+      "epoch": 0.00088,
+      "grad_norm": 0.8860541210154026,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 8.9293,
+      "step": 88
+    },
+    {
+      "epoch": 0.00089,
+      "grad_norm": 0.8822605600899943,
+      "learning_rate": 0.000267,
+      "loss": 8.9037,
+      "step": 89
+    },
+    {
+      "epoch": 0.0009,
+      "grad_norm": 0.8817151929172502,
+      "learning_rate": 0.00027,
+      "loss": 8.8766,
+      "step": 90
+    },
+    {
+      "epoch": 0.00091,
+      "grad_norm": 0.877617615465877,
+      "learning_rate": 0.000273,
+      "loss": 8.8478,
+      "step": 91
+    },
+    {
+      "epoch": 0.00092,
+      "grad_norm": 0.8822716293479064,
+      "learning_rate": 0.000276,
+      "loss": 8.8156,
+      "step": 92
+    },
+    {
+      "epoch": 0.00093,
+      "grad_norm": 0.8823661552266111,
+      "learning_rate": 0.000279,
+      "loss": 8.7863,
+      "step": 93
+    },
+    {
+      "epoch": 0.00094,
+      "grad_norm": 0.8830384482321385,
+      "learning_rate": 0.000282,
+      "loss": 8.7609,
+      "step": 94
+    },
+    {
+      "epoch": 0.00095,
+      "grad_norm": 0.8735042737334501,
+      "learning_rate": 0.000285,
+      "loss": 8.7321,
+      "step": 95
+    },
+    {
+      "epoch": 0.00096,
+      "grad_norm": 0.8799214796836804,
+      "learning_rate": 0.000288,
+      "loss": 8.7028,
+      "step": 96
+    },
+    {
+      "epoch": 0.00097,
+      "grad_norm": 0.8704594748643596,
+      "learning_rate": 0.000291,
+      "loss": 8.6791,
+      "step": 97
+    },
+    {
+      "epoch": 0.00098,
+      "grad_norm": 0.8706415983834461,
+      "learning_rate": 0.000294,
+      "loss": 8.642,
+      "step": 98
+    },
+    {
+      "epoch": 0.00099,
+      "grad_norm": 0.8683426041650804,
+      "learning_rate": 0.000297,
+      "loss": 8.62,
+      "step": 99
+    },
+    {
+      "epoch": 0.001,
+      "grad_norm": 0.8690594926543161,
+      "learning_rate": 0.00030000000000000003,
+      "loss": 8.5941,
+      "step": 100
+    },
+    {
+      "epoch": 0.00101,
+      "grad_norm": 0.8744725161423202,
+      "learning_rate": 0.00030300000000000005,
+      "loss": 8.5597,
+      "step": 101
+    },
+    {
+      "epoch": 0.00102,
+      "grad_norm": 0.8626538117869429,
+      "learning_rate": 0.000306,
+      "loss": 8.5407,
+      "step": 102
+    },
+    {
+      "epoch": 0.00103,
+      "grad_norm": 0.8630292491448714,
+      "learning_rate": 0.000309,
+      "loss": 8.5165,
+      "step": 103
+    },
+    {
+      "epoch": 0.00104,
+      "grad_norm": 0.8566580756350954,
+      "learning_rate": 0.000312,
+      "loss": 8.4948,
+      "step": 104
+    },
+    {
+      "epoch": 0.00105,
+      "grad_norm": 0.8588931967033124,
+      "learning_rate": 0.000315,
+      "loss": 8.4689,
+      "step": 105
+    },
+    {
+      "epoch": 0.00106,
+      "grad_norm": 0.8531867230327145,
+      "learning_rate": 0.000318,
+      "loss": 8.4397,
+      "step": 106
+    },
+    {
+      "epoch": 0.00107,
+      "grad_norm": 0.8474808010000593,
+      "learning_rate": 0.000321,
+      "loss": 8.4149,
+      "step": 107
+    },
+    {
+      "epoch": 0.00108,
+      "grad_norm": 0.858890949163445,
+      "learning_rate": 0.000324,
+      "loss": 8.3866,
+      "step": 108
+    },
+    {
+      "epoch": 0.00109,
+      "grad_norm": 0.862504115159085,
+      "learning_rate": 0.000327,
+      "loss": 8.3673,
+      "step": 109
+    },
+    {
+      "epoch": 0.0011,
+      "grad_norm": 0.8797254817902618,
+      "learning_rate": 0.00033,
+      "loss": 8.3453,
+      "step": 110
+    },
+    {
+      "epoch": 0.00111,
+      "grad_norm": 0.8938450574121157,
+      "learning_rate": 0.000333,
+      "loss": 8.3162,
+      "step": 111
+    },
+    {
+      "epoch": 0.00112,
+      "grad_norm": 0.8984693362159062,
+      "learning_rate": 0.00033600000000000004,
+      "loss": 8.2961,
+      "step": 112
+    },
+    {
+      "epoch": 0.00113,
+      "grad_norm": 0.8741969532880793,
+      "learning_rate": 0.000339,
+      "loss": 8.2543,
+      "step": 113
+    },
+    {
+      "epoch": 0.00114,
+      "grad_norm": 0.8263135137442741,
+      "learning_rate": 0.000342,
+      "loss": 8.2446,
+      "step": 114
+    },
+    {
+      "epoch": 0.00115,
+      "grad_norm": 0.8311105019716521,
+      "learning_rate": 0.00034500000000000004,
+      "loss": 8.2164,
+      "step": 115
+    },
+    {
+      "epoch": 0.00116,
+      "grad_norm": 0.8585271561560018,
+      "learning_rate": 0.000348,
+      "loss": 8.1938,
+      "step": 116
+    },
+    {
+      "epoch": 0.00117,
+      "grad_norm": 0.8687047969468357,
+      "learning_rate": 0.000351,
+      "loss": 8.1623,
+      "step": 117
+    },
+    {
+      "epoch": 0.00118,
+      "grad_norm": 0.8460500876754268,
+      "learning_rate": 0.000354,
+      "loss": 8.1456,
+      "step": 118
+    },
+    {
+      "epoch": 0.00119,
+      "grad_norm": 0.80734714043103,
+      "learning_rate": 0.000357,
+      "loss": 8.131,
+      "step": 119
+    },
+    {
+      "epoch": 0.0012,
+      "grad_norm": 0.7912470130916918,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 8.103,
+      "step": 120
+    },
+    {
+      "epoch": 0.00121,
+      "grad_norm": 0.811181199244069,
+      "learning_rate": 0.000363,
+      "loss": 8.0751,
+      "step": 121
+    },
+    {
+      "epoch": 0.00122,
+      "grad_norm": 0.8641427784894331,
+      "learning_rate": 0.000366,
+      "loss": 8.0581,
+      "step": 122
+    },
+    {
+      "epoch": 0.00123,
+      "grad_norm": 1.059706847686213,
+      "learning_rate": 0.000369,
+      "loss": 8.038,
+      "step": 123
+    },
+    {
+      "epoch": 0.00124,
+      "grad_norm": 1.1727027216994725,
+      "learning_rate": 0.000372,
+      "loss": 8.021,
+      "step": 124
+    },
+    {
+      "epoch": 0.00125,
+      "grad_norm": 0.8130002892684417,
+      "learning_rate": 0.000375,
+      "loss": 7.9874,
+      "step": 125
+    },
+    {
+      "epoch": 0.00126,
+      "grad_norm": 0.9195848585380069,
+      "learning_rate": 0.000378,
+      "loss": 7.9767,
+      "step": 126
+    },
+    {
+      "epoch": 0.00127,
+      "grad_norm": 1.0843433185909894,
+      "learning_rate": 0.000381,
+      "loss": 7.9572,
+      "step": 127
+    },
+    {
+      "epoch": 0.00128,
+      "grad_norm": 0.7822919696307823,
+      "learning_rate": 0.000384,
+      "loss": 7.9285,
+      "step": 128
+    },
+    {
+      "epoch": 0.00129,
+      "grad_norm": 0.8822428605457112,
+      "learning_rate": 0.00038700000000000003,
+      "loss": 7.9179,
+      "step": 129
+    },
+    {
+      "epoch": 0.0013,
+      "grad_norm": 0.844355538937723,
+      "learning_rate": 0.00039000000000000005,
+      "loss": 7.8895,
+      "step": 130
+    },
+    {
+      "epoch": 0.00131,
+      "grad_norm": 0.7902535444057679,
+      "learning_rate": 0.000393,
+      "loss": 7.8528,
+      "step": 131
+    },
+    {
+      "epoch": 0.00132,
+      "grad_norm": 0.8777082094723105,
+      "learning_rate": 0.00039600000000000003,
+      "loss": 7.8441,
+      "step": 132
+    },
+    {
+      "epoch": 0.00133,
+      "grad_norm": 0.7973277807473979,
+      "learning_rate": 0.00039900000000000005,
+      "loss": 7.8195,
+      "step": 133
+    },
+    {
+      "epoch": 0.00134,
+      "grad_norm": 0.7889088832890946,
+      "learning_rate": 0.000402,
+      "loss": 7.8056,
+      "step": 134
+    },
+    {
+      "epoch": 0.00135,
+      "grad_norm": 0.7461125825498439,
+      "learning_rate": 0.00040500000000000003,
+      "loss": 7.7735,
+      "step": 135
+    },
+    {
+      "epoch": 0.00136,
+      "grad_norm": 0.7727896835908762,
+      "learning_rate": 0.00040800000000000005,
+      "loss": 7.7579,
+      "step": 136
+    },
+    {
+      "epoch": 0.00137,
+      "grad_norm": 0.6932995987295251,
+      "learning_rate": 0.000411,
+      "loss": 7.7341,
+      "step": 137
+    },
+    {
+      "epoch": 0.00138,
+      "grad_norm": 0.758084762416224,
+      "learning_rate": 0.00041400000000000003,
+      "loss": 7.7117,
+      "step": 138
+    },
+    {
+      "epoch": 0.00139,
+      "grad_norm": 0.7171019453691133,
+      "learning_rate": 0.00041700000000000005,
+      "loss": 7.6963,
+      "step": 139
+    },
+    {
+      "epoch": 0.0014,
+      "grad_norm": 0.6814920611933867,
+      "learning_rate": 0.00042000000000000007,
+      "loss": 7.6775,
+      "step": 140
+    },
+    {
+      "epoch": 0.00141,
+      "grad_norm": 0.7091532995122851,
+      "learning_rate": 0.000423,
+      "loss": 7.6638,
+      "step": 141
+    },
+    {
+      "epoch": 0.00142,
+      "grad_norm": 0.6928279523561562,
+      "learning_rate": 0.00042599999999999995,
+      "loss": 7.6501,
+      "step": 142
+    },
+    {
+      "epoch": 0.00143,
+      "grad_norm": 0.6614572727332786,
+      "learning_rate": 0.00042899999999999997,
+      "loss": 7.6195,
+      "step": 143
+    },
+    {
+      "epoch": 0.00144,
+      "grad_norm": 0.6903462553659518,
+      "learning_rate": 0.000432,
+      "loss": 7.6015,
+      "step": 144
+    },
+    {
+      "epoch": 0.00145,
+      "grad_norm": 0.690019772183536,
+      "learning_rate": 0.000435,
+      "loss": 7.5953,
+      "step": 145
+    },
+    {
+      "epoch": 0.00146,
+      "grad_norm": 0.6908198257220046,
+      "learning_rate": 0.00043799999999999997,
+      "loss": 7.5557,
+      "step": 146
+    },
+    {
+      "epoch": 0.00147,
+      "grad_norm": 0.7009866965495668,
+      "learning_rate": 0.000441,
+      "loss": 7.5482,
+      "step": 147
+    },
+    {
+      "epoch": 0.00148,
+      "grad_norm": 0.6832764187147686,
+      "learning_rate": 0.000444,
+      "loss": 7.5366,
+      "step": 148
+    },
+    {
+      "epoch": 0.00149,
+      "grad_norm": 0.59797192318343,
+      "learning_rate": 0.00044699999999999997,
+      "loss": 7.5272,
+      "step": 149
+    },
+    {
+      "epoch": 0.0015,
+      "grad_norm": 0.6655702435683013,
+      "learning_rate": 0.00045,
+      "loss": 7.4963,
+      "step": 150
+    },
+    {
+      "epoch": 0.00151,
+      "grad_norm": 0.732396941583091,
+      "learning_rate": 0.000453,
+      "loss": 7.48,
+      "step": 151
+    },
+    {
+      "epoch": 0.00152,
+      "grad_norm": 0.5836278900992692,
+      "learning_rate": 0.000456,
+      "loss": 7.4694,
+      "step": 152
+    },
+    {
+      "epoch": 0.00153,
+      "grad_norm": 0.6777912087785298,
+      "learning_rate": 0.000459,
+      "loss": 7.4588,
+      "step": 153
+    },
+    {
+      "epoch": 0.00154,
+      "grad_norm": 0.727978180952039,
+      "learning_rate": 0.000462,
+      "loss": 7.442,
+      "step": 154
+    },
+    {
+      "epoch": 0.00155,
+      "grad_norm": 0.7368922682622268,
+      "learning_rate": 0.000465,
+      "loss": 7.4241,
+      "step": 155
+    },
+    {
+      "epoch": 0.00156,
+      "grad_norm": 0.8391325557731037,
+      "learning_rate": 0.000468,
+      "loss": 7.4013,
+      "step": 156
+    },
+    {
+      "epoch": 0.00157,
+      "grad_norm": 0.8289929528374833,
+      "learning_rate": 0.000471,
+      "loss": 7.3995,
+      "step": 157
+    },
+    {
+      "epoch": 0.00158,
+      "grad_norm": 0.5070337070851558,
+      "learning_rate": 0.00047400000000000003,
+      "loss": 7.3713,
+      "step": 158
+    },
+    {
+      "epoch": 0.00159,
+      "grad_norm": 0.783946493417518,
+      "learning_rate": 0.000477,
+      "loss": 7.3668,
+      "step": 159
+    },
+    {
+      "epoch": 0.0016,
+      "grad_norm": 0.6957053326984224,
+      "learning_rate": 0.00048,
+      "loss": 7.3475,
+      "step": 160
+    },
+    {
+      "epoch": 0.00161,
+      "grad_norm": 0.547833885334286,
+      "learning_rate": 0.00048300000000000003,
+      "loss": 7.3204,
+      "step": 161
+    },
+    {
+      "epoch": 0.00162,
+      "grad_norm": 0.8547649122041628,
+      "learning_rate": 0.00048600000000000005,
+      "loss": 7.3325,
+      "step": 162
+    },
+    {
+      "epoch": 0.00163,
+      "grad_norm": 0.8673949773728752,
+      "learning_rate": 0.0004890000000000001,
+      "loss": 7.316,
+      "step": 163
+    },
+    {
+      "epoch": 0.00164,
+      "grad_norm": 1.0619539506126108,
+      "learning_rate": 0.000492,
+      "loss": 7.3191,
+      "step": 164
+    },
+    {
+      "epoch": 0.00165,
+      "grad_norm": 0.6324744711420325,
+      "learning_rate": 0.000495,
+      "loss": 7.2864,
+      "step": 165
+    },
+    {
+      "epoch": 0.00166,
+      "grad_norm": 0.5662260261966024,
+      "learning_rate": 0.0004980000000000001,
+      "loss": 7.26,
+      "step": 166
+    },
+    {
+      "epoch": 0.00167,
+      "grad_norm": 0.7262900850309921,
+      "learning_rate": 0.000501,
+      "loss": 7.2554,
+      "step": 167
+    },
+    {
+      "epoch": 0.00168,
+      "grad_norm": 0.6121691436587496,
+      "learning_rate": 0.000504,
+      "loss": 7.2353,
+      "step": 168
+    },
+    {
+      "epoch": 0.00169,
+      "grad_norm": 0.5390794603769147,
+      "learning_rate": 0.0005070000000000001,
+      "loss": 7.2263,
+      "step": 169
+    },
+    {
+      "epoch": 0.0017,
+      "grad_norm": 0.5999036994585554,
+      "learning_rate": 0.00051,
+      "loss": 7.213,
+      "step": 170
+    },
+    {
+      "epoch": 0.00171,
+      "grad_norm": 0.4637320512013434,
+      "learning_rate": 0.000513,
+      "loss": 7.1933,
+      "step": 171
+    },
+    {
+      "epoch": 0.00172,
+      "grad_norm": 0.5250975302523401,
+      "learning_rate": 0.000516,
+      "loss": 7.1953,
+      "step": 172
+    },
+    {
+      "epoch": 0.00173,
+      "grad_norm": 0.40559164125903624,
+      "learning_rate": 0.0005189999999999999,
+      "loss": 7.1764,
+      "step": 173
+    },
+    {
+      "epoch": 0.00174,
+      "grad_norm": 0.4505921111310584,
+      "learning_rate": 0.000522,
+      "loss": 7.1953,
+      "step": 174
+    },
+    {
+      "epoch": 0.00175,
+      "grad_norm": 0.4234331150208657,
+      "learning_rate": 0.000525,
+      "loss": 7.1572,
+      "step": 175
+    },
+    {
+      "epoch": 0.00176,
+      "grad_norm": 0.3852967422981744,
+      "learning_rate": 0.0005279999999999999,
+      "loss": 7.1322,
+      "step": 176
+    },
+    {
+      "epoch": 0.00177,
+      "grad_norm": 0.3685443025565043,
+      "learning_rate": 0.000531,
+      "loss": 7.1378,
+      "step": 177
+    },
+    {
+      "epoch": 0.00178,
+      "grad_norm": 0.44280593644992733,
+      "learning_rate": 0.000534,
+      "loss": 7.1301,
+      "step": 178
+    },
+    {
+      "epoch": 0.00179,
+      "grad_norm": 0.3638226120256115,
+      "learning_rate": 0.000537,
+      "loss": 7.1191,
+      "step": 179
+    },
+    {
+      "epoch": 0.0018,
+      "grad_norm": 0.37841703582661185,
+      "learning_rate": 0.00054,
+      "loss": 7.0921,
+      "step": 180
+    },
+    {
+      "epoch": 0.00181,
+      "grad_norm": 0.3275750999054276,
+      "learning_rate": 0.000543,
+      "loss": 7.0801,
+      "step": 181
+    },
+    {
+      "epoch": 0.00182,
+      "grad_norm": 0.3469517461565544,
+      "learning_rate": 0.000546,
+      "loss": 7.0774,
+      "step": 182
+    },
+    {
+      "epoch": 0.00183,
+      "grad_norm": 0.3965623823212328,
+      "learning_rate": 0.000549,
+      "loss": 7.0674,
+      "step": 183
+    },
+    {
+      "epoch": 0.00184,
+      "grad_norm": 0.47244712960577356,
+      "learning_rate": 0.000552,
+      "loss": 7.0582,
+      "step": 184
+    },
+    {
+      "epoch": 0.00185,
+      "grad_norm": 0.7068086356604425,
+      "learning_rate": 0.000555,
+      "loss": 7.0369,
+      "step": 185
+    },
+    {
+      "epoch": 0.00186,
+      "grad_norm": 0.9665650200874053,
+      "learning_rate": 0.000558,
+      "loss": 7.0604,
+      "step": 186
+    },
+    {
+      "epoch": 0.00187,
+      "grad_norm": 1.1379887499575514,
+      "learning_rate": 0.000561,
+      "loss": 7.0366,
+      "step": 187
+    },
+    {
+      "epoch": 0.00188,
+      "grad_norm": 0.5005933831438132,
+      "learning_rate": 0.000564,
+      "loss": 7.0008,
+      "step": 188
+    },
+    {
+      "epoch": 0.00189,
+      "grad_norm": 0.4490325126563288,
+      "learning_rate": 0.000567,
+      "loss": 6.997,
+      "step": 189
+    },
+    {
+      "epoch": 0.0019,
+      "grad_norm": 0.6949112483193859,
+      "learning_rate": 0.00057,
+      "loss": 6.9846,
+      "step": 190
+    },
+    {
+      "epoch": 0.00191,
+      "grad_norm": 0.4887612962658467,
+      "learning_rate": 0.000573,
+      "loss": 6.9724,
+      "step": 191
+    },
+    {
+      "epoch": 0.00192,
+      "grad_norm": 0.5374763602633008,
+      "learning_rate": 0.000576,
+      "loss": 6.9655,
+      "step": 192
+    },
+    {
+      "epoch": 0.00193,
+      "grad_norm": 0.4491815623326969,
+      "learning_rate": 0.000579,
+      "loss": 6.9637,
+      "step": 193
+    },
+    {
+      "epoch": 0.00194,
+      "grad_norm": 0.4044031823800156,
+      "learning_rate": 0.000582,
+      "loss": 6.9565,
+      "step": 194
+    },
+    {
+      "epoch": 0.00195,
+      "grad_norm": 0.5115147380417242,
+      "learning_rate": 0.000585,
+      "loss": 6.9386,
+      "step": 195
+    },
+    {
+      "epoch": 0.00196,
+      "grad_norm": 0.45947827433809557,
+      "learning_rate": 0.000588,
+      "loss": 6.9258,
+      "step": 196
+    },
+    {
+      "epoch": 0.00197,
+      "grad_norm": 0.5289316721591154,
+      "learning_rate": 0.000591,
+      "loss": 6.9226,
+      "step": 197
+    },
+    {
+      "epoch": 0.00198,
+      "grad_norm": 0.4416511613975406,
+      "learning_rate": 0.000594,
+      "loss": 6.9132,
+      "step": 198
+    },
+    {
+      "epoch": 0.00199,
+      "grad_norm": 0.36314866008934127,
+      "learning_rate": 0.0005970000000000001,
+      "loss": 6.8916,
+      "step": 199
+    },
+    {
+      "epoch": 0.002,
+      "grad_norm": 0.4299454881914127,
+      "learning_rate": 0.0006000000000000001,
+      "loss": 6.8932,
+      "step": 200
+    },
+    {
+      "epoch": 0.00201,
+      "grad_norm": 0.2786890859011363,
+      "learning_rate": 0.000603,
+      "loss": 6.8645,
+      "step": 201
+    },
+    {
+      "epoch": 0.00202,
+      "grad_norm": 0.4553990879307791,
+      "learning_rate": 0.0006060000000000001,
+      "loss": 6.855,
+      "step": 202
+    },
+    {
+      "epoch": 0.00203,
+      "grad_norm": 0.49491513980041124,
+      "learning_rate": 0.0006090000000000001,
+      "loss": 6.8592,
+      "step": 203
+    },
+    {
+      "epoch": 0.00204,
+      "grad_norm": 0.5750090076580947,
+      "learning_rate": 0.000612,
+      "loss": 6.8457,
+      "step": 204
+    },
+    {
+      "epoch": 0.00205,
+      "grad_norm": 0.6904749130436038,
+      "learning_rate": 0.000615,
+      "loss": 6.8381,
+      "step": 205
+    },
+    {
+      "epoch": 0.00206,
+      "grad_norm": 0.7582947936777445,
+      "learning_rate": 0.000618,
+      "loss": 6.83,
+      "step": 206
+    },
+    {
+      "epoch": 0.00207,
+      "grad_norm": 0.728748472942146,
+      "learning_rate": 0.000621,
+      "loss": 6.8214,
+      "step": 207
+    },
+    {
+      "epoch": 0.00208,
+      "grad_norm": 0.5163586812963157,
+      "learning_rate": 0.000624,
+      "loss": 6.8116,
+      "step": 208
+    },
+    {
+      "epoch": 0.00209,
+      "grad_norm": 0.5726761174567752,
+      "learning_rate": 0.000627,
+      "loss": 6.7933,
+      "step": 209
+    },
+    {
+      "epoch": 0.0021,
+      "grad_norm": 0.6890311463252623,
+      "learning_rate": 0.00063,
+      "loss": 6.7854,
+      "step": 210
+    },
+    {
+      "epoch": 0.00211,
+      "grad_norm": 0.9174002778722206,
+      "learning_rate": 0.000633,
+      "loss": 6.7849,
+      "step": 211
+    },
+    {
+      "epoch": 0.00212,
+      "grad_norm": 0.8086617968740898,
+      "learning_rate": 0.000636,
+      "loss": 6.7808,
+      "step": 212
+    },
+    {
+      "epoch": 0.00213,
+      "grad_norm": 0.6685717599500662,
+      "learning_rate": 0.000639,
+      "loss": 6.7542,
+      "step": 213
+    },
+    {
+      "epoch": 0.00214,
+      "grad_norm": 0.511917016650173,
+      "learning_rate": 0.000642,
+      "loss": 6.7483,
+      "step": 214
+    },
+    {
+      "epoch": 0.00215,
+      "grad_norm": 0.5132261185164837,
+      "learning_rate": 0.000645,
+      "loss": 6.7465,
+      "step": 215
+    },
+    {
+      "epoch": 0.00216,
+      "grad_norm": 0.3896647006337605,
+      "learning_rate": 0.000648,
+      "loss": 6.7354,
+      "step": 216
+    },
+    {
+      "epoch": 0.00217,
+      "grad_norm": 0.53153444147609,
+      "learning_rate": 0.000651,
+      "loss": 6.7114,
+      "step": 217
+    },
+    {
+      "epoch": 0.00218,
+      "grad_norm": 0.4560253950102483,
+      "learning_rate": 0.000654,
+      "loss": 6.7136,
+      "step": 218
+    },
+    {
+      "epoch": 0.00219,
+      "grad_norm": 0.38246603109839156,
+      "learning_rate": 0.000657,
+      "loss": 6.6847,
+      "step": 219
+    },
+    {
+      "epoch": 0.0022,
+      "grad_norm": 0.502249830770979,
+      "learning_rate": 0.00066,
+      "loss": 6.7061,
+      "step": 220
+    },
+    {
+      "epoch": 0.00221,
+      "grad_norm": 0.555840042257826,
+      "learning_rate": 0.0006630000000000001,
+      "loss": 6.6817,
+      "step": 221
+    },
+    {
+      "epoch": 0.00222,
+      "grad_norm": 0.7008290795132841,
+      "learning_rate": 0.000666,
+      "loss": 6.6751,
+      "step": 222
+    },
+    {
+      "epoch": 0.00223,
+      "grad_norm": 0.9665649898158697,
+      "learning_rate": 0.000669,
+      "loss": 6.6759,
+      "step": 223
+    },
+    {
+      "epoch": 0.00224,
+      "grad_norm": 1.0460190685952617,
+      "learning_rate": 0.0006720000000000001,
+      "loss": 6.6821,
+      "step": 224
+    },
+    {
+      "epoch": 0.00225,
+      "grad_norm": 0.9709238336565439,
+      "learning_rate": 0.000675,
+      "loss": 6.6643,
+      "step": 225
+    },
+    {
+      "epoch": 0.00226,
+      "grad_norm": 0.9675609159629996,
+      "learning_rate": 0.000678,
+      "loss": 6.6602,
+      "step": 226
+    },
+    {
+      "epoch": 0.00227,
+      "grad_norm": 0.6069059963900193,
+      "learning_rate": 0.0006810000000000001,
+      "loss": 6.6251,
+      "step": 227
+    },
+    {
+      "epoch": 0.00228,
+      "grad_norm": 0.6661980684886607,
+      "learning_rate": 0.000684,
+      "loss": 6.6314,
+      "step": 228
+    },
+    {
+      "epoch": 0.00229,
+      "grad_norm": 0.5755819115869941,
+      "learning_rate": 0.000687,
+      "loss": 6.6231,
+      "step": 229
+    },
+    {
+      "epoch": 0.0023,
+      "grad_norm": 0.48612508126201437,
+      "learning_rate": 0.0006900000000000001,
+      "loss": 6.6015,
+      "step": 230
+    },
+    {
+      "epoch": 0.00231,
+      "grad_norm": 0.42902206098374773,
+      "learning_rate": 0.000693,
+      "loss": 6.5844,
+      "step": 231
+    },
+    {
+      "epoch": 0.00232,
+      "grad_norm": 0.4617696239896233,
+      "learning_rate": 0.000696,
+      "loss": 6.5964,
+      "step": 232
+    },
+    {
+      "epoch": 0.00233,
+      "grad_norm": 0.397560103207551,
+      "learning_rate": 0.0006990000000000001,
+      "loss": 6.5819,
+      "step": 233
+    },
+    {
+      "epoch": 0.00234,
+      "grad_norm": 0.39436477469071923,
+      "learning_rate": 0.000702,
+      "loss": 6.5732,
+      "step": 234
+    },
+    {
+      "epoch": 0.00235,
+      "grad_norm": 0.37818254545129,
+      "learning_rate": 0.000705,
+      "loss": 6.5584,
+      "step": 235
+    },
+    {
+      "epoch": 0.00236,
+      "grad_norm": 0.39793300295732814,
+      "learning_rate": 0.000708,
+      "loss": 6.539,
+      "step": 236
+    },
+    {
+      "epoch": 0.00237,
+      "grad_norm": 0.32880148477167265,
+      "learning_rate": 0.0007109999999999999,
+      "loss": 6.5486,
+      "step": 237
+    },
+    {
+      "epoch": 0.00238,
+      "grad_norm": 0.33186649759843,
+      "learning_rate": 0.000714,
+      "loss": 6.5374,
+      "step": 238
+    },
+    {
+      "epoch": 0.00239,
+      "grad_norm": 0.3861082150171924,
+      "learning_rate": 0.000717,
+      "loss": 6.5195,
+      "step": 239
+    },
+    {
+      "epoch": 0.0024,
+      "grad_norm": 0.3818382545466509,
+      "learning_rate": 0.0007199999999999999,
+      "loss": 6.5368,
+      "step": 240
+    },
+    {
+      "epoch": 0.00241,
+      "grad_norm": 0.26517003068907236,
+      "learning_rate": 0.000723,
+      "loss": 6.5167,
+      "step": 241
+    },
+    {
+      "epoch": 0.00242,
+      "grad_norm": 0.30652105574179844,
+      "learning_rate": 0.000726,
+      "loss": 6.4934,
+      "step": 242
+    },
+    {
+      "epoch": 0.00243,
+      "grad_norm": 0.3382942246099826,
+      "learning_rate": 0.000729,
+      "loss": 6.4799,
+      "step": 243
+    },
+    {
+      "epoch": 0.00244,
+      "grad_norm": 0.4164388540502762,
+      "learning_rate": 0.000732,
+      "loss": 6.4843,
+      "step": 244
+    },
+    {
+      "epoch": 0.00245,
+      "grad_norm": 0.4035007765909141,
+      "learning_rate": 0.000735,
+      "loss": 6.4741,
+      "step": 245
+    },
+    {
+      "epoch": 0.00246,
+      "grad_norm": 0.4484502106991885,
+      "learning_rate": 0.000738,
+      "loss": 6.458,
+      "step": 246
+    },
+    {
+      "epoch": 0.00247,
+      "grad_norm": 0.6057401118193197,
+      "learning_rate": 0.000741,
+      "loss": 6.4543,
+      "step": 247
+    },
+    {
+      "epoch": 0.00248,
+      "grad_norm": 1.1775332556723501,
+      "learning_rate": 0.000744,
+      "loss": 6.4781,
+      "step": 248
+    },
+    {
+      "epoch": 0.00249,
+      "grad_norm": 1.0888595785598245,
+      "learning_rate": 0.000747,
+      "loss": 6.4635,
+      "step": 249
+    },
+    {
+      "epoch": 0.0025,
+      "grad_norm": 0.536872031808477,
+      "learning_rate": 0.00075,
+      "loss": 6.4479,
+      "step": 250
+    },
+    {
+      "epoch": 0.00251,
+      "grad_norm": 0.7926645181932281,
+      "learning_rate": 0.000753,
+      "loss": 6.4169,
+      "step": 251
+    },
+    {
+      "epoch": 0.00252,
+      "grad_norm": 0.6863348407685264,
+      "learning_rate": 0.000756,
+      "loss": 6.4273,
+      "step": 252
+    },
+    {
+      "epoch": 0.00253,
+      "grad_norm": 0.7123800606299509,
+      "learning_rate": 0.000759,
+      "loss": 6.4195,
+      "step": 253
+    },
+    {
+      "epoch": 0.00254,
+      "grad_norm": 0.839405849029746,
+      "learning_rate": 0.000762,
+      "loss": 6.4177,
+      "step": 254
+    },
+    {
+      "epoch": 0.00255,
+      "grad_norm": 0.76501143998226,
+      "learning_rate": 0.0007650000000000001,
+      "loss": 6.4159,
+      "step": 255
+    },
+    {
+      "epoch": 0.00256,
+      "grad_norm": 0.597608858095952,
+      "learning_rate": 0.000768,
+      "loss": 6.3889,
+      "step": 256
+    },
+    {
+      "epoch": 0.00257,
+      "grad_norm": 0.6526176951631347,
+      "learning_rate": 0.000771,
+      "loss": 6.3981,
+      "step": 257
+    },
+    {
+      "epoch": 0.00258,
+      "grad_norm": 0.651228257980475,
+      "learning_rate": 0.0007740000000000001,
+      "loss": 6.3725,
+      "step": 258
+    },
+    {
+      "epoch": 0.00259,
+      "grad_norm": 0.5603901273931662,
+      "learning_rate": 0.000777,
+      "loss": 6.3719,
+      "step": 259
+    },
+    {
+      "epoch": 0.0026,
+      "grad_norm": 0.41845020316479425,
+      "learning_rate": 0.0007800000000000001,
+      "loss": 6.3536,
+      "step": 260
+    },
+    {
+      "epoch": 0.00261,
+      "grad_norm": 0.5144884019867095,
+      "learning_rate": 0.0007830000000000001,
+      "loss": 6.3665,
+      "step": 261
+    },
+    {
+      "epoch": 0.00262,
+      "grad_norm": 0.5548811083770797,
+      "learning_rate": 0.000786,
+      "loss": 6.3412,
+      "step": 262
+    },
+    {
+      "epoch": 0.00263,
+      "grad_norm": 0.702068573310266,
+      "learning_rate": 0.0007890000000000001,
+      "loss": 6.353,
+      "step": 263
+    },
+    {
+      "epoch": 0.00264,
+      "grad_norm": 0.9481897048028406,
+      "learning_rate": 0.0007920000000000001,
+      "loss": 6.3404,
+      "step": 264
+    },
+    {
+      "epoch": 0.00265,
+      "grad_norm": 1.2297805755386195,
+      "learning_rate": 0.000795,
+      "loss": 6.3478,
+      "step": 265
+    },
+    {
+      "epoch": 0.00266,
+      "grad_norm": 0.5034998722006886,
+      "learning_rate": 0.0007980000000000001,
+      "loss": 6.3233,
+      "step": 266
+    },
+    {
+      "epoch": 0.00267,
+      "grad_norm": 0.8457797300321087,
+      "learning_rate": 0.0008010000000000001,
+      "loss": 6.3268,
+      "step": 267
+    },
+    {
+      "epoch": 0.00268,
+      "grad_norm": 0.7624901894608749,
+      "learning_rate": 0.000804,
+      "loss": 6.3226,
+      "step": 268
+    },
+    {
+      "epoch": 0.00269,
+      "grad_norm": 0.6803898428780553,
+      "learning_rate": 0.0008070000000000001,
+      "loss": 6.3045,
+      "step": 269
+    },
+    {
+      "epoch": 0.0027,
+      "grad_norm": 0.5891673657315365,
+      "learning_rate": 0.0008100000000000001,
+      "loss": 6.3,
+      "step": 270
+    },
+    {
+      "epoch": 0.00271,
+      "grad_norm": 0.6321969571669588,
+      "learning_rate": 0.000813,
+      "loss": 6.3081,
+      "step": 271
+    },
+    {
+      "epoch": 0.00272,
+      "grad_norm": 0.6080230974854919,
+      "learning_rate": 0.0008160000000000001,
+      "loss": 6.2911,
+      "step": 272
+    },
+    {
+      "epoch": 0.00273,
+      "grad_norm": 0.577176950863229,
+      "learning_rate": 0.0008190000000000001,
+      "loss": 6.2786,
+      "step": 273
+    },
+    {
+      "epoch": 0.00274,
+      "grad_norm": 0.46970800663022055,
+      "learning_rate": 0.000822,
+      "loss": 6.2573,
+      "step": 274
+    },
+    {
+      "epoch": 0.00275,
+      "grad_norm": 0.5095773122618286,
+      "learning_rate": 0.0008250000000000001,
+      "loss": 6.2676,
+      "step": 275
+    },
+    {
+      "epoch": 0.00276,
+      "grad_norm": 0.421367493059458,
+      "learning_rate": 0.0008280000000000001,
+      "loss": 6.2547,
+      "step": 276
+    },
+    {
+      "epoch": 0.00277,
+      "grad_norm": 0.4229723742956301,
+      "learning_rate": 0.0008310000000000001,
+      "loss": 6.2503,
+      "step": 277
+    },
+    {
+      "epoch": 0.00278,
+      "grad_norm": 0.4631977178825306,
+      "learning_rate": 0.0008340000000000001,
+      "loss": 6.2346,
+      "step": 278
+    },
+    {
+      "epoch": 0.00279,
+      "grad_norm": 0.41870110969580987,
+      "learning_rate": 0.0008370000000000001,
+      "loss": 6.2332,
+      "step": 279
+    },
+    {
+      "epoch": 0.0028,
+      "grad_norm": 0.4083314739680453,
+      "learning_rate": 0.0008400000000000001,
+      "loss": 6.2161,
+      "step": 280
+    },
+    {
+      "epoch": 0.00281,
+      "grad_norm": 0.42451645247510567,
+      "learning_rate": 0.0008430000000000001,
+      "loss": 6.2058,
+      "step": 281
+    },
+    {
+      "epoch": 0.00282,
+      "grad_norm": 0.4811013283391871,
+      "learning_rate": 0.000846,
+      "loss": 6.206,
+      "step": 282
+    },
+    {
+      "epoch": 0.00283,
+      "grad_norm": 0.6798083705841664,
+      "learning_rate": 0.0008489999999999999,
+      "loss": 6.2015,
+      "step": 283
+    },
+    {
+      "epoch": 0.00284,
+      "grad_norm": 1.0382201143248402,
+      "learning_rate": 0.0008519999999999999,
+      "loss": 6.2055,
+      "step": 284
+    },
+    {
+      "epoch": 0.00285,
+      "grad_norm": 1.115942818342409,
+      "learning_rate": 0.000855,
+      "loss": 6.2129,
+      "step": 285
+    },
+    {
+      "epoch": 0.00286,
+      "grad_norm": 0.8889955339821247,
+      "learning_rate": 0.0008579999999999999,
+      "loss": 6.187,
+      "step": 286
+    },
+    {
+      "epoch": 0.00287,
+      "grad_norm": 1.2422801265585652,
+      "learning_rate": 0.000861,
+      "loss": 6.209,
+      "step": 287
+    },
+    {
+      "epoch": 0.00288,
+      "grad_norm": 0.8315932342234975,
+      "learning_rate": 0.000864,
+      "loss": 6.174,
+      "step": 288
+    },
+    {
+      "epoch": 0.00289,
+      "grad_norm": 1.2914759013339998,
+      "learning_rate": 0.0008669999999999999,
+      "loss": 6.2078,
+      "step": 289
+    },
+    {
+      "epoch": 0.0029,
+      "grad_norm": 0.8376507056381004,
+      "learning_rate": 0.00087,
+      "loss": 6.1757,
+      "step": 290
+    },
+    {
+      "epoch": 0.00291,
+      "grad_norm": 0.8412780961911104,
+      "learning_rate": 0.000873,
+      "loss": 6.1658,
+      "step": 291
+    },
+    {
+      "epoch": 0.00292,
+      "grad_norm": 1.047021583757866,
+      "learning_rate": 0.0008759999999999999,
+      "loss": 6.1758,
+      "step": 292
+    },
+    {
+      "epoch": 0.00293,
+      "grad_norm": 0.8919470282886952,
+      "learning_rate": 0.000879,
+      "loss": 6.151,
+      "step": 293
+    },
+    {
+      "epoch": 0.00294,
+      "grad_norm": 0.665529628519212,
+      "learning_rate": 0.000882,
+      "loss": 6.159,
+      "step": 294
+    },
+    {
+      "epoch": 0.00295,
+      "grad_norm": 0.5169660787787601,
+      "learning_rate": 0.0008849999999999999,
+      "loss": 6.1239,
+      "step": 295
+    },
+    {
+      "epoch": 0.00296,
+      "grad_norm": 0.5611538425989948,
+      "learning_rate": 0.000888,
+      "loss": 6.1363,
+      "step": 296
+    },
+    {
+      "epoch": 0.00297,
+      "grad_norm": 0.46398604023920087,
+      "learning_rate": 0.000891,
+      "loss": 6.1045,
+      "step": 297
+    },
+    {
+      "epoch": 0.00298,
+      "grad_norm": 0.4361556326298739,
+      "learning_rate": 0.0008939999999999999,
+      "loss": 6.1198,
+      "step": 298
+    },
+    {
+      "epoch": 0.00299,
+      "grad_norm": 0.4319584905904094,
+      "learning_rate": 0.000897,
+      "loss": 6.0941,
+      "step": 299
+    },
+    {
+      "epoch": 0.003,
+      "grad_norm": 0.4255386299160817,
+      "learning_rate": 0.0009,
+      "loss": 6.0936,
+      "step": 300
+    },
+    {
+      "epoch": 0.00301,
+      "grad_norm": 0.3316584659066082,
+      "learning_rate": 0.0009029999999999999,
+      "loss": 6.0857,
+      "step": 301
+    },
+    {
+      "epoch": 0.00302,
+      "grad_norm": 0.37299869635167304,
+      "learning_rate": 0.000906,
+      "loss": 6.0685,
+      "step": 302
+    },
+    {
+      "epoch": 0.00303,
+      "grad_norm": 0.40148217950038195,
+      "learning_rate": 0.000909,
+      "loss": 6.0805,
+      "step": 303
+    },
+    {
+      "epoch": 0.00304,
+      "grad_norm": 0.420191340163935,
+      "learning_rate": 0.000912,
+      "loss": 6.0758,
+      "step": 304
+    },
+    {
+      "epoch": 0.00305,
+      "grad_norm": 0.45307668264044143,
+      "learning_rate": 0.000915,
+      "loss": 6.0736,
+      "step": 305
+    },
+    {
+      "epoch": 0.00306,
+      "grad_norm": 0.6122731249830943,
+      "learning_rate": 0.000918,
+      "loss": 6.0651,
+      "step": 306
+    },
+    {
+      "epoch": 0.00307,
+      "grad_norm": 0.851197326038436,
+      "learning_rate": 0.000921,
+      "loss": 6.0633,
+      "step": 307
+    },
+    {
+      "epoch": 0.00308,
+      "grad_norm": 1.1284591769922636,
+      "learning_rate": 0.000924,
+      "loss": 6.0582,
+      "step": 308
+    },
+    {
+      "epoch": 0.00309,
+      "grad_norm": 0.9596545216263644,
+      "learning_rate": 0.000927,
+      "loss": 6.0709,
+      "step": 309
+    },
+    {
+      "epoch": 0.0031,
+      "grad_norm": 1.039007932956353,
+      "learning_rate": 0.00093,
+      "loss": 6.0624,
+      "step": 310
+    },
+    {
+      "epoch": 0.00311,
+      "grad_norm": 0.9855401820369791,
+      "learning_rate": 0.000933,
+      "loss": 6.0524,
+      "step": 311
+    },
+    {
+      "epoch": 0.00312,
+      "grad_norm": 1.0163701418335827,
+      "learning_rate": 0.000936,
+      "loss": 6.041,
+      "step": 312
+    },
+    {
+      "epoch": 0.00313,
+      "grad_norm": 1.0223663613867633,
+      "learning_rate": 0.0009390000000000001,
+      "loss": 6.0491,
+      "step": 313
+    },
+    {
+      "epoch": 0.00314,
+      "grad_norm": 1.0997292958340223,
+      "learning_rate": 0.000942,
+      "loss": 6.0641,
+      "step": 314
+    },
+    {
+      "epoch": 0.00315,
+      "grad_norm": 0.932276773939602,
+      "learning_rate": 0.000945,
+      "loss": 6.0354,
+      "step": 315
+    },
+    {
+      "epoch": 0.00316,
+      "grad_norm": 0.8624268848533463,
+      "learning_rate": 0.0009480000000000001,
+      "loss": 6.0096,
+      "step": 316
+    },
+    {
+      "epoch": 0.00317,
+      "grad_norm": 0.6867359398602113,
+      "learning_rate": 0.000951,
+      "loss": 6.0237,
+      "step": 317
+    },
+    {
+      "epoch": 0.00318,
+      "grad_norm": 0.5777711812516898,
+      "learning_rate": 0.000954,
+      "loss": 6.014,
+      "step": 318
+    },
+    {
+      "epoch": 0.00319,
+      "grad_norm": 0.5907201170091796,
+      "learning_rate": 0.0009570000000000001,
+      "loss": 6.0042,
+      "step": 319
+    },
+    {
+      "epoch": 0.0032,
+      "grad_norm": 0.5929910498481646,
+      "learning_rate": 0.00096,
+      "loss": 6.0021,
+      "step": 320
+    },
+    {
+      "epoch": 0.00321,
+      "grad_norm": 0.6559636432249029,
+      "learning_rate": 0.000963,
+      "loss": 5.9891,
+      "step": 321
+    },
+    {
+      "epoch": 0.00322,
+      "grad_norm": 0.5844133161497509,
+      "learning_rate": 0.0009660000000000001,
+      "loss": 5.9766,
+      "step": 322
+    },
+    {
+      "epoch": 0.00323,
+      "grad_norm": 0.5466606066369618,
+      "learning_rate": 0.000969,
+      "loss": 5.9736,
+      "step": 323
+    },
+    {
+      "epoch": 0.00324,
+      "grad_norm": 0.563270781105711,
+      "learning_rate": 0.0009720000000000001,
+      "loss": 5.9778,
+      "step": 324
+    },
+    {
+      "epoch": 0.00325,
+      "grad_norm": 0.5312990845923178,
+      "learning_rate": 0.0009750000000000001,
+      "loss": 5.9405,
+      "step": 325
+    },
+    {
+      "epoch": 0.00326,
+      "grad_norm": 0.5118566622058196,
+      "learning_rate": 0.0009780000000000001,
+      "loss": 5.9566,
+      "step": 326
+    },
+    {
+      "epoch": 0.00327,
+      "grad_norm": 0.5259315695578027,
+      "learning_rate": 0.000981,
+      "loss": 5.9568,
+      "step": 327
+    },
+    {
+      "epoch": 0.00328,
+      "grad_norm": 0.5410551164101933,
+      "learning_rate": 0.000984,
+      "loss": 5.9324,
+      "step": 328
+    },
+    {
+      "epoch": 0.00329,
+      "grad_norm": 0.48301414107035934,
+      "learning_rate": 0.000987,
+      "loss": 5.931,
+      "step": 329
+    },
+    {
+      "epoch": 0.0033,
+      "grad_norm": 0.5975532498257996,
+      "learning_rate": 0.00099,
+      "loss": 5.9265,
+      "step": 330
+    },
+    {
+      "epoch": 0.00331,
+      "grad_norm": 0.9746373555768076,
+      "learning_rate": 0.0009930000000000002,
+      "loss": 5.9381,
+      "step": 331
+    },
+    {
+      "epoch": 0.00332,
+      "grad_norm": 1.1994973147641799,
+      "learning_rate": 0.0009960000000000001,
+      "loss": 5.9385,
+      "step": 332
+    },
+    {
+      "epoch": 0.00333,
+      "grad_norm": 0.5144992648158865,
+      "learning_rate": 0.000999,
+      "loss": 5.8989,
+      "step": 333
+    },
+    {
+      "epoch": 0.00334,
+      "grad_norm": 0.9073672400240862,
+      "learning_rate": 0.001002,
+      "loss": 5.9262,
+      "step": 334
+    },
+    {
+      "epoch": 0.00335,
+      "grad_norm": 0.7561451103694465,
+      "learning_rate": 0.001005,
+      "loss": 5.9186,
+      "step": 335
+    },
+    {
+      "epoch": 0.00336,
+      "grad_norm": 0.7872757919528415,
+      "learning_rate": 0.001008,
+      "loss": 5.9134,
+      "step": 336
+    },
+    {
+      "epoch": 0.00337,
+      "grad_norm": 0.7536020827923614,
+      "learning_rate": 0.0010110000000000002,
+      "loss": 5.8884,
+      "step": 337
+    },
+    {
+      "epoch": 0.00338,
+      "grad_norm": 1.0523353883962923,
+      "learning_rate": 0.0010140000000000001,
+      "loss": 5.9132,
+      "step": 338
+    },
+    {
+      "epoch": 0.00339,
+      "grad_norm": 1.2857238182949966,
+      "learning_rate": 0.0010170000000000001,
+      "loss": 5.9334,
+      "step": 339
+    },
+    {
+      "epoch": 0.0034,
+      "grad_norm": 0.7361708327689132,
+      "learning_rate": 0.00102,
+      "loss": 5.885,
+      "step": 340
+    },
+    {
+      "epoch": 0.00341,
+      "grad_norm": 0.6901997441262301,
+      "learning_rate": 0.001023,
+      "loss": 5.869,
+      "step": 341
+    },
+    {
+      "epoch": 0.00342,
+      "grad_norm": 0.6381033550571615,
+      "learning_rate": 0.001026,
+      "loss": 5.8702,
+      "step": 342
+    },
+    {
+      "epoch": 0.00343,
+      "grad_norm": 0.6071718978500397,
+      "learning_rate": 0.0010290000000000002,
+      "loss": 5.8743,
+      "step": 343
+    },
+    {
+      "epoch": 0.00344,
+      "grad_norm": 0.5857796625429044,
+      "learning_rate": 0.001032,
+      "loss": 5.861,
+      "step": 344
+    },
+    {
+      "epoch": 0.00345,
+      "grad_norm": 0.626640702848716,
+      "learning_rate": 0.001035,
+      "loss": 5.8537,
+      "step": 345
+    },
+    {
+      "epoch": 0.00346,
+      "grad_norm": 0.6755670022907736,
+      "learning_rate": 0.0010379999999999999,
+      "loss": 5.8603,
+      "step": 346
+    },
+    {
+      "epoch": 0.00347,
+      "grad_norm": 0.9144508249400731,
+      "learning_rate": 0.001041,
+      "loss": 5.8338,
+      "step": 347
+    },
+    {
+      "epoch": 0.00348,
+      "grad_norm": 1.2125876856754099,
+      "learning_rate": 0.001044,
+      "loss": 5.8634,
+      "step": 348
+    },
+    {
+      "epoch": 0.00349,
+      "grad_norm": 0.6928695941460523,
+      "learning_rate": 0.001047,
+      "loss": 5.8236,
+      "step": 349
+    },
+    {
+      "epoch": 0.0035,
+      "grad_norm": 0.7654262923967496,
+      "learning_rate": 0.00105,
+      "loss": 5.8502,
+      "step": 350
+    },
+    {
+      "epoch": 0.00351,
+      "grad_norm": 0.8300223804260752,
+      "learning_rate": 0.001053,
+      "loss": 5.8507,
+      "step": 351
+    },
+    {
+      "epoch": 0.00352,
+      "grad_norm": 1.1393832643973667,
+      "learning_rate": 0.0010559999999999999,
+      "loss": 5.841,
+      "step": 352
+    },
+    {
+      "epoch": 0.00353,
+      "grad_norm": 0.7670875434573843,
+      "learning_rate": 0.001059,
+      "loss": 5.8302,
+      "step": 353
+    },
+    {
+      "epoch": 0.00354,
+      "grad_norm": 0.8617169684849714,
+      "learning_rate": 0.001062,
+      "loss": 5.8072,
+      "step": 354
+    },
+    {
+      "epoch": 0.00355,
+      "grad_norm": 0.8787230305505044,
+      "learning_rate": 0.001065,
+      "loss": 5.8221,
+      "step": 355
+    },
+    {
+      "epoch": 0.00356,
+      "grad_norm": 0.9037602020080988,
+      "learning_rate": 0.001068,
+      "loss": 5.8164,
+      "step": 356
+    },
+    {
+      "epoch": 0.00357,
+      "grad_norm": 0.785887699185926,
+      "learning_rate": 0.001071,
+      "loss": 5.8055,
+      "step": 357
+    },
+    {
+      "epoch": 0.00358,
+      "grad_norm": 0.6152742029666318,
+      "learning_rate": 0.001074,
+      "loss": 5.7887,
+      "step": 358
+    },
+    {
+      "epoch": 0.00359,
+      "grad_norm": 0.5372063086433791,
+      "learning_rate": 0.001077,
+      "loss": 5.78,
+      "step": 359
+    },
+    {
+      "epoch": 0.0036,
+      "grad_norm": 0.5078257678271803,
+      "learning_rate": 0.00108,
+      "loss": 5.7825,
+      "step": 360
+    },
+    {
+      "epoch": 0.00361,
+      "grad_norm": 0.4885651334266738,
+      "learning_rate": 0.001083,
+      "loss": 5.7748,
+      "step": 361
+    },
+    {
+      "epoch": 0.00362,
+      "grad_norm": 0.5495429650143561,
+      "learning_rate": 0.001086,
+      "loss": 5.7596,
+      "step": 362
+    },
+    {
+      "epoch": 0.00363,
+      "grad_norm": 0.5626950540152672,
+      "learning_rate": 0.001089,
+      "loss": 5.7515,
+      "step": 363
+    },
+    {
+      "epoch": 0.00364,
+      "grad_norm": 0.6199658617744055,
+      "learning_rate": 0.001092,
+      "loss": 5.766,
+      "step": 364
+    },
+    {
+      "epoch": 0.00365,
+      "grad_norm": 0.7392438146286566,
+      "learning_rate": 0.001095,
+      "loss": 5.7655,
+      "step": 365
+    },
+    {
+      "epoch": 0.00366,
+      "grad_norm": 0.9864875055616179,
+      "learning_rate": 0.001098,
+      "loss": 5.7524,
+      "step": 366
+    },
+    {
+      "epoch": 0.00367,
+      "grad_norm": 1.1449768044417052,
+      "learning_rate": 0.001101,
+      "loss": 5.7648,
+      "step": 367
+    },
+    {
+      "epoch": 0.00368,
+      "grad_norm": 0.9013400729864322,
+      "learning_rate": 0.001104,
+      "loss": 5.755,
+      "step": 368
+    },
+    {
+      "epoch": 0.00369,
+      "grad_norm": 0.9093377711089596,
+      "learning_rate": 0.001107,
+      "loss": 5.7659,
+      "step": 369
+    },
+    {
+      "epoch": 0.0037,
+      "grad_norm": 0.575923493278037,
+      "learning_rate": 0.00111,
+      "loss": 5.7328,
+      "step": 370
+    },
+    {
+      "epoch": 0.00371,
+      "grad_norm": 0.6737016310188764,
+      "learning_rate": 0.001113,
+      "loss": 5.7102,
+      "step": 371
+    },
+    {
+      "epoch": 0.00372,
+      "grad_norm": 0.4833347808689347,
+      "learning_rate": 0.001116,
+      "loss": 5.7236,
+      "step": 372
+    },
+    {
+      "epoch": 0.00373,
+      "grad_norm": 0.6361357392920576,
+      "learning_rate": 0.001119,
+      "loss": 5.7181,
+      "step": 373
+    },
+    {
+      "epoch": 0.00374,
+      "grad_norm": 0.6286357986456886,
+      "learning_rate": 0.001122,
+      "loss": 5.7192,
+      "step": 374
+    },
+    {
+      "epoch": 0.00375,
+      "grad_norm": 0.7140127832546589,
+      "learning_rate": 0.0011250000000000001,
+      "loss": 5.7248,
+      "step": 375
+    },
+    {
+      "epoch": 0.00376,
+      "grad_norm": 0.728891228424708,
+      "learning_rate": 0.001128,
+      "loss": 5.7207,
+      "step": 376
+    },
+    {
+      "epoch": 0.00377,
+      "grad_norm": 0.7251122752592066,
+      "learning_rate": 0.001131,
+      "loss": 5.7141,
+      "step": 377
+    },
+    {
+      "epoch": 0.00378,
+      "grad_norm": 0.8109517942362439,
+      "learning_rate": 0.001134,
+      "loss": 5.7121,
+      "step": 378
+    },
+    {
+      "epoch": 0.00379,
+      "grad_norm": 0.7548765882892476,
+      "learning_rate": 0.001137,
+      "loss": 5.6981,
+      "step": 379
+    },
+    {
+      "epoch": 0.0038,
+      "grad_norm": 0.5982490555873449,
+      "learning_rate": 0.00114,
+      "loss": 5.7128,
+      "step": 380
+    },
+    {
+      "epoch": 0.00381,
+      "grad_norm": 0.5479723067602768,
+      "learning_rate": 0.0011430000000000001,
+      "loss": 5.6793,
+      "step": 381
+    },
+    {
+      "epoch": 0.00382,
+      "grad_norm": 0.5400365110175976,
+      "learning_rate": 0.001146,
+      "loss": 5.6631,
+      "step": 382
+    },
+    {
+      "epoch": 0.00383,
+      "grad_norm": 0.4406698702316126,
+      "learning_rate": 0.001149,
+      "loss": 5.673,
+      "step": 383
+    },
+    {
+      "epoch": 0.00384,
+      "grad_norm": 0.5495584711003424,
+      "learning_rate": 0.001152,
+      "loss": 5.6782,
+      "step": 384
+    },
+    {
+      "epoch": 0.00385,
+      "grad_norm": 0.7047837665038742,
+      "learning_rate": 0.001155,
+      "loss": 5.6686,
+      "step": 385
+    },
+    {
+      "epoch": 0.00386,
+      "grad_norm": 1.0039450355838517,
+      "learning_rate": 0.001158,
+      "loss": 5.6846,
+      "step": 386
+    },
+    {
+      "epoch": 0.00387,
+      "grad_norm": 1.2552299335364856,
+      "learning_rate": 0.0011610000000000001,
+      "loss": 5.6713,
+      "step": 387
+    },
+    {
+      "epoch": 0.00388,
+      "grad_norm": 1.0630057286422998,
+      "learning_rate": 0.001164,
+      "loss": 5.6615,
+      "step": 388
+    },
+    {
+      "epoch": 0.00389,
+      "grad_norm": 1.4085777330550793,
+      "learning_rate": 0.001167,
+      "loss": 5.6963,
+      "step": 389
+    },
+    {
+      "epoch": 0.0039,
+      "grad_norm": 0.7893066659624004,
+      "learning_rate": 0.00117,
+      "loss": 5.651,
+      "step": 390
+    },
+    {
+      "epoch": 0.00391,
+      "grad_norm": 0.8891270576556106,
+      "learning_rate": 0.001173,
+      "loss": 5.666,
+      "step": 391
+    },
+    {
+      "epoch": 0.00392,
+      "grad_norm": 1.1058013192110903,
+      "learning_rate": 0.001176,
+      "loss": 5.6796,
+      "step": 392
+    },
+    {
+      "epoch": 0.00393,
+      "grad_norm": 1.170614508410806,
+      "learning_rate": 0.0011790000000000001,
+      "loss": 5.6646,
+      "step": 393
+    },
+    {
+      "epoch": 0.00394,
+      "grad_norm": 0.8391276502601887,
+      "learning_rate": 0.001182,
+      "loss": 5.6402,
+      "step": 394
+    },
+    {
+      "epoch": 0.00395,
+      "grad_norm": 0.9435882620236007,
+      "learning_rate": 0.001185,
+      "loss": 5.6277,
+      "step": 395
+    },
+    {
+      "epoch": 0.00396,
+      "grad_norm": 0.7925001626557522,
+      "learning_rate": 0.001188,
+      "loss": 5.6404,
+      "step": 396
+    },
+    {
+      "epoch": 0.00397,
+      "grad_norm": 0.8633162203152536,
+      "learning_rate": 0.001191,
+      "loss": 5.6366,
+      "step": 397
+    },
+    {
+      "epoch": 0.00398,
+      "grad_norm": 0.9359127674730449,
+      "learning_rate": 0.0011940000000000002,
+      "loss": 5.6437,
+      "step": 398
+    },
+    {
+      "epoch": 0.00399,
+      "grad_norm": 1.0926478209626875,
+      "learning_rate": 0.0011970000000000001,
+      "loss": 5.6494,
+      "step": 399
+    },
+    {
+      "epoch": 0.004,
+      "grad_norm": 0.8943926064407558,
+      "learning_rate": 0.0012000000000000001,
+      "loss": 5.6306,
+      "step": 400
+    },
+    {
+      "epoch": 0.00401,
+      "grad_norm": 1.211025202532141,
+      "learning_rate": 0.001203,
+      "loss": 5.6241,
+      "step": 401
+    },
+    {
+      "epoch": 0.00402,
+      "grad_norm": 0.8585006093020132,
+      "learning_rate": 0.001206,
+      "loss": 5.6196,
+      "step": 402
+    },
+    {
+      "epoch": 0.00403,
+      "grad_norm": 0.8708424012246267,
+      "learning_rate": 0.001209,
+      "loss": 5.618,
+      "step": 403
+    },
+    {
+      "epoch": 0.00404,
+      "grad_norm": 0.6771235232466043,
+      "learning_rate": 0.0012120000000000002,
+      "loss": 5.6057,
+      "step": 404
+    },
+    {
+      "epoch": 0.00405,
+      "grad_norm": 0.5089480196948696,
+      "learning_rate": 0.0012150000000000002,
+      "loss": 5.5986,
+      "step": 405
+    },
+    {
+      "epoch": 0.00406,
+      "grad_norm": 0.5324784457955185,
+      "learning_rate": 0.0012180000000000001,
+      "loss": 5.583,
+      "step": 406
+    },
+    {
+      "epoch": 0.00407,
+      "grad_norm": 0.4806328504890235,
+      "learning_rate": 0.0012209999999999999,
+      "loss": 5.575,
+      "step": 407
+    },
+    {
+      "epoch": 0.00408,
+      "grad_norm": 0.5340674298116082,
+      "learning_rate": 0.001224,
+      "loss": 5.5941,
+      "step": 408
+    },
+    {
+      "epoch": 0.00409,
+      "grad_norm": 0.6817510563164704,
+      "learning_rate": 0.001227,
+      "loss": 5.5739,
+      "step": 409
+    },
+    {
+      "epoch": 0.0041,
+      "grad_norm": 0.8230982603577015,
+      "learning_rate": 0.00123,
+      "loss": 5.5739,
+      "step": 410
+    },
+    {
+      "epoch": 0.00411,
+      "grad_norm": 0.7982833366881755,
+      "learning_rate": 0.001233,
+      "loss": 5.5886,
+      "step": 411
+    },
+    {
+      "epoch": 0.00412,
+      "grad_norm": 0.7882120301866252,
+      "learning_rate": 0.001236,
+      "loss": 5.5767,
+      "step": 412
+    },
+    {
+      "epoch": 0.00413,
+      "grad_norm": 0.9078160072473371,
+      "learning_rate": 0.0012389999999999999,
+      "loss": 5.5798,
+      "step": 413
+    },
+    {
+      "epoch": 0.00414,
+      "grad_norm": 0.8046291024914881,
+      "learning_rate": 0.001242,
+      "loss": 5.5582,
+      "step": 414
+    },
+    {
+      "epoch": 0.00415,
+      "grad_norm": 0.9449024826812693,
+      "learning_rate": 0.001245,
+      "loss": 5.5633,
+      "step": 415
+    },
+    {
+      "epoch": 0.00416,
+      "grad_norm": 0.9578358959807691,
+      "learning_rate": 0.001248,
+      "loss": 5.5489,
+      "step": 416
+    },
+    {
+      "epoch": 0.00417,
+      "grad_norm": 0.7364680005190741,
+      "learning_rate": 0.001251,
+      "loss": 5.575,
+      "step": 417
+    },
+    {
+      "epoch": 0.00418,
+      "grad_norm": 0.5600093653837771,
+      "learning_rate": 0.001254,
+      "loss": 5.5419,
+      "step": 418
+    },
+    {
+      "epoch": 0.00419,
+      "grad_norm": 0.7369458002937045,
+      "learning_rate": 0.0012569999999999999,
+      "loss": 5.535,
+      "step": 419
+    },
+    {
+      "epoch": 0.0042,
+      "grad_norm": 0.7566412883958042,
+      "learning_rate": 0.00126,
+      "loss": 5.5478,
+      "step": 420
+    },
+    {
+      "epoch": 0.00421,
+      "grad_norm": 0.9341471688658377,
+      "learning_rate": 0.001263,
+      "loss": 5.5468,
+      "step": 421
+    },
+    {
+      "epoch": 0.00422,
+      "grad_norm": 0.9387048270351058,
+      "learning_rate": 0.001266,
+      "loss": 5.5395,
+      "step": 422
+    },
+    {
+      "epoch": 0.00423,
+      "grad_norm": 0.738543672170714,
+      "learning_rate": 0.001269,
+      "loss": 5.5309,
+      "step": 423
+    },
+    {
+      "epoch": 0.00424,
+      "grad_norm": 0.879163854006119,
+      "learning_rate": 0.001272,
+      "loss": 5.5379,
+      "step": 424
+    },
+    {
+      "epoch": 0.00425,
+      "grad_norm": 0.878245832078137,
+      "learning_rate": 0.001275,
+      "loss": 5.5393,
+      "step": 425
+    },
+    {
+      "epoch": 0.00426,
+      "grad_norm": 0.8393572375675296,
+      "learning_rate": 0.001278,
+      "loss": 5.5388,
+      "step": 426
+    },
+    {
+      "epoch": 0.00427,
+      "grad_norm": 0.8175993205302655,
+      "learning_rate": 0.001281,
+      "loss": 5.5188,
+      "step": 427
+    },
+    {
+      "epoch": 0.00428,
+      "grad_norm": 0.8492227718152501,
+      "learning_rate": 0.001284,
+      "loss": 5.5011,
+      "step": 428
+    },
+    {
+      "epoch": 0.00429,
+      "grad_norm": 0.8455500448937461,
+      "learning_rate": 0.001287,
+      "loss": 5.5167,
+      "step": 429
+    },
+    {
+      "epoch": 0.0043,
+      "grad_norm": 0.9588196540360735,
+      "learning_rate": 0.00129,
+      "loss": 5.5126,
+      "step": 430
+    },
+    {
+      "epoch": 0.00431,
+      "grad_norm": 1.0358439149859766,
+      "learning_rate": 0.001293,
+      "loss": 5.5121,
+      "step": 431
+    },
+    {
+      "epoch": 0.00432,
+      "grad_norm": 0.853137595287236,
+      "learning_rate": 0.001296,
+      "loss": 5.5152,
+      "step": 432
+    },
+    {
+      "epoch": 0.00433,
+      "grad_norm": 0.9144896540159448,
+      "learning_rate": 0.001299,
+      "loss": 5.5075,
+      "step": 433
+    },
+    {
+      "epoch": 0.00434,
+      "grad_norm": 1.0340397416077374,
+      "learning_rate": 0.001302,
+      "loss": 5.5131,
+      "step": 434
+    },
+    {
+      "epoch": 0.00435,
+      "grad_norm": 1.1136200661191735,
+      "learning_rate": 0.001305,
+      "loss": 5.5153,
+      "step": 435
+    },
+    {
+      "epoch": 0.00436,
+      "grad_norm": 0.7998503424321469,
+      "learning_rate": 0.001308,
+      "loss": 5.4814,
+      "step": 436
+    },
+    {
+      "epoch": 0.00437,
+      "grad_norm": 0.8862208467810537,
+      "learning_rate": 0.001311,
+      "loss": 5.5052,
+      "step": 437
+    },
+    {
+      "epoch": 0.00438,
+      "grad_norm": 0.85557749799579,
+      "learning_rate": 0.001314,
+      "loss": 5.4855,
+      "step": 438
+    },
+    {
+      "epoch": 0.00439,
+      "grad_norm": 0.6596001138977952,
+      "learning_rate": 0.001317,
+      "loss": 5.5056,
+      "step": 439
+    },
+    {
+      "epoch": 0.0044,
+      "grad_norm": 0.5461926920380444,
+      "learning_rate": 0.00132,
+      "loss": 5.4734,
+      "step": 440
+    },
+    {
+      "epoch": 0.00441,
+      "grad_norm": 0.5325344576484976,
+      "learning_rate": 0.001323,
+      "loss": 5.4692,
+      "step": 441
+    },
+    {
+      "epoch": 0.00442,
+      "grad_norm": 0.46029396349038315,
+      "learning_rate": 0.0013260000000000001,
+      "loss": 5.4603,
+      "step": 442
+    },
+    {
+      "epoch": 0.00443,
+      "grad_norm": 0.5200620875251907,
+      "learning_rate": 0.001329,
+      "loss": 5.4641,
+      "step": 443
+    },
+    {
+      "epoch": 0.00444,
+      "grad_norm": 0.511034817927936,
+      "learning_rate": 0.001332,
+      "loss": 5.4632,
+      "step": 444
+    },
+    {
+      "epoch": 0.00445,
+      "grad_norm": 0.61375364791033,
+      "learning_rate": 0.001335,
+      "loss": 5.4483,
+      "step": 445
+    },
+    {
+      "epoch": 0.00446,
+      "grad_norm": 0.7540282970336214,
+      "learning_rate": 0.001338,
+      "loss": 5.4549,
+      "step": 446
+    },
+    {
+      "epoch": 0.00447,
+      "grad_norm": 0.7743861790351634,
+      "learning_rate": 0.001341,
+      "loss": 5.456,
+      "step": 447
+    },
+    {
+      "epoch": 0.00448,
+      "grad_norm": 0.6949785247448689,
+      "learning_rate": 0.0013440000000000001,
+      "loss": 5.4375,
+      "step": 448
+    },
+    {
+      "epoch": 0.00449,
+      "grad_norm": 0.8972954522362333,
+      "learning_rate": 0.001347,
+      "loss": 5.4453,
+      "step": 449
+    },
+    {
+      "epoch": 0.0045,
+      "grad_norm": 1.0136292885280909,
+      "learning_rate": 0.00135,
+      "loss": 5.4524,
+      "step": 450
+    },
+    {
+      "epoch": 0.00451,
+      "grad_norm": 0.7959348815359711,
+      "learning_rate": 0.001353,
+      "loss": 5.4372,
+      "step": 451
+    },
+    {
+      "epoch": 0.00452,
+      "grad_norm": 0.750530581913797,
+      "learning_rate": 0.001356,
+      "loss": 5.4212,
+      "step": 452
+    },
+    {
+      "epoch": 0.00453,
+      "grad_norm": 0.718332283553841,
+      "learning_rate": 0.001359,
+      "loss": 5.4094,
+      "step": 453
+    },
+    {
+      "epoch": 0.00454,
+      "grad_norm": 0.8243339574967999,
+      "learning_rate": 0.0013620000000000001,
+      "loss": 5.4327,
+      "step": 454
+    },
+    {
+      "epoch": 0.00455,
+      "grad_norm": 0.8060545663764288,
+      "learning_rate": 0.0013650000000000001,
+      "loss": 5.4278,
+      "step": 455
+    },
+    {
+      "epoch": 0.00456,
+      "grad_norm": 0.9387057405661987,
+      "learning_rate": 0.001368,
+      "loss": 5.4287,
+      "step": 456
+    },
+    {
+      "epoch": 0.00457,
+      "grad_norm": 1.110172512819111,
+      "learning_rate": 0.001371,
+      "loss": 5.4304,
+      "step": 457
+    },
+    {
+      "epoch": 0.00458,
+      "grad_norm": 0.7485802071411273,
+      "learning_rate": 0.001374,
+      "loss": 5.4279,
+      "step": 458
+    },
+    {
+      "epoch": 0.00459,
+      "grad_norm": 0.846395295484429,
+      "learning_rate": 0.0013770000000000002,
+      "loss": 5.4177,
+      "step": 459
+    },
+    {
+      "epoch": 0.0046,
+      "grad_norm": 1.2095188964594632,
+      "learning_rate": 0.0013800000000000002,
+      "loss": 5.4166,
+      "step": 460
+    },
+    {
+      "epoch": 0.00461,
+      "grad_norm": 1.1548058188436976,
+      "learning_rate": 0.0013830000000000001,
+      "loss": 5.417,
+      "step": 461
+    },
+    {
+      "epoch": 0.00462,
+      "grad_norm": 0.9626057997692408,
+      "learning_rate": 0.001386,
+      "loss": 5.4177,
+      "step": 462
+    },
+    {
+      "epoch": 0.00463,
+      "grad_norm": 1.1365427244526745,
+      "learning_rate": 0.001389,
+      "loss": 5.4083,
+      "step": 463
+    },
+    {
+      "epoch": 0.00464,
+      "grad_norm": 0.7154214826672701,
+      "learning_rate": 0.001392,
+      "loss": 5.4147,
+      "step": 464
+    },
+    {
+      "epoch": 0.00465,
+      "grad_norm": 0.5933778225768791,
+      "learning_rate": 0.0013950000000000002,
+      "loss": 5.3806,
+      "step": 465
+    },
+    {
+      "epoch": 0.00466,
+      "grad_norm": 0.6213055581786315,
+      "learning_rate": 0.0013980000000000002,
+      "loss": 5.3973,
+      "step": 466
+    },
+    {
+      "epoch": 0.00467,
+      "grad_norm": 0.5608640811659587,
+      "learning_rate": 0.0014010000000000001,
+      "loss": 5.3871,
+      "step": 467
+    },
+    {
+      "epoch": 0.00468,
+      "grad_norm": 0.4459725756410885,
+      "learning_rate": 0.001404,
+      "loss": 5.3713,
+      "step": 468
+    },
+    {
+      "epoch": 0.00469,
+      "grad_norm": 0.46857319789964524,
+      "learning_rate": 0.001407,
+      "loss": 5.3733,
+      "step": 469
+    },
+    {
+      "epoch": 0.0047,
+      "grad_norm": 0.4864537455422831,
+      "learning_rate": 0.00141,
+      "loss": 5.3823,
+      "step": 470
+    },
+    {
+      "epoch": 0.00471,
+      "grad_norm": 0.5233417273033707,
+      "learning_rate": 0.001413,
+      "loss": 5.3595,
+      "step": 471
+    },
+    {
+      "epoch": 0.00472,
+      "grad_norm": 0.7276814872840428,
+      "learning_rate": 0.001416,
+      "loss": 5.376,
+      "step": 472
+    },
+    {
+      "epoch": 0.00473,
+      "grad_norm": 0.9313958457119089,
+      "learning_rate": 0.001419,
+      "loss": 5.3908,
+      "step": 473
+    },
+    {
+      "epoch": 0.00474,
+      "grad_norm": 0.9969581851520253,
+      "learning_rate": 0.0014219999999999999,
+      "loss": 5.3782,
+      "step": 474
+    },
+    {
+      "epoch": 0.00475,
+      "grad_norm": 0.7172709684261298,
+      "learning_rate": 0.001425,
+      "loss": 5.3626,
+      "step": 475
+    },
+    {
+      "epoch": 0.00476,
+      "grad_norm": 0.749267541315669,
+      "learning_rate": 0.001428,
+      "loss": 5.395,
+      "step": 476
+    },
+    {
+      "epoch": 0.00477,
+      "grad_norm": 0.7586220730764037,
+      "learning_rate": 0.001431,
+      "loss": 5.35,
+      "step": 477
+    },
+    {
+      "epoch": 0.00478,
+      "grad_norm": 0.8456343691773762,
+      "learning_rate": 0.001434,
+      "loss": 5.378,
+      "step": 478
+    },
+    {
+      "epoch": 0.00479,
+      "grad_norm": 1.0643326937248148,
+      "learning_rate": 0.001437,
+      "loss": 5.3628,
+      "step": 479
+    },
+    {
+      "epoch": 0.0048,
+      "grad_norm": 0.9414664669888669,
+      "learning_rate": 0.0014399999999999999,
+      "loss": 5.3552,
+      "step": 480
+    },
+    {
+      "epoch": 0.00481,
+      "grad_norm": 0.9003358375425434,
+      "learning_rate": 0.001443,
+      "loss": 5.3593,
+      "step": 481
+    },
+    {
+      "epoch": 0.00482,
+      "grad_norm": 0.8978331557974801,
+      "learning_rate": 0.001446,
+      "loss": 5.3398,
+      "step": 482
+    },
+    {
+      "epoch": 0.00483,
+      "grad_norm": 1.1289938883763697,
+      "learning_rate": 0.001449,
+      "loss": 5.3578,
+      "step": 483
+    },
+    {
+      "epoch": 0.00484,
+      "grad_norm": 1.1013965257300222,
+      "learning_rate": 0.001452,
+      "loss": 5.3611,
+      "step": 484
+    },
+    {
+      "epoch": 0.00485,
+      "grad_norm": 0.9590590729365723,
+      "learning_rate": 0.001455,
+      "loss": 5.3653,
+      "step": 485
+    },
+    {
+      "epoch": 0.00486,
+      "grad_norm": 0.8987758447866343,
+      "learning_rate": 0.001458,
+      "loss": 5.3363,
+      "step": 486
+    },
+    {
+      "epoch": 0.00487,
+      "grad_norm": 1.1924974582473045,
+      "learning_rate": 0.001461,
+      "loss": 5.3676,
+      "step": 487
+    },
+    {
+      "epoch": 0.00488,
+      "grad_norm": 0.945268328759551,
+      "learning_rate": 0.001464,
+      "loss": 5.3393,
+      "step": 488
+    },
+    {
+      "epoch": 0.00489,
+      "grad_norm": 0.9295239913179795,
+      "learning_rate": 0.001467,
+      "loss": 5.3304,
+      "step": 489
+    },
+    {
+      "epoch": 0.0049,
+      "grad_norm": 0.8440135575742712,
+      "learning_rate": 0.00147,
+      "loss": 5.3401,
+      "step": 490
+    },
+    {
+      "epoch": 0.00491,
+      "grad_norm": 0.8518595255895461,
+      "learning_rate": 0.001473,
+      "loss": 5.3387,
+      "step": 491
+    },
+    {
+      "epoch": 0.00492,
+      "grad_norm": 0.8521564820940866,
+      "learning_rate": 0.001476,
+      "loss": 5.3306,
+      "step": 492
+    },
+    {
+      "epoch": 0.00493,
+      "grad_norm": 1.037786125041147,
+      "learning_rate": 0.001479,
+      "loss": 5.337,
+      "step": 493
+    },
+    {
+      "epoch": 0.00494,
+      "grad_norm": 0.8602402216970677,
+      "learning_rate": 0.001482,
+      "loss": 5.3365,
+      "step": 494
+    },
+    {
+      "epoch": 0.00495,
+      "grad_norm": 0.833325441881814,
+      "learning_rate": 0.001485,
+      "loss": 5.3402,
+      "step": 495
+    },
+    {
+      "epoch": 0.00496,
+      "grad_norm": 0.7919536178225737,
+      "learning_rate": 0.001488,
+      "loss": 5.324,
+      "step": 496
+    },
+    {
+      "epoch": 0.00497,
+      "grad_norm": 0.873050532284563,
+      "learning_rate": 0.001491,
+      "loss": 5.3076,
+      "step": 497
+    },
+    {
+      "epoch": 0.00498,
+      "grad_norm": 0.8401174504098841,
+      "learning_rate": 0.001494,
+      "loss": 5.3331,
+      "step": 498
+    },
+    {
+      "epoch": 0.00499,
+      "grad_norm": 0.7077904347820982,
+      "learning_rate": 0.001497,
+      "loss": 5.3023,
+      "step": 499
+    },
+    {
+      "epoch": 0.005,
+      "grad_norm": 0.722193970534152,
+      "learning_rate": 0.0015,
+      "loss": 5.3084,
+      "step": 500
+    },
+    {
+      "epoch": 0.00501,
+      "grad_norm": 0.6788524699363854,
+      "learning_rate": 0.001503,
+      "loss": 5.3068,
+      "step": 501
+    },
+    {
+      "epoch": 0.00502,
+      "grad_norm": 0.4688637134987383,
+      "learning_rate": 0.001506,
+      "loss": 5.3027,
+      "step": 502
+    },
+    {
+      "epoch": 0.00503,
+      "grad_norm": 0.5347885266184398,
+      "learning_rate": 0.0015090000000000001,
+      "loss": 5.3107,
+      "step": 503
+    },
+    {
+      "epoch": 0.00504,
+      "grad_norm": 0.4568589587316421,
+      "learning_rate": 0.001512,
+      "loss": 5.2883,
+      "step": 504
+    },
+    {
+      "epoch": 0.00505,
+      "grad_norm": 0.6059364079243326,
+      "learning_rate": 0.001515,
+      "loss": 5.2843,
+      "step": 505
+    },
+    {
+      "epoch": 0.00506,
+      "grad_norm": 0.5690871559871206,
+      "learning_rate": 0.001518,
+      "loss": 5.2813,
+      "step": 506
+    },
+    {
+      "epoch": 0.00507,
+      "grad_norm": 0.6408278920334604,
+      "learning_rate": 0.001521,
+      "loss": 5.279,
+      "step": 507
+    },
+    {
+      "epoch": 0.00508,
+      "grad_norm": 0.6374798421453328,
+      "learning_rate": 0.001524,
+      "loss": 5.2759,
+      "step": 508
+    },
+    {
+      "epoch": 0.00509,
+      "grad_norm": 0.5508180980448808,
+      "learning_rate": 0.0015270000000000001,
+      "loss": 5.2603,
+      "step": 509
+    },
+    {
+      "epoch": 0.0051,
+      "grad_norm": 0.4678264081626914,
+      "learning_rate": 0.0015300000000000001,
+      "loss": 5.2734,
+      "step": 510
+    },
+    {
+      "epoch": 0.00511,
+      "grad_norm": 0.4947375080326457,
+      "learning_rate": 0.001533,
+      "loss": 5.252,
+      "step": 511
+    },
+    {
+      "epoch": 0.00512,
+      "grad_norm": 0.4646141343190718,
+      "learning_rate": 0.001536,
+      "loss": 5.2454,
+      "step": 512
+    },
+    {
+      "epoch": 0.00513,
+      "grad_norm": 0.4614608859817223,
+      "learning_rate": 0.001539,
+      "loss": 5.2616,
+      "step": 513
+    },
+    {
+      "epoch": 0.00514,
+      "grad_norm": 0.47514560040822135,
+      "learning_rate": 0.001542,
+      "loss": 5.2462,
+      "step": 514
+    },
+    {
+      "epoch": 0.00515,
+      "grad_norm": 0.5516172292908152,
+      "learning_rate": 0.0015450000000000001,
+      "loss": 5.2495,
+      "step": 515
+    },
+    {
+      "epoch": 0.00516,
+      "grad_norm": 0.6762883843976203,
+      "learning_rate": 0.0015480000000000001,
+      "loss": 5.2514,
+      "step": 516
+    },
+    {
+      "epoch": 0.00517,
+      "grad_norm": 0.8220059659316783,
+      "learning_rate": 0.001551,
+      "loss": 5.2491,
+      "step": 517
+    },
+    {
+      "epoch": 0.00518,
+      "grad_norm": 0.8472303407674783,
+      "learning_rate": 0.001554,
+      "loss": 5.2384,
+      "step": 518
+    },
+    {
+      "epoch": 0.00519,
+      "grad_norm": 0.8352554081583162,
+      "learning_rate": 0.001557,
+      "loss": 5.2403,
+      "step": 519
+    },
+    {
+      "epoch": 0.0052,
+      "grad_norm": 0.9289527635802941,
+      "learning_rate": 0.0015600000000000002,
+      "loss": 5.2527,
+      "step": 520
+    },
+    {
+      "epoch": 0.00521,
+      "grad_norm": 1.112613217213008,
+      "learning_rate": 0.0015630000000000002,
+      "loss": 5.242,
+      "step": 521
+    },
+    {
+      "epoch": 0.00522,
+      "grad_norm": 0.9402831205824588,
+      "learning_rate": 0.0015660000000000001,
+      "loss": 5.2388,
+      "step": 522
+    },
+    {
+      "epoch": 0.00523,
+      "grad_norm": 1.103453237567065,
+      "learning_rate": 0.001569,
+      "loss": 5.2627,
+      "step": 523
+    },
+    {
+      "epoch": 0.00524,
+      "grad_norm": 1.0118007660209627,
+      "learning_rate": 0.001572,
+      "loss": 5.268,
+      "step": 524
+    },
+    {
+      "epoch": 0.00525,
+      "grad_norm": 0.9435955195760587,
+      "learning_rate": 0.001575,
+      "loss": 5.2606,
+      "step": 525
+    },
+    {
+      "epoch": 0.00526,
+      "grad_norm": 0.8731456366832249,
+      "learning_rate": 0.0015780000000000002,
+      "loss": 5.2355,
+      "step": 526
+    },
+    {
+      "epoch": 0.00527,
+      "grad_norm": 0.7556141048030679,
+      "learning_rate": 0.0015810000000000002,
+      "loss": 5.244,
+      "step": 527
+    },
+    {
+      "epoch": 0.00528,
+      "grad_norm": 0.8959181971092488,
+      "learning_rate": 0.0015840000000000001,
+      "loss": 5.2287,
+      "step": 528
+    },
+    {
+      "epoch": 0.00529,
+      "grad_norm": 1.10956087267792,
+      "learning_rate": 0.001587,
+      "loss": 5.2485,
+      "step": 529
+    },
+    {
+      "epoch": 0.0053,
+      "grad_norm": 0.7373688170406403,
+      "learning_rate": 0.00159,
+      "loss": 5.2473,
+      "step": 530
+    },
+    {
+      "epoch": 0.00531,
+      "grad_norm": 0.9440929198745536,
+      "learning_rate": 0.001593,
+      "loss": 5.2364,
+      "step": 531
+    },
+    {
+      "epoch": 0.00532,
+      "grad_norm": 1.2761807884581948,
+      "learning_rate": 0.0015960000000000002,
+      "loss": 5.2627,
+      "step": 532
+    },
+    {
+      "epoch": 0.00533,
+      "grad_norm": 0.7254918561906474,
+      "learning_rate": 0.0015990000000000002,
+      "loss": 5.2411,
+      "step": 533
+    },
+    {
+      "epoch": 0.00534,
+      "grad_norm": 0.9458095799105631,
+      "learning_rate": 0.0016020000000000001,
+      "loss": 5.2364,
+      "step": 534
+    },
+    {
+      "epoch": 0.00535,
+      "grad_norm": 1.1518877149230409,
+      "learning_rate": 0.001605,
+      "loss": 5.2538,
+      "step": 535
+    },
+    {
+      "epoch": 0.00536,
+      "grad_norm": 1.0003795894462209,
+      "learning_rate": 0.001608,
+      "loss": 5.2516,
+      "step": 536
+    },
+    {
+      "epoch": 0.00537,
+      "grad_norm": 1.0837750602504435,
+      "learning_rate": 0.0016110000000000002,
+      "loss": 5.245,
+      "step": 537
+    },
+    {
+      "epoch": 0.00538,
+      "grad_norm": 0.8341741073056014,
+      "learning_rate": 0.0016140000000000002,
+      "loss": 5.2215,
+      "step": 538
+    },
+    {
+      "epoch": 0.00539,
+      "grad_norm": 0.8405248729905289,
+      "learning_rate": 0.0016170000000000002,
+      "loss": 5.218,
+      "step": 539
+    },
+    {
+      "epoch": 0.0054,
+      "grad_norm": 1.0652634270364731,
+      "learning_rate": 0.0016200000000000001,
+      "loss": 5.2348,
+      "step": 540
+    },
+    {
+      "epoch": 0.00541,
+      "grad_norm": 1.02504417487003,
+      "learning_rate": 0.001623,
+      "loss": 5.2275,
+      "step": 541
+    },
+    {
+      "epoch": 0.00542,
+      "grad_norm": 1.0410836109463524,
+      "learning_rate": 0.001626,
+      "loss": 5.2461,
+      "step": 542
+    },
+    {
+      "epoch": 0.00543,
+      "grad_norm": 1.1820143116937272,
+      "learning_rate": 0.0016290000000000002,
+      "loss": 5.2251,
+      "step": 543
+    },
+    {
+      "epoch": 0.00544,
+      "grad_norm": 0.7766425096978492,
+      "learning_rate": 0.0016320000000000002,
+      "loss": 5.227,
+      "step": 544
+    },
+    {
+      "epoch": 0.00545,
+      "grad_norm": 0.7162190734488068,
+      "learning_rate": 0.0016350000000000002,
+      "loss": 5.2344,
+      "step": 545
+    },
+    {
+      "epoch": 0.00546,
+      "grad_norm": 0.6138609629035112,
+      "learning_rate": 0.0016380000000000001,
+      "loss": 5.1997,
+      "step": 546
+    },
+    {
+      "epoch": 0.00547,
+      "grad_norm": 0.5845964576550613,
+      "learning_rate": 0.001641,
+      "loss": 5.199,
+      "step": 547
+    },
+    {
+      "epoch": 0.00548,
+      "grad_norm": 0.5000045746748227,
+      "learning_rate": 0.001644,
+      "loss": 5.1903,
+      "step": 548
+    },
+    {
+      "epoch": 0.00549,
+      "grad_norm": 0.45001097530426093,
+      "learning_rate": 0.0016470000000000002,
+      "loss": 5.1848,
+      "step": 549
+    },
+    {
+      "epoch": 0.0055,
+      "grad_norm": 0.48852089918970865,
+      "learning_rate": 0.0016500000000000002,
+      "loss": 5.1966,
+      "step": 550
+    },
+    {
+      "epoch": 0.00551,
+      "grad_norm": 0.4504277526474317,
+      "learning_rate": 0.0016530000000000002,
+      "loss": 5.1901,
+      "step": 551
+    },
+    {
+      "epoch": 0.00552,
+      "grad_norm": 0.3849177997836439,
+      "learning_rate": 0.0016560000000000001,
+      "loss": 5.1829,
+      "step": 552
+    },
+    {
+      "epoch": 0.00553,
+      "grad_norm": 0.3648459807382568,
+      "learning_rate": 0.001659,
+      "loss": 5.1708,
+      "step": 553
+    },
+    {
+      "epoch": 0.00554,
+      "grad_norm": 0.36806425815097327,
+      "learning_rate": 0.0016620000000000003,
+      "loss": 5.1683,
+      "step": 554
+    },
+    {
+      "epoch": 0.00555,
+      "grad_norm": 0.399857849170916,
+      "learning_rate": 0.0016650000000000002,
+      "loss": 5.174,
+      "step": 555
+    },
+    {
+      "epoch": 0.00556,
+      "grad_norm": 0.5317181922092629,
+      "learning_rate": 0.0016680000000000002,
+      "loss": 5.1608,
+      "step": 556
+    },
+    {
+      "epoch": 0.00557,
+      "grad_norm": 0.7831777956168835,
+      "learning_rate": 0.0016710000000000002,
+      "loss": 5.1496,
+      "step": 557
+    },
+    {
+      "epoch": 0.00558,
+      "grad_norm": 1.0341253030675983,
+      "learning_rate": 0.0016740000000000001,
+      "loss": 5.1915,
+      "step": 558
+    },
+    {
+      "epoch": 0.00559,
+      "grad_norm": 0.9310990839840096,
+      "learning_rate": 0.001677,
+      "loss": 5.1724,
+      "step": 559
+    },
+    {
+      "epoch": 0.0056,
+      "grad_norm": 1.1320536331440925,
+      "learning_rate": 0.0016800000000000003,
+      "loss": 5.1553,
+      "step": 560
+    },
+    {
+      "epoch": 0.00561,
+      "grad_norm": 1.0798635762834115,
+      "learning_rate": 0.0016830000000000003,
+      "loss": 5.1888,
+      "step": 561
+    },
+    {
+      "epoch": 0.00562,
+      "grad_norm": 0.9955409946448898,
+      "learning_rate": 0.0016860000000000002,
+      "loss": 5.1785,
+      "step": 562
+    },
+    {
+      "epoch": 0.00563,
+      "grad_norm": 0.8046398619546209,
+      "learning_rate": 0.001689,
+      "loss": 5.1727,
+      "step": 563
+    },
+    {
+      "epoch": 0.00564,
+      "grad_norm": 0.8640806614647399,
+      "learning_rate": 0.001692,
+      "loss": 5.1564,
+      "step": 564
+    },
+    {
+      "epoch": 0.00565,
+      "grad_norm": 0.9094590673653183,
+      "learning_rate": 0.001695,
+      "loss": 5.1591,
+      "step": 565
+    },
+    {
+      "epoch": 0.00566,
+      "grad_norm": 1.0763832222921514,
+      "learning_rate": 0.0016979999999999999,
+      "loss": 5.1686,
+      "step": 566
+    },
+    {
+      "epoch": 0.00567,
+      "grad_norm": 0.9106969464088174,
+      "learning_rate": 0.0017009999999999998,
+      "loss": 5.1536,
+      "step": 567
+    },
+    {
+      "epoch": 0.00568,
+      "grad_norm": 1.0612496613421336,
+      "learning_rate": 0.0017039999999999998,
+      "loss": 5.175,
+      "step": 568
+    },
+    {
+      "epoch": 0.00569,
+      "grad_norm": 1.030428797640198,
+      "learning_rate": 0.001707,
+      "loss": 5.1588,
+      "step": 569
+    },
+    {
+      "epoch": 0.0057,
+      "grad_norm": 1.0168109110540526,
+      "learning_rate": 0.00171,
+      "loss": 5.1716,
+      "step": 570
+    },
+    {
+      "epoch": 0.00571,
+      "grad_norm": 1.1348044931505399,
+      "learning_rate": 0.001713,
+      "loss": 5.1856,
+      "step": 571
+    },
+    {
+      "epoch": 0.00572,
+      "grad_norm": 0.761146802779985,
+      "learning_rate": 0.0017159999999999999,
+      "loss": 5.1497,
+      "step": 572
+    },
+    {
+      "epoch": 0.00573,
+      "grad_norm": 0.7662650183514703,
+      "learning_rate": 0.0017189999999999998,
+      "loss": 5.1536,
+      "step": 573
+    },
+    {
+      "epoch": 0.00574,
+      "grad_norm": 0.8733185334950944,
+      "learning_rate": 0.001722,
+      "loss": 5.1641,
+      "step": 574
+    },
+    {
+      "epoch": 0.00575,
+      "grad_norm": 0.8629871927891676,
+      "learning_rate": 0.001725,
+      "loss": 5.142,
+      "step": 575
+    },
+    {
+      "epoch": 0.00576,
+      "grad_norm": 0.8467410686880783,
+      "learning_rate": 0.001728,
+      "loss": 5.137,
+      "step": 576
+    },
+    {
+      "epoch": 0.00577,
+      "grad_norm": 0.8007452144545169,
+      "learning_rate": 0.001731,
+      "loss": 5.1615,
+      "step": 577
+    },
+    {
+      "epoch": 0.00578,
+      "grad_norm": 0.6405300314411674,
+      "learning_rate": 0.0017339999999999999,
+      "loss": 5.1325,
+      "step": 578
+    },
+    {
+      "epoch": 0.00579,
+      "grad_norm": 0.70587507977218,
+      "learning_rate": 0.0017369999999999998,
+      "loss": 5.1208,
+      "step": 579
+    },
+    {
+      "epoch": 0.0058,
+      "grad_norm": 0.6985854652221088,
+      "learning_rate": 0.00174,
+      "loss": 5.1378,
+      "step": 580
+    },
+    {
+      "epoch": 0.00581,
+      "grad_norm": 0.6137529113435678,
+      "learning_rate": 0.001743,
+      "loss": 5.1348,
+      "step": 581
+    },
+    {
+      "epoch": 0.00582,
+      "grad_norm": 0.48264514730128444,
+      "learning_rate": 0.001746,
+      "loss": 5.1045,
+      "step": 582
+    },
+    {
+      "epoch": 0.00583,
+      "grad_norm": 0.5771367675923135,
+      "learning_rate": 0.001749,
+      "loss": 5.1293,
+      "step": 583
+    },
+    {
+      "epoch": 0.00584,
+      "grad_norm": 0.6936037332792736,
+      "learning_rate": 0.0017519999999999999,
+      "loss": 5.1229,
+      "step": 584
+    },
+    {
+      "epoch": 0.00585,
+      "grad_norm": 0.7595444862659105,
+      "learning_rate": 0.0017549999999999998,
+      "loss": 5.1095,
+      "step": 585
+    },
+    {
+      "epoch": 0.00586,
+      "grad_norm": 0.8605796861767604,
+      "learning_rate": 0.001758,
+      "loss": 5.1202,
+      "step": 586
+    },
+    {
+      "epoch": 0.00587,
+      "grad_norm": 0.907920304070859,
+      "learning_rate": 0.001761,
+      "loss": 5.1079,
+      "step": 587
+    },
+    {
+      "epoch": 0.00588,
+      "grad_norm": 0.9246177820411534,
+      "learning_rate": 0.001764,
+      "loss": 5.1163,
+      "step": 588
+    },
+    {
+      "epoch": 0.00589,
+      "grad_norm": 0.8704696950275989,
+      "learning_rate": 0.001767,
+      "loss": 5.1211,
+      "step": 589
+    },
+    {
+      "epoch": 0.0059,
+      "grad_norm": 0.7193563853719346,
+      "learning_rate": 0.0017699999999999999,
+      "loss": 5.101,
+      "step": 590
+    },
+    {
+      "epoch": 0.00591,
+      "grad_norm": 0.6949293332158505,
+      "learning_rate": 0.001773,
+      "loss": 5.0928,
+      "step": 591
+    },
+    {
+      "epoch": 0.00592,
+      "grad_norm": 0.654055845037553,
+      "learning_rate": 0.001776,
+      "loss": 5.1147,
+      "step": 592
+    },
+    {
+      "epoch": 0.00593,
+      "grad_norm": 0.7140132542710956,
+      "learning_rate": 0.001779,
+      "loss": 5.1161,
+      "step": 593
+    },
+    {
+      "epoch": 0.00594,
+      "grad_norm": 0.6270820258277058,
+      "learning_rate": 0.001782,
+      "loss": 5.107,
+      "step": 594
+    },
+    {
+      "epoch": 0.00595,
+      "grad_norm": 0.5627466921921568,
+      "learning_rate": 0.001785,
+      "loss": 5.0962,
+      "step": 595
+    },
+    {
+      "epoch": 0.00596,
+      "grad_norm": 0.6408060930916085,
+      "learning_rate": 0.0017879999999999999,
+      "loss": 5.0966,
+      "step": 596
+    },
+    {
+      "epoch": 0.00597,
+      "grad_norm": 0.7315483645065906,
+      "learning_rate": 0.001791,
+      "loss": 5.077,
+      "step": 597
+    },
+    {
+      "epoch": 0.00598,
+      "grad_norm": 0.7860917802127037,
+      "learning_rate": 0.001794,
+      "loss": 5.0917,
+      "step": 598
+    },
+    {
+      "epoch": 0.00599,
+      "grad_norm": 0.7396694082174577,
+      "learning_rate": 0.001797,
+      "loss": 5.0901,
+      "step": 599
+    },
+    {
+      "epoch": 0.006,
+      "grad_norm": 0.678011019966431,
+      "learning_rate": 0.0018,
+      "loss": 5.0859,
+      "step": 600
+    },
+    {
+      "epoch": 0.00601,
+      "grad_norm": 0.8039567544190375,
+      "learning_rate": 0.001803,
+      "loss": 5.0817,
+      "step": 601
+    },
+    {
+      "epoch": 0.00602,
+      "grad_norm": 0.8362811130198314,
+      "learning_rate": 0.0018059999999999999,
+      "loss": 5.0736,
+      "step": 602
+    },
+    {
+      "epoch": 0.00603,
+      "grad_norm": 0.7437701841167941,
+      "learning_rate": 0.001809,
+      "loss": 5.0897,
+      "step": 603
+    },
+    {
+      "epoch": 0.00604,
+      "grad_norm": 0.7865788629191941,
+      "learning_rate": 0.001812,
+      "loss": 5.0711,
+      "step": 604
+    },
+    {
+      "epoch": 0.00605,
+      "grad_norm": 0.7065527487675252,
+      "learning_rate": 0.001815,
+      "loss": 5.0798,
+      "step": 605
+    },
+    {
+      "epoch": 0.00606,
+      "grad_norm": 0.6576721805135337,
+      "learning_rate": 0.001818,
+      "loss": 5.0782,
+      "step": 606
+    },
+    {
+      "epoch": 0.00607,
+      "grad_norm": 0.6528064751937639,
+      "learning_rate": 0.001821,
+      "loss": 5.0831,
+      "step": 607
+    },
+    {
+      "epoch": 0.00608,
+      "grad_norm": 0.6182798778507802,
+      "learning_rate": 0.001824,
+      "loss": 5.0615,
+      "step": 608
+    },
+    {
+      "epoch": 0.00609,
+      "grad_norm": 0.8163285355074863,
+      "learning_rate": 0.001827,
+      "loss": 5.079,
+      "step": 609
+    },
+    {
+      "epoch": 0.0061,
+      "grad_norm": 0.795800426894062,
+      "learning_rate": 0.00183,
+      "loss": 5.053,
+      "step": 610
+    },
+    {
+      "epoch": 0.00611,
+      "grad_norm": 0.7510077277929101,
+      "learning_rate": 0.001833,
+      "loss": 5.0545,
+      "step": 611
+    },
+    {
+      "epoch": 0.00612,
+      "grad_norm": 0.7863529213037952,
+      "learning_rate": 0.001836,
+      "loss": 5.0625,
+      "step": 612
+    },
+    {
+      "epoch": 0.00613,
+      "grad_norm": 0.718541559578873,
+      "learning_rate": 0.001839,
+      "loss": 5.0526,
+      "step": 613
+    },
+    {
+      "epoch": 0.00614,
+      "grad_norm": 0.648240484004962,
+      "learning_rate": 0.001842,
+      "loss": 5.0729,
+      "step": 614
+    },
+    {
+      "epoch": 0.00615,
+      "grad_norm": 0.6572100797643513,
+      "learning_rate": 0.001845,
+      "loss": 5.0607,
+      "step": 615
+    },
+    {
+      "epoch": 0.00616,
+      "grad_norm": 0.598653088729871,
+      "learning_rate": 0.001848,
+      "loss": 5.0431,
+      "step": 616
+    },
+    {
+      "epoch": 0.00617,
+      "grad_norm": 0.6867319038647199,
+      "learning_rate": 0.001851,
+      "loss": 5.058,
+      "step": 617
+    },
+    {
+      "epoch": 0.00618,
+      "grad_norm": 0.8065427766951909,
+      "learning_rate": 0.001854,
+      "loss": 5.0471,
+      "step": 618
+    },
+    {
+      "epoch": 0.00619,
+      "grad_norm": 0.952336607614539,
+      "learning_rate": 0.001857,
+      "loss": 5.0614,
+      "step": 619
+    },
+    {
+      "epoch": 0.0062,
+      "grad_norm": 0.937144329881144,
+      "learning_rate": 0.00186,
+      "loss": 5.0503,
+      "step": 620
+    },
+    {
+      "epoch": 0.00621,
+      "grad_norm": 0.861992409216581,
+      "learning_rate": 0.001863,
+      "loss": 5.0519,
+      "step": 621
+    },
+    {
+      "epoch": 0.00622,
+      "grad_norm": 0.7937303039626423,
+      "learning_rate": 0.001866,
+      "loss": 5.046,
+      "step": 622
+    },
+    {
+      "epoch": 0.00623,
+      "grad_norm": 0.7729108287648654,
+      "learning_rate": 0.001869,
+      "loss": 5.0315,
+      "step": 623
+    },
+    {
+      "epoch": 0.00624,
+      "grad_norm": 0.7644499866680247,
+      "learning_rate": 0.001872,
+      "loss": 5.0449,
+      "step": 624
+    },
+    {
+      "epoch": 0.00625,
+      "grad_norm": 0.833886296843224,
+      "learning_rate": 0.001875,
+      "loss": 5.0598,
+      "step": 625
+    },
+    {
+      "epoch": 0.00626,
+      "grad_norm": 0.8372671439007264,
+      "learning_rate": 0.0018780000000000001,
+      "loss": 5.0351,
+      "step": 626
+    },
+    {
+      "epoch": 0.00627,
+      "grad_norm": 0.8603722936304884,
+      "learning_rate": 0.001881,
+      "loss": 5.0306,
+      "step": 627
+    },
+    {
+      "epoch": 0.00628,
+      "grad_norm": 0.7718433014027412,
+      "learning_rate": 0.001884,
+      "loss": 5.0548,
+      "step": 628
+    },
+    {
+      "epoch": 0.00629,
+      "grad_norm": 0.6787638881219394,
+      "learning_rate": 0.001887,
+      "loss": 5.0263,
+      "step": 629
+    },
+    {
+      "epoch": 0.0063,
+      "grad_norm": 0.6841857077673691,
+      "learning_rate": 0.00189,
+      "loss": 5.0201,
+      "step": 630
+    },
+    {
+      "epoch": 0.00631,
+      "grad_norm": 0.7297958952205305,
+      "learning_rate": 0.0018930000000000002,
+      "loss": 5.0163,
+      "step": 631
+    },
+    {
+      "epoch": 0.00632,
+      "grad_norm": 0.8611511166687711,
+      "learning_rate": 0.0018960000000000001,
+      "loss": 5.0388,
+      "step": 632
+    },
+    {
+      "epoch": 0.00633,
+      "grad_norm": 0.9514927337678543,
+      "learning_rate": 0.001899,
+      "loss": 5.0438,
+      "step": 633
+    },
+    {
+      "epoch": 0.00634,
+      "grad_norm": 0.8956137982105657,
+      "learning_rate": 0.001902,
+      "loss": 5.031,
+      "step": 634
+    },
+    {
+      "epoch": 0.00635,
+      "grad_norm": 0.952460918530188,
+      "learning_rate": 0.001905,
+      "loss": 5.0218,
+      "step": 635
+    },
+    {
+      "epoch": 0.00636,
+      "grad_norm": 1.0747774477703274,
+      "learning_rate": 0.001908,
+      "loss": 5.0402,
+      "step": 636
+    },
+    {
+      "epoch": 0.00637,
+      "grad_norm": 0.8336825219104492,
+      "learning_rate": 0.0019110000000000002,
+      "loss": 5.0466,
+      "step": 637
+    },
+    {
+      "epoch": 0.00638,
+      "grad_norm": 0.8986808700713379,
+      "learning_rate": 0.0019140000000000001,
+      "loss": 5.0353,
+      "step": 638
+    },
+    {
+      "epoch": 0.00639,
+      "grad_norm": 1.0369332378334974,
+      "learning_rate": 0.001917,
+      "loss": 5.0213,
+      "step": 639
+    },
+    {
+      "epoch": 0.0064,
+      "grad_norm": 1.0511699002679498,
+      "learning_rate": 0.00192,
+      "loss": 5.0287,
+      "step": 640
+    },
+    {
+      "epoch": 0.00641,
+      "grad_norm": 1.0967175650217074,
+      "learning_rate": 0.001923,
+      "loss": 5.0426,
+      "step": 641
+    },
+    {
+      "epoch": 0.00642,
+      "grad_norm": 0.9933027226251017,
+      "learning_rate": 0.001926,
+      "loss": 5.0352,
+      "step": 642
+    },
+    {
+      "epoch": 0.00643,
+      "grad_norm": 0.9768777735178511,
+      "learning_rate": 0.0019290000000000002,
+      "loss": 5.0523,
+      "step": 643
+    },
+    {
+      "epoch": 0.00644,
+      "grad_norm": 1.001060805060802,
+      "learning_rate": 0.0019320000000000001,
+      "loss": 5.0363,
+      "step": 644
+    },
+    {
+      "epoch": 0.00645,
+      "grad_norm": 1.0724455418770003,
+      "learning_rate": 0.001935,
+      "loss": 5.0522,
+      "step": 645
+    },
+    {
+      "epoch": 0.00646,
+      "grad_norm": 1.1074452374796593,
+      "learning_rate": 0.001938,
+      "loss": 5.0611,
+      "step": 646
+    },
+    {
+      "epoch": 0.00647,
+      "grad_norm": 0.9675974316517354,
+      "learning_rate": 0.001941,
+      "loss": 5.062,
+      "step": 647
+    },
+    {
+      "epoch": 0.00648,
+      "grad_norm": 0.7269750089664031,
+      "learning_rate": 0.0019440000000000002,
+      "loss": 5.0382,
+      "step": 648
+    },
+    {
+      "epoch": 0.00649,
+      "grad_norm": 0.6773093371497811,
+      "learning_rate": 0.0019470000000000002,
+      "loss": 5.0209,
+      "step": 649
+    },
+    {
+      "epoch": 0.0065,
+      "grad_norm": 0.6721153823123059,
+      "learning_rate": 0.0019500000000000001,
+      "loss": 5.0458,
+      "step": 650
+    },
+    {
+      "epoch": 0.00651,
+      "grad_norm": 0.6257734701697928,
+      "learning_rate": 0.001953,
+      "loss": 5.022,
+      "step": 651
+    },
+    {
+      "epoch": 0.00652,
+      "grad_norm": 0.6412471052264174,
+      "learning_rate": 0.0019560000000000003,
+      "loss": 5.0157,
+      "step": 652
+    },
+    {
+      "epoch": 0.00653,
+      "grad_norm": 0.7030358108381808,
+      "learning_rate": 0.0019590000000000002,
+      "loss": 5.0108,
+      "step": 653
+    },
+    {
+      "epoch": 0.00654,
+      "grad_norm": 0.6417272703555769,
+      "learning_rate": 0.001962,
+      "loss": 4.994,
+      "step": 654
+    },
+    {
+      "epoch": 0.00655,
+      "grad_norm": 0.6123939930365004,
+      "learning_rate": 0.001965,
+      "loss": 5.0066,
+      "step": 655
+    },
+    {
+      "epoch": 0.00656,
+      "grad_norm": 0.6761646659152818,
+      "learning_rate": 0.001968,
+      "loss": 4.9975,
+      "step": 656
+    },
+    {
+      "epoch": 0.00657,
+      "grad_norm": 0.9015512681716132,
+      "learning_rate": 0.001971,
+      "loss": 4.9965,
+      "step": 657
+    },
+    {
+      "epoch": 0.00658,
+      "grad_norm": 1.1070883554163469,
+      "learning_rate": 0.001974,
+      "loss": 5.032,
+      "step": 658
+    },
+    {
+      "epoch": 0.00659,
+      "grad_norm": 0.8166390486888542,
+      "learning_rate": 0.001977,
+      "loss": 5.0022,
+      "step": 659
+    },
+    {
+      "epoch": 0.0066,
+      "grad_norm": 0.7886404692657704,
+      "learning_rate": 0.00198,
+      "loss": 4.9952,
+      "step": 660
+    },
+    {
+      "epoch": 0.00661,
+      "grad_norm": 0.6930157053874706,
+      "learning_rate": 0.001983,
+      "loss": 4.9751,
+      "step": 661
+    },
+    {
+      "epoch": 0.00662,
+      "grad_norm": 0.6021137246671374,
+      "learning_rate": 0.0019860000000000004,
+      "loss": 4.9738,
+      "step": 662
+    },
+    {
+      "epoch": 0.00663,
+      "grad_norm": 0.5342549595827694,
+      "learning_rate": 0.0019890000000000003,
+      "loss": 4.9858,
+      "step": 663
+    },
+    {
+      "epoch": 0.00664,
+      "grad_norm": 0.5549032292488858,
+      "learning_rate": 0.0019920000000000003,
+      "loss": 4.9668,
+      "step": 664
+    },
+    {
+      "epoch": 0.00665,
+      "grad_norm": 0.5630527450969944,
+      "learning_rate": 0.0019950000000000002,
+      "loss": 4.9437,
+      "step": 665
+    },
+    {
+      "epoch": 0.00666,
+      "grad_norm": 0.5436432724284425,
+      "learning_rate": 0.001998,
+      "loss": 4.9653,
+      "step": 666
+    },
+    {
+      "epoch": 0.00667,
+      "grad_norm": 0.5455501294317395,
+      "learning_rate": 0.002001,
+      "loss": 4.9611,
+      "step": 667
+    },
+    {
+      "epoch": 0.00668,
+      "grad_norm": 0.5771340900782089,
+      "learning_rate": 0.002004,
+      "loss": 4.9815,
+      "step": 668
+    },
+    {
+      "epoch": 0.00669,
+      "grad_norm": 0.5741027689717983,
+      "learning_rate": 0.002007,
+      "loss": 4.9757,
+      "step": 669
+    },
+    {
+      "epoch": 0.0067,
+      "grad_norm": 0.6058348461614704,
+      "learning_rate": 0.00201,
+      "loss": 4.969,
+      "step": 670
+    },
+    {
+      "epoch": 0.00671,
+      "grad_norm": 0.566759025441364,
+      "learning_rate": 0.002013,
+      "loss": 4.9434,
+      "step": 671
+    },
+    {
+      "epoch": 0.00672,
+      "grad_norm": 0.5204858890697072,
+      "learning_rate": 0.002016,
+      "loss": 4.9491,
+      "step": 672
+    },
+    {
+      "epoch": 0.00673,
+      "grad_norm": 0.5776654990466389,
+      "learning_rate": 0.002019,
+      "loss": 4.9473,
+      "step": 673
+    },
+    {
+      "epoch": 0.00674,
+      "grad_norm": 0.5899189939484214,
+      "learning_rate": 0.0020220000000000004,
+      "loss": 4.9512,
+      "step": 674
+    },
+    {
+      "epoch": 0.00675,
+      "grad_norm": 0.73017970913878,
+      "learning_rate": 0.0020250000000000003,
+      "loss": 4.954,
+      "step": 675
+    },
+    {
+      "epoch": 0.00676,
+      "grad_norm": 0.8714534217285366,
+      "learning_rate": 0.0020280000000000003,
+      "loss": 4.9527,
+      "step": 676
+    },
+    {
+      "epoch": 0.00677,
+      "grad_norm": 0.9455529231086871,
+      "learning_rate": 0.0020310000000000003,
+      "loss": 4.966,
+      "step": 677
+    },
+    {
+      "epoch": 0.00678,
+      "grad_norm": 0.7654589233287328,
+      "learning_rate": 0.0020340000000000002,
+      "loss": 4.9721,
+      "step": 678
+    },
+    {
+      "epoch": 0.00679,
+      "grad_norm": 0.7453542715563302,
+      "learning_rate": 0.002037,
+      "loss": 4.9514,
+      "step": 679
+    },
+    {
+      "epoch": 0.0068,
+      "grad_norm": 0.8198542095591255,
+      "learning_rate": 0.00204,
+      "loss": 4.9599,
+      "step": 680
+    },
+    {
+      "epoch": 0.00681,
+      "grad_norm": 1.0035327070670854,
+      "learning_rate": 0.002043,
+      "loss": 4.9608,
+      "step": 681
+    },
+    {
+      "epoch": 0.00682,
+      "grad_norm": 1.1272493939475496,
+      "learning_rate": 0.002046,
+      "loss": 4.9829,
+      "step": 682
+    },
+    {
+      "epoch": 0.00683,
+      "grad_norm": 0.9375943776902779,
+      "learning_rate": 0.002049,
+      "loss": 4.9573,
+      "step": 683
+    },
+    {
+      "epoch": 0.00684,
+      "grad_norm": 0.988255846210588,
+      "learning_rate": 0.002052,
+      "loss": 4.9739,
+      "step": 684
+    },
+    {
+      "epoch": 0.00685,
+      "grad_norm": 1.2166901935378913,
+      "learning_rate": 0.0020550000000000004,
+      "loss": 4.9877,
+      "step": 685
+    },
+    {
+      "epoch": 0.00686,
+      "grad_norm": 0.8834239063706651,
+      "learning_rate": 0.0020580000000000004,
+      "loss": 4.9625,
+      "step": 686
+    },
+    {
+      "epoch": 0.00687,
+      "grad_norm": 0.9980240769781185,
+      "learning_rate": 0.0020610000000000003,
+      "loss": 4.9587,
+      "step": 687
+    },
+    {
+      "epoch": 0.00688,
+      "grad_norm": 1.1254414916617237,
+      "learning_rate": 0.002064,
+      "loss": 4.9804,
+      "step": 688
+    },
+    {
+      "epoch": 0.00689,
+      "grad_norm": 1.0306142304378458,
+      "learning_rate": 0.002067,
+      "loss": 4.9703,
+      "step": 689
+    },
+    {
+      "epoch": 0.0069,
+      "grad_norm": 1.1841729949788287,
+      "learning_rate": 0.00207,
+      "loss": 4.9832,
+      "step": 690
+    },
+    {
+      "epoch": 0.00691,
+      "grad_norm": 0.94703321985356,
+      "learning_rate": 0.0020729999999999998,
+      "loss": 4.946,
+      "step": 691
+    },
+    {
+      "epoch": 0.00692,
+      "grad_norm": 1.0500733994835558,
+      "learning_rate": 0.0020759999999999997,
+      "loss": 4.9769,
+      "step": 692
+    },
+    {
+      "epoch": 0.00693,
+      "grad_norm": 0.8703789610160195,
+      "learning_rate": 0.0020789999999999997,
+      "loss": 4.9707,
+      "step": 693
+    },
+    {
+      "epoch": 0.00694,
+      "grad_norm": 0.8616077913099822,
+      "learning_rate": 0.002082,
+      "loss": 4.9464,
+      "step": 694
+    },
+    {
+      "epoch": 0.00695,
+      "grad_norm": 1.1391889561194677,
+      "learning_rate": 0.002085,
+      "loss": 4.9626,
+      "step": 695
+    },
+    {
+      "epoch": 0.00696,
+      "grad_norm": 0.9584412207038412,
+      "learning_rate": 0.002088,
+      "loss": 4.9606,
+      "step": 696
+    },
+    {
+      "epoch": 0.00697,
+      "grad_norm": 1.0011041706204022,
+      "learning_rate": 0.002091,
+      "loss": 4.9624,
+      "step": 697
+    },
+    {
+      "epoch": 0.00698,
+      "grad_norm": 1.3127609957868622,
+      "learning_rate": 0.002094,
+      "loss": 4.9728,
+      "step": 698
+    },
+    {
+      "epoch": 0.00699,
+      "grad_norm": 0.8537920758308664,
+      "learning_rate": 0.002097,
+      "loss": 4.9491,
+      "step": 699
+    },
+    {
+      "epoch": 0.007,
+      "grad_norm": 1.0018321540367086,
+      "learning_rate": 0.0021,
+      "loss": 4.9652,
+      "step": 700
+    },
+    {
+      "epoch": 0.00701,
+      "grad_norm": 0.9481035749179906,
+      "learning_rate": 0.002103,
+      "loss": 4.9422,
+      "step": 701
+    },
+    {
+      "epoch": 0.00702,
+      "grad_norm": 0.8258463386850557,
+      "learning_rate": 0.002106,
+      "loss": 4.945,
+      "step": 702
+    },
+    {
+      "epoch": 0.00703,
+      "grad_norm": 0.6305963851301584,
+      "learning_rate": 0.0021089999999999998,
+      "loss": 4.9167,
+      "step": 703
+    },
+    {
+      "epoch": 0.00704,
+      "grad_norm": 0.6428932601365419,
+      "learning_rate": 0.0021119999999999997,
+      "loss": 4.9232,
+      "step": 704
+    },
+    {
+      "epoch": 0.00705,
+      "grad_norm": 0.6965384868722897,
+      "learning_rate": 0.002115,
+      "loss": 4.9266,
+      "step": 705
+    },
+    {
+      "epoch": 0.00706,
+      "grad_norm": 0.7410784964260257,
+      "learning_rate": 0.002118,
+      "loss": 4.9344,
+      "step": 706
+    },
+    {
+      "epoch": 0.00707,
+      "grad_norm": 0.721339398040246,
+      "learning_rate": 0.002121,
+      "loss": 4.9083,
+      "step": 707
+    },
+    {
+      "epoch": 0.00708,
+      "grad_norm": 0.675102725576719,
+      "learning_rate": 0.002124,
+      "loss": 4.8875,
+      "step": 708
+    },
+    {
+      "epoch": 0.00709,
+      "grad_norm": 0.6428565761724286,
+      "learning_rate": 0.002127,
+      "loss": 4.9151,
+      "step": 709
+    },
+    {
+      "epoch": 0.0071,
+      "grad_norm": 0.5795546215215135,
+      "learning_rate": 0.00213,
+      "loss": 4.8959,
+      "step": 710
+    },
+    {
+      "epoch": 0.00711,
+      "grad_norm": 0.6667417841414063,
+      "learning_rate": 0.002133,
+      "loss": 4.8739,
+      "step": 711
+    },
+    {
+      "epoch": 0.00712,
+      "grad_norm": 0.7094343142489271,
+      "learning_rate": 0.002136,
+      "loss": 4.8945,
+      "step": 712
+    },
+    {
+      "epoch": 0.00713,
+      "grad_norm": 0.6503826772007358,
+      "learning_rate": 0.002139,
+      "loss": 4.9061,
+      "step": 713
+    },
+    {
+      "epoch": 0.00714,
+      "grad_norm": 0.5916220730293257,
+      "learning_rate": 0.002142,
+      "loss": 4.8928,
+      "step": 714
+    },
+    {
+      "epoch": 0.00715,
+      "grad_norm": 0.6248268778602033,
+      "learning_rate": 0.0021449999999999998,
+      "loss": 4.8651,
+      "step": 715
+    },
+    {
+      "epoch": 0.00716,
+      "grad_norm": 0.7056512941727522,
+      "learning_rate": 0.002148,
+      "loss": 4.8619,
+      "step": 716
+    },
+    {
+      "epoch": 0.00717,
+      "grad_norm": 0.7609530945312164,
+      "learning_rate": 0.002151,
+      "loss": 4.8823,
+      "step": 717
+    },
+    {
+      "epoch": 0.00718,
+      "grad_norm": 0.7494152535556156,
+      "learning_rate": 0.002154,
+      "loss": 4.8579,
+      "step": 718
+    },
+    {
+      "epoch": 0.00719,
+      "grad_norm": 0.7503709346262388,
+      "learning_rate": 0.002157,
+      "loss": 4.8753,
+      "step": 719
+    },
+    {
+      "epoch": 0.0072,
+      "grad_norm": 0.7326012456404516,
+      "learning_rate": 0.00216,
+      "loss": 4.8631,
+      "step": 720
+    },
+    {
+      "epoch": 0.00721,
+      "grad_norm": 0.7261591024591523,
+      "learning_rate": 0.002163,
+      "loss": 4.867,
+      "step": 721
+    },
+    {
+      "epoch": 0.00722,
+      "grad_norm": 0.782783835861699,
+      "learning_rate": 0.002166,
+      "loss": 4.843,
+      "step": 722
+    },
+    {
+      "epoch": 0.00723,
+      "grad_norm": 0.9405574246546834,
+      "learning_rate": 0.002169,
+      "loss": 4.859,
+      "step": 723
+    },
+    {
+      "epoch": 0.00724,
+      "grad_norm": 0.9635066655828881,
+      "learning_rate": 0.002172,
+      "loss": 4.8537,
+      "step": 724
+    },
+    {
+      "epoch": 0.00725,
+      "grad_norm": 1.040794527733107,
+      "learning_rate": 0.002175,
+      "loss": 4.8644,
+      "step": 725
+    },
+    {
+      "epoch": 0.00726,
+      "grad_norm": 0.8259351691267115,
+      "learning_rate": 0.002178,
+      "loss": 4.8549,
+      "step": 726
+    },
+    {
+      "epoch": 0.00727,
+      "grad_norm": 0.9246523356382073,
+      "learning_rate": 0.0021809999999999998,
+      "loss": 4.8481,
+      "step": 727
+    },
+    {
+      "epoch": 0.00728,
+      "grad_norm": 0.8492007448518646,
+      "learning_rate": 0.002184,
+      "loss": 4.8551,
+      "step": 728
+    },
+    {
+      "epoch": 0.00729,
+      "grad_norm": 0.8802732367664233,
+      "learning_rate": 0.002187,
+      "loss": 4.8519,
+      "step": 729
+    },
+    {
+      "epoch": 0.0073,
+      "grad_norm": 0.8143690086226398,
+      "learning_rate": 0.00219,
+      "loss": 4.8583,
+      "step": 730
+    },
+    {
+      "epoch": 0.00731,
+      "grad_norm": 0.7978098404882568,
+      "learning_rate": 0.002193,
+      "loss": 4.846,
+      "step": 731
+    },
+    {
+      "epoch": 0.00732,
+      "grad_norm": 0.8726405483808616,
+      "learning_rate": 0.002196,
+      "loss": 4.8538,
+      "step": 732
+    },
+    {
+      "epoch": 0.00733,
+      "grad_norm": 0.852624212596407,
+      "learning_rate": 0.002199,
+      "loss": 4.8348,
+      "step": 733
+    },
+    {
+      "epoch": 0.00734,
+      "grad_norm": 0.8516024625713395,
+      "learning_rate": 0.002202,
+      "loss": 4.8503,
+      "step": 734
+    },
+    {
+      "epoch": 0.00735,
+      "grad_norm": 1.0296852244953179,
+      "learning_rate": 0.002205,
+      "loss": 4.8752,
+      "step": 735
+    },
+    {
+      "epoch": 0.00736,
+      "grad_norm": 0.8907291733705317,
+      "learning_rate": 0.002208,
+      "loss": 4.8478,
+      "step": 736
+    },
+    {
+      "epoch": 0.00737,
+      "grad_norm": 0.847923173569621,
+      "learning_rate": 0.002211,
+      "loss": 4.8138,
+      "step": 737
+    },
+    {
+      "epoch": 0.00738,
+      "grad_norm": 1.0201562870165168,
+      "learning_rate": 0.002214,
+      "loss": 4.8304,
+      "step": 738
+    },
+    {
+      "epoch": 0.00739,
+      "grad_norm": 0.8864771113941665,
+      "learning_rate": 0.0022170000000000002,
+      "loss": 4.8243,
+      "step": 739
+    },
+    {
+      "epoch": 0.0074,
+      "grad_norm": 0.8107829600567034,
+      "learning_rate": 0.00222,
+      "loss": 4.8172,
+      "step": 740
+    },
+    {
+      "epoch": 0.00741,
+      "grad_norm": 0.9926562607182313,
+      "learning_rate": 0.002223,
+      "loss": 4.7981,
+      "step": 741
+    },
+    {
+      "epoch": 0.00742,
+      "grad_norm": 1.1169095501625559,
+      "learning_rate": 0.002226,
+      "loss": 4.8325,
+      "step": 742
+    },
+    {
+      "epoch": 0.00743,
+      "grad_norm": 1.0154903575297356,
+      "learning_rate": 0.002229,
+      "loss": 4.8033,
+      "step": 743
+    },
+    {
+      "epoch": 0.00744,
+      "grad_norm": 0.8796310645612629,
+      "learning_rate": 0.002232,
+      "loss": 4.8021,
+      "step": 744
+    },
+    {
+      "epoch": 0.00745,
+      "grad_norm": 0.8347673276949072,
+      "learning_rate": 0.002235,
+      "loss": 4.8034,
+      "step": 745
+    },
+    {
+      "epoch": 0.00746,
+      "grad_norm": 1.0163090629032976,
+      "learning_rate": 0.002238,
+      "loss": 4.8122,
+      "step": 746
+    },
+    {
+      "epoch": 0.00747,
+      "grad_norm": 0.9716623005872416,
+      "learning_rate": 0.002241,
+      "loss": 4.7984,
+      "step": 747
+    },
+    {
+      "epoch": 0.00748,
+      "grad_norm": 1.0278655828642824,
+      "learning_rate": 0.002244,
+      "loss": 4.8089,
+      "step": 748
+    },
+    {
+      "epoch": 0.00749,
+      "grad_norm": 1.0480541445842007,
+      "learning_rate": 0.002247,
+      "loss": 4.797,
+      "step": 749
+    },
+    {
+      "epoch": 0.0075,
+      "grad_norm": 0.8547105592298554,
+      "learning_rate": 0.0022500000000000003,
+      "loss": 4.8161,
+      "step": 750
+    },
+    {
+      "epoch": 0.00751,
+      "grad_norm": 1.124867997489978,
+      "learning_rate": 0.0022530000000000002,
+      "loss": 4.8181,
+      "step": 751
+    },
+    {
+      "epoch": 0.00752,
+      "grad_norm": 0.9791690369641056,
+      "learning_rate": 0.002256,
+      "loss": 4.777,
+      "step": 752
+    },
+    {
+      "epoch": 0.00753,
+      "grad_norm": 1.076442658365417,
+      "learning_rate": 0.002259,
+      "loss": 4.8205,
+      "step": 753
+    },
+    {
+      "epoch": 0.00754,
+      "grad_norm": 1.2763721488952993,
+      "learning_rate": 0.002262,
+      "loss": 4.8438,
+      "step": 754
+    },
+    {
+      "epoch": 0.00755,
+      "grad_norm": 0.956893666746658,
+      "learning_rate": 0.002265,
+      "loss": 4.7969,
+      "step": 755
+    },
+    {
+      "epoch": 0.00756,
+      "grad_norm": 0.9171753465968743,
+      "learning_rate": 0.002268,
+      "loss": 4.7762,
+      "step": 756
+    },
+    {
+      "epoch": 0.00757,
+      "grad_norm": 1.0908406750664337,
+      "learning_rate": 0.002271,
+      "loss": 4.7959,
+      "step": 757
+    },
+    {
+      "epoch": 0.00758,
+      "grad_norm": 0.9176733613061218,
+      "learning_rate": 0.002274,
+      "loss": 4.7709,
+      "step": 758
+    },
+    {
+      "epoch": 0.00759,
+      "grad_norm": 0.8908245026099332,
+      "learning_rate": 0.002277,
+      "loss": 4.7862,
+      "step": 759
+    },
+    {
+      "epoch": 0.0076,
+      "grad_norm": 0.9024826649233609,
+      "learning_rate": 0.00228,
+      "loss": 4.7716,
+      "step": 760
+    },
+    {
+      "epoch": 0.00761,
+      "grad_norm": 0.8163892030302853,
+      "learning_rate": 0.002283,
+      "loss": 4.7408,
+      "step": 761
+    },
+    {
+      "epoch": 0.00762,
+      "grad_norm": 0.8370737395441165,
+      "learning_rate": 0.0022860000000000003,
+      "loss": 4.7247,
+      "step": 762
+    },
+    {
+      "epoch": 0.00763,
+      "grad_norm": 0.8480960736158745,
+      "learning_rate": 0.0022890000000000002,
+      "loss": 4.7641,
+      "step": 763
+    },
+    {
+      "epoch": 0.00764,
+      "grad_norm": 0.7323677069229925,
+      "learning_rate": 0.002292,
+      "loss": 4.7635,
+      "step": 764
+    },
+    {
+      "epoch": 0.00765,
+      "grad_norm": 0.517428476031845,
+      "learning_rate": 0.002295,
+      "loss": 4.7402,
+      "step": 765
+    },
+    {
+      "epoch": 0.00766,
+      "grad_norm": 0.5449748434883583,
+      "learning_rate": 0.002298,
+      "loss": 4.7349,
+      "step": 766
+    },
+    {
+      "epoch": 0.00767,
+      "grad_norm": 0.519905380789997,
+      "learning_rate": 0.002301,
+      "loss": 4.7297,
+      "step": 767
+    },
+    {
+      "epoch": 0.00768,
+      "grad_norm": 0.5353014095205901,
+      "learning_rate": 0.002304,
+      "loss": 4.7326,
+      "step": 768
+    },
+    {
+      "epoch": 0.00769,
+      "grad_norm": 0.6021476092106454,
+      "learning_rate": 0.002307,
+      "loss": 4.7369,
+      "step": 769
+    },
+    {
+      "epoch": 0.0077,
+      "grad_norm": 0.6291707943467315,
+      "learning_rate": 0.00231,
+      "loss": 4.6951,
+      "step": 770
+    },
+    {
+      "epoch": 0.00771,
+      "grad_norm": 0.6259748544734598,
+      "learning_rate": 0.002313,
+      "loss": 4.726,
+      "step": 771
+    },
+    {
+      "epoch": 0.00772,
+      "grad_norm": 0.6411937087975925,
+      "learning_rate": 0.002316,
+      "loss": 4.71,
+      "step": 772
+    },
+    {
+      "epoch": 0.00773,
+      "grad_norm": 0.6087209583882061,
+      "learning_rate": 0.0023190000000000003,
+      "loss": 4.7089,
+      "step": 773
+    },
+    {
+      "epoch": 0.00774,
+      "grad_norm": 0.6187878154543281,
+      "learning_rate": 0.0023220000000000003,
+      "loss": 4.7216,
+      "step": 774
+    },
+    {
+      "epoch": 0.00775,
+      "grad_norm": 0.6698244521960122,
+      "learning_rate": 0.0023250000000000002,
+      "loss": 4.7125,
+      "step": 775
+    },
+    {
+      "epoch": 0.00776,
+      "grad_norm": 0.681313335099601,
+      "learning_rate": 0.002328,
+      "loss": 4.716,
+      "step": 776
+    },
+    {
+      "epoch": 0.00777,
+      "grad_norm": 0.564623702064127,
+      "learning_rate": 0.002331,
+      "loss": 4.6866,
+      "step": 777
+    },
+    {
+      "epoch": 0.00778,
+      "grad_norm": 0.5709713255684672,
+      "learning_rate": 0.002334,
+      "loss": 4.6954,
+      "step": 778
+    },
+    {
+      "epoch": 0.00779,
+      "grad_norm": 0.6149821587836622,
+      "learning_rate": 0.002337,
+      "loss": 4.6565,
+      "step": 779
+    },
+    {
+      "epoch": 0.0078,
+      "grad_norm": 0.632842020953678,
+      "learning_rate": 0.00234,
+      "loss": 4.6725,
+      "step": 780
+    },
+    {
+      "epoch": 0.00781,
+      "grad_norm": 0.7467970568022195,
+      "learning_rate": 0.002343,
+      "loss": 4.678,
+      "step": 781
+    },
+    {
+      "epoch": 0.00782,
+      "grad_norm": 0.9047857384333676,
+      "learning_rate": 0.002346,
+      "loss": 4.6801,
+      "step": 782
+    },
+    {
+      "epoch": 0.00783,
+      "grad_norm": 1.0576276659724262,
+      "learning_rate": 0.002349,
+      "loss": 4.6678,
+      "step": 783
+    },
+    {
+      "epoch": 0.00784,
+      "grad_norm": 1.1751809929780095,
+      "learning_rate": 0.002352,
+      "loss": 4.6921,
+      "step": 784
+    },
+    {
+      "epoch": 0.00785,
+      "grad_norm": 0.8865561583369859,
+      "learning_rate": 0.0023550000000000003,
+      "loss": 4.6774,
+      "step": 785
+    },
+    {
+      "epoch": 0.00786,
+      "grad_norm": 1.1372694124786462,
+      "learning_rate": 0.0023580000000000003,
+      "loss": 4.7016,
+      "step": 786
+    },
+    {
+      "epoch": 0.00787,
+      "grad_norm": 1.0092158733190273,
+      "learning_rate": 0.0023610000000000003,
+      "loss": 4.6918,
+      "step": 787
+    },
+    {
+      "epoch": 0.00788,
+      "grad_norm": 1.1115636173510615,
+      "learning_rate": 0.002364,
+      "loss": 4.7036,
+      "step": 788
+    },
+    {
+      "epoch": 0.00789,
+      "grad_norm": 0.9627507497820444,
+      "learning_rate": 0.002367,
+      "loss": 4.7196,
+      "step": 789
+    },
+    {
+      "epoch": 0.0079,
+      "grad_norm": 1.0391564883926698,
+      "learning_rate": 0.00237,
+      "loss": 4.706,
+      "step": 790
+    },
+    {
+      "epoch": 0.00791,
+      "grad_norm": 0.9857967273862618,
+      "learning_rate": 0.002373,
+      "loss": 4.6613,
+      "step": 791
+    },
+    {
+      "epoch": 0.00792,
+      "grad_norm": 1.07807376567954,
+      "learning_rate": 0.002376,
+      "loss": 4.6969,
+      "step": 792
+    },
+    {
+      "epoch": 0.00793,
+      "grad_norm": 1.1762161150555115,
+      "learning_rate": 0.002379,
+      "loss": 4.7002,
+      "step": 793
+    },
+    {
+      "epoch": 0.00794,
+      "grad_norm": 0.8247319634032979,
+      "learning_rate": 0.002382,
+      "loss": 4.6948,
+      "step": 794
+    },
+    {
+      "epoch": 0.00795,
+      "grad_norm": 0.8498192334201926,
+      "learning_rate": 0.002385,
+      "loss": 4.676,
+      "step": 795
+    },
+    {
+      "epoch": 0.00796,
+      "grad_norm": 0.7871292688998256,
+      "learning_rate": 0.0023880000000000004,
+      "loss": 4.6606,
+      "step": 796
+    },
+    {
+      "epoch": 0.00797,
+      "grad_norm": 0.7600244620970749,
+      "learning_rate": 0.0023910000000000003,
+      "loss": 4.7021,
+      "step": 797
+    },
+    {
+      "epoch": 0.00798,
+      "grad_norm": 0.7960964641914068,
+      "learning_rate": 0.0023940000000000003,
+      "loss": 4.6698,
+      "step": 798
+    },
+    {
+      "epoch": 0.00799,
+      "grad_norm": 0.740030757660218,
+      "learning_rate": 0.0023970000000000003,
+      "loss": 4.6941,
+      "step": 799
+    },
+    {
+      "epoch": 0.008,
+      "grad_norm": 0.7620304314170098,
+      "learning_rate": 0.0024000000000000002,
+      "loss": 4.6524,
+      "step": 800
+    },
+    {
+      "epoch": 0.00801,
+      "grad_norm": 0.7109641363313225,
+      "learning_rate": 0.002403,
+      "loss": 4.6746,
+      "step": 801
+    },
+    {
+      "epoch": 0.00802,
+      "grad_norm": 0.6915734125547375,
+      "learning_rate": 0.002406,
+      "loss": 4.6614,
+      "step": 802
+    },
+    {
+      "epoch": 0.00803,
+      "grad_norm": 0.6786740334805601,
+      "learning_rate": 0.002409,
+      "loss": 4.6408,
+      "step": 803
+    },
+    {
+      "epoch": 0.00804,
+      "grad_norm": 0.6631767318006158,
+      "learning_rate": 0.002412,
+      "loss": 4.6361,
+      "step": 804
+    },
+    {
+      "epoch": 0.00805,
+      "grad_norm": 0.6460763714070807,
+      "learning_rate": 0.002415,
+      "loss": 4.631,
+      "step": 805
+    },
+    {
+      "epoch": 0.00806,
+      "grad_norm": 0.5689446567265103,
+      "learning_rate": 0.002418,
+      "loss": 4.6413,
+      "step": 806
+    },
+    {
+      "epoch": 0.00807,
+      "grad_norm": 0.7294155919129667,
+      "learning_rate": 0.0024210000000000004,
+      "loss": 4.6407,
+      "step": 807
+    },
+    {
+      "epoch": 0.00808,
+      "grad_norm": 0.9713010260235329,
+      "learning_rate": 0.0024240000000000004,
+      "loss": 4.6372,
+      "step": 808
+    },
+    {
+      "epoch": 0.00809,
+      "grad_norm": 1.039651350431828,
+      "learning_rate": 0.0024270000000000003,
+      "loss": 4.6452,
+      "step": 809
+    },
+    {
+      "epoch": 0.0081,
+      "grad_norm": 0.7987971948237619,
+      "learning_rate": 0.0024300000000000003,
+      "loss": 4.6558,
+      "step": 810
+    },
+    {
+      "epoch": 0.00811,
+      "grad_norm": 0.866506786468452,
+      "learning_rate": 0.0024330000000000003,
+      "loss": 4.6332,
+      "step": 811
+    },
+    {
+      "epoch": 0.00812,
+      "grad_norm": 0.8940070034144846,
+      "learning_rate": 0.0024360000000000002,
+      "loss": 4.6162,
+      "step": 812
+    },
+    {
+      "epoch": 0.00813,
+      "grad_norm": 0.7601015395596301,
+      "learning_rate": 0.0024389999999999998,
+      "loss": 4.6418,
+      "step": 813
+    },
+    {
+      "epoch": 0.00814,
+      "grad_norm": 0.959371802874838,
+      "learning_rate": 0.0024419999999999997,
+      "loss": 4.6561,
+      "step": 814
+    },
+    {
+      "epoch": 0.00815,
+      "grad_norm": 1.0501943727451555,
+      "learning_rate": 0.0024449999999999997,
+      "loss": 4.6758,
+      "step": 815
+    },
+    {
+      "epoch": 0.00816,
+      "grad_norm": 1.0657571457398118,
+      "learning_rate": 0.002448,
+      "loss": 4.6479,
+      "step": 816
+    },
+    {
+      "epoch": 0.00817,
+      "grad_norm": 0.9466339757774411,
+      "learning_rate": 0.002451,
+      "loss": 4.6581,
+      "step": 817
+    },
+    {
+      "epoch": 0.00818,
+      "grad_norm": 1.0180120345579544,
+      "learning_rate": 0.002454,
+      "loss": 4.6439,
+      "step": 818
+    },
+    {
+      "epoch": 0.00819,
+      "grad_norm": 0.9541848583096257,
+      "learning_rate": 0.002457,
+      "loss": 4.6558,
+      "step": 819
+    },
+    {
+      "epoch": 0.0082,
+      "grad_norm": 0.8369907334148766,
+      "learning_rate": 0.00246,
+      "loss": 4.6334,
+      "step": 820
+    },
+    {
+      "epoch": 0.00821,
+      "grad_norm": 0.9636511535948165,
+      "learning_rate": 0.002463,
+      "loss": 4.6201,
+      "step": 821
+    },
+    {
+      "epoch": 0.00822,
+      "grad_norm": 0.9781201110522014,
+      "learning_rate": 0.002466,
+      "loss": 4.6234,
+      "step": 822
+    },
+    {
+      "epoch": 0.00823,
+      "grad_norm": 0.7544698826164155,
+      "learning_rate": 0.002469,
+      "loss": 4.6177,
+      "step": 823
+    },
+    {
+      "epoch": 0.00824,
+      "grad_norm": 0.8351046210351873,
+      "learning_rate": 0.002472,
+      "loss": 4.6395,
+      "step": 824
+    },
+    {
+      "epoch": 0.00825,
+      "grad_norm": 0.9579203449788568,
+      "learning_rate": 0.0024749999999999998,
+      "loss": 4.6552,
+      "step": 825
+    },
+    {
+      "epoch": 0.00826,
+      "grad_norm": 0.6855903524790949,
+      "learning_rate": 0.0024779999999999997,
+      "loss": 4.6015,
+      "step": 826
+    },
+    {
+      "epoch": 0.00827,
+      "grad_norm": 0.5831382107833026,
+      "learning_rate": 0.002481,
+      "loss": 4.6271,
+      "step": 827
+    },
+    {
+      "epoch": 0.00828,
+      "grad_norm": 0.6353397555552095,
+      "learning_rate": 0.002484,
+      "loss": 4.6166,
+      "step": 828
+    },
+    {
+      "epoch": 0.00829,
+      "grad_norm": 0.7710602439494972,
+      "learning_rate": 0.002487,
+      "loss": 4.619,
+      "step": 829
+    },
+    {
+      "epoch": 0.0083,
+      "grad_norm": 0.8708544431374072,
+      "learning_rate": 0.00249,
+      "loss": 4.6284,
+      "step": 830
+    },
+    {
+      "epoch": 0.00831,
+      "grad_norm": 1.0453965047267408,
+      "learning_rate": 0.002493,
+      "loss": 4.6196,
+      "step": 831
+    },
+    {
+      "epoch": 0.00832,
+      "grad_norm": 1.1447270695034462,
+      "learning_rate": 0.002496,
+      "loss": 4.6327,
+      "step": 832
+    },
+    {
+      "epoch": 0.00833,
+      "grad_norm": 1.0246810397019857,
+      "learning_rate": 0.002499,
+      "loss": 4.6176,
+      "step": 833
+    },
+    {
+      "epoch": 0.00834,
+      "grad_norm": 0.9810039542594958,
+      "learning_rate": 0.002502,
+      "loss": 4.6274,
+      "step": 834
+    },
+    {
+      "epoch": 0.00835,
+      "grad_norm": 0.8876541499902995,
+      "learning_rate": 0.002505,
+      "loss": 4.6197,
+      "step": 835
+    },
+    {
+      "epoch": 0.00836,
+      "grad_norm": 0.903016402602156,
+      "learning_rate": 0.002508,
+      "loss": 4.6457,
+      "step": 836
+    },
+    {
+      "epoch": 0.00837,
+      "grad_norm": 0.9089070943446178,
+      "learning_rate": 0.0025109999999999998,
+      "loss": 4.6001,
+      "step": 837
+    },
+    {
+      "epoch": 0.00838,
+      "grad_norm": 1.0056159571167094,
+      "learning_rate": 0.0025139999999999997,
+      "loss": 4.6276,
+      "step": 838
+    },
+    {
+      "epoch": 0.00839,
+      "grad_norm": 0.9338401740679543,
+      "learning_rate": 0.002517,
+      "loss": 4.6258,
+      "step": 839
+    },
+    {
+      "epoch": 0.0084,
+      "grad_norm": 0.8595710991386702,
+      "learning_rate": 0.00252,
+      "loss": 4.6076,
+      "step": 840
+    },
+    {
+      "epoch": 0.00841,
+      "grad_norm": 0.8408169361023179,
+      "learning_rate": 0.002523,
+      "loss": 4.5981,
+      "step": 841
+    },
+    {
+      "epoch": 0.00842,
+      "grad_norm": 0.9223461051136856,
+      "learning_rate": 0.002526,
+      "loss": 4.6018,
+      "step": 842
+    },
+    {
+      "epoch": 0.00843,
+      "grad_norm": 0.8967298269444864,
+      "learning_rate": 0.002529,
+      "loss": 4.5992,
+      "step": 843
+    },
+    {
+      "epoch": 0.00844,
+      "grad_norm": 1.0054950179651305,
+      "learning_rate": 0.002532,
+      "loss": 4.6384,
+      "step": 844
+    },
+    {
+      "epoch": 0.00845,
+      "grad_norm": 0.9421873102957943,
+      "learning_rate": 0.002535,
+      "loss": 4.597,
+      "step": 845
+    },
+    {
+      "epoch": 0.00846,
+      "grad_norm": 0.7813468730903476,
+      "learning_rate": 0.002538,
+      "loss": 4.612,
+      "step": 846
+    },
+    {
+      "epoch": 0.00847,
+      "grad_norm": 0.7336538825841323,
+      "learning_rate": 0.002541,
+      "loss": 4.5922,
+      "step": 847
+    },
+    {
+      "epoch": 0.00848,
+      "grad_norm": 0.7490910216536115,
+      "learning_rate": 0.002544,
+      "loss": 4.5881,
+      "step": 848
+    },
+    {
+      "epoch": 0.00849,
+      "grad_norm": 0.7281767237417216,
+      "learning_rate": 0.002547,
+      "loss": 4.5791,
+      "step": 849
+    },
+    {
+      "epoch": 0.0085,
+      "grad_norm": 0.6108432594850526,
+      "learning_rate": 0.00255,
+      "loss": 4.5847,
+      "step": 850
+    },
+    {
+      "epoch": 0.00851,
+      "grad_norm": 0.6024151823798927,
+      "learning_rate": 0.002553,
+      "loss": 4.5988,
+      "step": 851
+    },
+    {
+      "epoch": 0.00852,
+      "grad_norm": 0.4780185378158163,
+      "learning_rate": 0.002556,
+      "loss": 4.5828,
+      "step": 852
+    },
+    {
+      "epoch": 0.00853,
+      "grad_norm": 0.5435830639563773,
+      "learning_rate": 0.002559,
+      "loss": 4.5614,
+      "step": 853
+    },
+    {
+      "epoch": 0.00854,
+      "grad_norm": 0.6028384776402124,
+      "learning_rate": 0.002562,
+      "loss": 4.5542,
+      "step": 854
+    },
+    {
+      "epoch": 0.00855,
+      "grad_norm": 0.5885480983638159,
+      "learning_rate": 0.002565,
+      "loss": 4.5306,
+      "step": 855
+    },
+    {
+      "epoch": 0.00856,
+      "grad_norm": 0.6060410896547319,
+      "learning_rate": 0.002568,
+      "loss": 4.5634,
+      "step": 856
+    },
+    {
+      "epoch": 0.00857,
+      "grad_norm": 0.5609885323605243,
+      "learning_rate": 0.002571,
+      "loss": 4.5495,
+      "step": 857
+    },
+    {
+      "epoch": 0.00858,
+      "grad_norm": 0.5199675893089152,
+      "learning_rate": 0.002574,
+      "loss": 4.541,
+      "step": 858
+    },
+    {
+      "epoch": 0.00859,
+      "grad_norm": 0.5200210947429836,
+      "learning_rate": 0.002577,
+      "loss": 4.5318,
+      "step": 859
+    },
+    {
+      "epoch": 0.0086,
+      "grad_norm": 0.5623426902483657,
+      "learning_rate": 0.00258,
+      "loss": 4.5597,
+      "step": 860
+    },
+    {
+      "epoch": 0.00861,
+      "grad_norm": 0.6385070114593622,
+      "learning_rate": 0.0025830000000000002,
+      "loss": 4.5763,
+      "step": 861
+    },
+    {
+      "epoch": 0.00862,
+      "grad_norm": 0.7215967762933786,
+      "learning_rate": 0.002586,
+      "loss": 4.5292,
+      "step": 862
+    },
+    {
+      "epoch": 0.00863,
+      "grad_norm": 0.8778870896923412,
+      "learning_rate": 0.002589,
+      "loss": 4.559,
+      "step": 863
+    },
+    {
+      "epoch": 0.00864,
+      "grad_norm": 1.1440074056398795,
+      "learning_rate": 0.002592,
+      "loss": 4.547,
+      "step": 864
+    },
+    {
+      "epoch": 0.00865,
+      "grad_norm": 0.9219223071258862,
+      "learning_rate": 0.002595,
+      "loss": 4.5491,
+      "step": 865
+    },
+    {
+      "epoch": 0.00866,
+      "grad_norm": 1.065047382603863,
+      "learning_rate": 0.002598,
+      "loss": 4.5845,
+      "step": 866
+    },
+    {
+      "epoch": 0.00867,
+      "grad_norm": 0.9527212983309642,
+      "learning_rate": 0.002601,
+      "loss": 4.5736,
+      "step": 867
+    },
+    {
+      "epoch": 0.00868,
+      "grad_norm": 0.7943624191664302,
+      "learning_rate": 0.002604,
+      "loss": 4.5474,
+      "step": 868
+    },
+    {
+      "epoch": 0.00869,
+      "grad_norm": 0.671885521142916,
+      "learning_rate": 0.002607,
+      "loss": 4.5536,
+      "step": 869
+    },
+    {
+      "epoch": 0.0087,
+      "grad_norm": 0.6527559883208905,
+      "learning_rate": 0.00261,
+      "loss": 4.54,
+      "step": 870
+    },
+    {
+      "epoch": 0.00871,
+      "grad_norm": 0.6266922867904314,
+      "learning_rate": 0.002613,
+      "loss": 4.5522,
+      "step": 871
+    },
+    {
+      "epoch": 0.00872,
+      "grad_norm": 0.5879044949928981,
+      "learning_rate": 0.002616,
+      "loss": 4.5345,
+      "step": 872
+    },
+    {
+      "epoch": 0.00873,
+      "grad_norm": 0.5844885657438569,
+      "learning_rate": 0.0026190000000000002,
+      "loss": 4.5441,
+      "step": 873
+    },
+    {
+      "epoch": 0.00874,
+      "grad_norm": 0.5286751889514939,
+      "learning_rate": 0.002622,
+      "loss": 4.5573,
+      "step": 874
+    },
+    {
+      "epoch": 0.00875,
+      "grad_norm": 0.5494502975686567,
+      "learning_rate": 0.002625,
+      "loss": 4.5051,
+      "step": 875
+    },
+    {
+      "epoch": 0.00876,
+      "grad_norm": 0.6269645893334619,
+      "learning_rate": 0.002628,
+      "loss": 4.5356,
+      "step": 876
+    },
+    {
+      "epoch": 0.00877,
+      "grad_norm": 0.6901692969476347,
+      "learning_rate": 0.002631,
+      "loss": 4.5537,
+      "step": 877
+    },
+    {
+      "epoch": 0.00878,
+      "grad_norm": 0.7433571199864201,
+      "learning_rate": 0.002634,
+      "loss": 4.4894,
+      "step": 878
+    },
+    {
+      "epoch": 0.00879,
+      "grad_norm": 0.7806263902694818,
+      "learning_rate": 0.002637,
+      "loss": 4.5303,
+      "step": 879
+    },
+    {
+      "epoch": 0.0088,
+      "grad_norm": 0.7694352427135146,
+      "learning_rate": 0.00264,
+      "loss": 4.5353,
+      "step": 880
+    },
+    {
+      "epoch": 0.00881,
+      "grad_norm": 0.6938035372615263,
+      "learning_rate": 0.002643,
+      "loss": 4.5351,
+      "step": 881
+    },
+    {
+      "epoch": 0.00882,
+      "grad_norm": 0.672108824408367,
+      "learning_rate": 0.002646,
+      "loss": 4.5331,
+      "step": 882
+    },
+    {
+      "epoch": 0.00883,
+      "grad_norm": 0.6847246415873477,
+      "learning_rate": 0.002649,
+      "loss": 4.524,
+      "step": 883
+    },
+    {
+      "epoch": 0.00884,
+      "grad_norm": 0.8034466162851082,
+      "learning_rate": 0.0026520000000000003,
+      "loss": 4.5123,
+      "step": 884
+    },
+    {
+      "epoch": 0.00885,
+      "grad_norm": 0.8504740488359851,
+      "learning_rate": 0.0026550000000000002,
+      "loss": 4.5174,
+      "step": 885
+    },
+    {
+      "epoch": 0.00886,
+      "grad_norm": 0.6814828631263485,
+      "learning_rate": 0.002658,
+      "loss": 4.526,
+      "step": 886
+    },
+    {
+      "epoch": 0.00887,
+      "grad_norm": 0.7764171535679266,
+      "learning_rate": 0.002661,
+      "loss": 4.5254,
+      "step": 887
+    },
+    {
+      "epoch": 0.00888,
+      "grad_norm": 0.864178556132455,
+      "learning_rate": 0.002664,
+      "loss": 4.5323,
+      "step": 888
+    },
+    {
+      "epoch": 0.00889,
+      "grad_norm": 0.9963366133988169,
+      "learning_rate": 0.002667,
+      "loss": 4.4915,
+      "step": 889
+    },
+    {
+      "epoch": 0.0089,
+      "grad_norm": 0.9684646116139818,
+      "learning_rate": 0.00267,
+      "loss": 4.5454,
+      "step": 890
+    },
+    {
+      "epoch": 0.00891,
+      "grad_norm": 0.8019607803647243,
+      "learning_rate": 0.002673,
+      "loss": 4.523,
+      "step": 891
+    },
+    {
+      "epoch": 0.00892,
+      "grad_norm": 0.9082109544296421,
+      "learning_rate": 0.002676,
+      "loss": 4.5431,
+      "step": 892
+    },
+    {
+      "epoch": 0.00893,
+      "grad_norm": 0.9728985136087724,
+      "learning_rate": 0.002679,
+      "loss": 4.5212,
+      "step": 893
+    },
+    {
+      "epoch": 0.00894,
+      "grad_norm": 1.0580229080074564,
+      "learning_rate": 0.002682,
+      "loss": 4.5156,
+      "step": 894
+    },
+    {
+      "epoch": 0.00895,
+      "grad_norm": 0.8166896686472195,
+      "learning_rate": 0.0026850000000000003,
+      "loss": 4.5188,
+      "step": 895
+    },
+    {
+      "epoch": 0.00896,
+      "grad_norm": 0.8942454442625324,
+      "learning_rate": 0.0026880000000000003,
+      "loss": 4.5121,
+      "step": 896
+    },
+    {
+      "epoch": 0.00897,
+      "grad_norm": 0.9627689980406442,
+      "learning_rate": 0.0026910000000000002,
+      "loss": 4.5406,
+      "step": 897
+    },
+    {
+      "epoch": 0.00898,
+      "grad_norm": 0.915209672072645,
+      "learning_rate": 0.002694,
+      "loss": 4.5291,
+      "step": 898
+    },
+    {
+      "epoch": 0.00899,
+      "grad_norm": 0.8008505248098504,
+      "learning_rate": 0.002697,
+      "loss": 4.5313,
+      "step": 899
+    },
+    {
+      "epoch": 0.009,
+      "grad_norm": 0.974662986907947,
+      "learning_rate": 0.0027,
+      "loss": 4.5627,
+      "step": 900
+    },
+    {
+      "epoch": 0.00901,
+      "grad_norm": 1.1878109077704273,
+      "learning_rate": 0.002703,
+      "loss": 4.5336,
+      "step": 901
+    },
+    {
+      "epoch": 0.00902,
+      "grad_norm": 0.7391945290133592,
+      "learning_rate": 0.002706,
+      "loss": 4.5248,
+      "step": 902
+    },
+    {
+      "epoch": 0.00903,
+      "grad_norm": 0.7310904041760188,
+      "learning_rate": 0.002709,
+      "loss": 4.5195,
+      "step": 903
+    },
+    {
+      "epoch": 0.00904,
+      "grad_norm": 0.763540341773903,
+      "learning_rate": 0.002712,
+      "loss": 4.4985,
+      "step": 904
+    },
+    {
+      "epoch": 0.00905,
+      "grad_norm": 0.922476362328922,
+      "learning_rate": 0.002715,
+      "loss": 4.5209,
+      "step": 905
+    },
+    {
+      "epoch": 0.00906,
+      "grad_norm": 0.9574009618515481,
+      "learning_rate": 0.002718,
+      "loss": 4.5379,
+      "step": 906
+    },
+    {
+      "epoch": 0.00907,
+      "grad_norm": 1.035774941473643,
+      "learning_rate": 0.0027210000000000003,
+      "loss": 4.5538,
+      "step": 907
+    },
+    {
+      "epoch": 0.00908,
+      "grad_norm": 1.0191319400722418,
+      "learning_rate": 0.0027240000000000003,
+      "loss": 4.5279,
+      "step": 908
+    },
+    {
+      "epoch": 0.00909,
+      "grad_norm": 1.1103950353838743,
+      "learning_rate": 0.0027270000000000003,
+      "loss": 4.554,
+      "step": 909
+    },
+    {
+      "epoch": 0.0091,
+      "grad_norm": 0.8691030580510016,
+      "learning_rate": 0.0027300000000000002,
+      "loss": 4.5543,
+      "step": 910
+    },
+    {
+      "epoch": 0.00911,
+      "grad_norm": 0.7723878752960539,
+      "learning_rate": 0.002733,
+      "loss": 4.5035,
+      "step": 911
+    },
+    {
+      "epoch": 0.00912,
+      "grad_norm": 0.8180906874107354,
+      "learning_rate": 0.002736,
+      "loss": 4.5117,
+      "step": 912
+    },
+    {
+      "epoch": 0.00913,
+      "grad_norm": 0.8148069350423363,
+      "learning_rate": 0.002739,
+      "loss": 4.505,
+      "step": 913
+    },
+    {
+      "epoch": 0.00914,
+      "grad_norm": 0.7118677869163097,
+      "learning_rate": 0.002742,
+      "loss": 4.5326,
+      "step": 914
+    },
+    {
+      "epoch": 0.00915,
+      "grad_norm": 0.67221021637466,
+      "learning_rate": 0.002745,
+      "loss": 4.4896,
+      "step": 915
+    },
+    {
+      "epoch": 0.00916,
+      "grad_norm": 0.7521734182214744,
+      "learning_rate": 0.002748,
+      "loss": 4.4742,
+      "step": 916
+    },
+    {
+      "epoch": 0.00917,
+      "grad_norm": 0.6660072639776646,
+      "learning_rate": 0.002751,
+      "loss": 4.4809,
+      "step": 917
+    },
+    {
+      "epoch": 0.00918,
+      "grad_norm": 0.5671697873930601,
+      "learning_rate": 0.0027540000000000004,
+      "loss": 4.4963,
+      "step": 918
+    },
+    {
+      "epoch": 0.00919,
+      "grad_norm": 0.5594822914263968,
+      "learning_rate": 0.0027570000000000003,
+      "loss": 4.4754,
+      "step": 919
+    },
+    {
+      "epoch": 0.0092,
+      "grad_norm": 0.6080421327199805,
+      "learning_rate": 0.0027600000000000003,
+      "loss": 4.4903,
+      "step": 920
+    },
+    {
+      "epoch": 0.00921,
+      "grad_norm": 0.728784609109995,
+      "learning_rate": 0.0027630000000000003,
+      "loss": 4.4759,
+      "step": 921
+    },
+    {
+      "epoch": 0.00922,
+      "grad_norm": 0.8477951683682543,
+      "learning_rate": 0.0027660000000000002,
+      "loss": 4.5076,
+      "step": 922
+    },
+    {
+      "epoch": 0.00923,
+      "grad_norm": 0.6016579194588629,
+      "learning_rate": 0.002769,
+      "loss": 4.4637,
+      "step": 923
+    },
+    {
+      "epoch": 0.00924,
+      "grad_norm": 0.6726777284073512,
+      "learning_rate": 0.002772,
+      "loss": 4.4919,
+      "step": 924
+    },
+    {
+      "epoch": 0.00925,
+      "grad_norm": 0.7061399279754877,
+      "learning_rate": 0.002775,
+      "loss": 4.4751,
+      "step": 925
+    },
+    {
+      "epoch": 0.00926,
+      "grad_norm": 0.6525850527724611,
+      "learning_rate": 0.002778,
+      "loss": 4.4844,
+      "step": 926
+    },
+    {
+      "epoch": 0.00927,
+      "grad_norm": 0.6775355167555962,
+      "learning_rate": 0.002781,
+      "loss": 4.4493,
+      "step": 927
+    },
+    {
+      "epoch": 0.00928,
+      "grad_norm": 0.660488843824027,
+      "learning_rate": 0.002784,
+      "loss": 4.493,
+      "step": 928
+    },
+    {
+      "epoch": 0.00929,
+      "grad_norm": 0.6375906920183253,
+      "learning_rate": 0.0027870000000000004,
+      "loss": 4.4714,
+      "step": 929
+    },
+    {
+      "epoch": 0.0093,
+      "grad_norm": 0.7170906127689122,
+      "learning_rate": 0.0027900000000000004,
+      "loss": 4.4557,
+      "step": 930
+    },
+    {
+      "epoch": 0.00931,
+      "grad_norm": 0.7727209257871167,
+      "learning_rate": 0.0027930000000000003,
+      "loss": 4.4706,
+      "step": 931
+    },
+    {
+      "epoch": 0.00932,
+      "grad_norm": 0.759902395005458,
+      "learning_rate": 0.0027960000000000003,
+      "loss": 4.4795,
+      "step": 932
+    },
+    {
+      "epoch": 0.00933,
+      "grad_norm": 0.9089287828754257,
+      "learning_rate": 0.0027990000000000003,
+      "loss": 4.4536,
+      "step": 933
+    },
+    {
+      "epoch": 0.00934,
+      "grad_norm": 0.9561562356933814,
+      "learning_rate": 0.0028020000000000002,
+      "loss": 4.4859,
+      "step": 934
+    },
+    {
+      "epoch": 0.00935,
+      "grad_norm": 1.010681147518413,
+      "learning_rate": 0.002805,
+      "loss": 4.5044,
+      "step": 935
+    },
+    {
+      "epoch": 0.00936,
+      "grad_norm": 0.960270788501537,
+      "learning_rate": 0.002808,
+      "loss": 4.5076,
+      "step": 936
+    },
+    {
+      "epoch": 0.00937,
+      "grad_norm": 0.8170703531029062,
+      "learning_rate": 0.002811,
+      "loss": 4.5151,
+      "step": 937
+    },
+    {
+      "epoch": 0.00938,
+      "grad_norm": 0.6779931914379209,
+      "learning_rate": 0.002814,
+      "loss": 4.4854,
+      "step": 938
+    },
+    {
+      "epoch": 0.00939,
+      "grad_norm": 0.6509733323350428,
+      "learning_rate": 0.002817,
+      "loss": 4.4771,
+      "step": 939
+    },
+    {
+      "epoch": 0.0094,
+      "grad_norm": 0.6633333111873998,
+      "learning_rate": 0.00282,
+      "loss": 4.4796,
+      "step": 940
+    },
+    {
+      "epoch": 0.00941,
+      "grad_norm": 0.7449615790040228,
+      "learning_rate": 0.002823,
+      "loss": 4.4576,
+      "step": 941
+    },
+    {
+      "epoch": 0.00942,
+      "grad_norm": 0.7696922222349031,
+      "learning_rate": 0.002826,
+      "loss": 4.4823,
+      "step": 942
+    },
+    {
+      "epoch": 0.00943,
+      "grad_norm": 0.6087986992050594,
+      "learning_rate": 0.002829,
+      "loss": 4.4688,
+      "step": 943
+    },
+    {
+      "epoch": 0.00944,
+      "grad_norm": 0.5228736659630974,
+      "learning_rate": 0.002832,
+      "loss": 4.4678,
+      "step": 944
+    },
+    {
+      "epoch": 0.00945,
+      "grad_norm": 0.5754533269053475,
+      "learning_rate": 0.002835,
+      "loss": 4.4713,
+      "step": 945
+    },
+    {
+      "epoch": 0.00946,
+      "grad_norm": 0.6164648444392735,
+      "learning_rate": 0.002838,
+      "loss": 4.4832,
+      "step": 946
+    },
+    {
+      "epoch": 0.00947,
+      "grad_norm": 0.6419505093155794,
+      "learning_rate": 0.0028409999999999998,
+      "loss": 4.4556,
+      "step": 947
+    },
+    {
+      "epoch": 0.00948,
+      "grad_norm": 0.556707209906786,
+      "learning_rate": 0.0028439999999999997,
+      "loss": 4.4351,
+      "step": 948
+    },
+    {
+      "epoch": 0.00949,
+      "grad_norm": 0.5031407625923785,
+      "learning_rate": 0.002847,
+      "loss": 4.4737,
+      "step": 949
+    },
+    {
+      "epoch": 0.0095,
+      "grad_norm": 0.5881138889925357,
+      "learning_rate": 0.00285,
+      "loss": 4.4655,
+      "step": 950
+    },
+    {
+      "epoch": 0.00951,
+      "grad_norm": 0.6194945571715194,
+      "learning_rate": 0.002853,
+      "loss": 4.4418,
+      "step": 951
+    },
+    {
+      "epoch": 0.00952,
+      "grad_norm": 0.6965416314433459,
+      "learning_rate": 0.002856,
+      "loss": 4.4704,
+      "step": 952
+    },
+    {
+      "epoch": 0.00953,
+      "grad_norm": 0.7506696864658969,
+      "learning_rate": 0.002859,
+      "loss": 4.4161,
+      "step": 953
+    },
+    {
+      "epoch": 0.00954,
+      "grad_norm": 0.6440899731036704,
+      "learning_rate": 0.002862,
+      "loss": 4.4703,
+      "step": 954
+    },
+    {
+      "epoch": 0.00955,
+      "grad_norm": 0.7716821398038454,
+      "learning_rate": 0.002865,
+      "loss": 4.4601,
+      "step": 955
+    },
+    {
+      "epoch": 0.00956,
+      "grad_norm": 0.9111507025485583,
+      "learning_rate": 0.002868,
+      "loss": 4.4578,
+      "step": 956
+    },
+    {
+      "epoch": 0.00957,
+      "grad_norm": 0.8443539201487298,
+      "learning_rate": 0.002871,
+      "loss": 4.4727,
+      "step": 957
+    },
+    {
+      "epoch": 0.00958,
+      "grad_norm": 0.7989694313181581,
+      "learning_rate": 0.002874,
+      "loss": 4.4582,
+      "step": 958
+    },
+    {
+      "epoch": 0.00959,
+      "grad_norm": 0.7547037856292312,
+      "learning_rate": 0.002877,
+      "loss": 4.4482,
+      "step": 959
+    },
+    {
+      "epoch": 0.0096,
+      "grad_norm": 0.7121103430025651,
+      "learning_rate": 0.0028799999999999997,
+      "loss": 4.4466,
+      "step": 960
+    },
+    {
+      "epoch": 0.00961,
+      "grad_norm": 0.6758285086831927,
+      "learning_rate": 0.002883,
+      "loss": 4.4209,
+      "step": 961
+    },
+    {
+      "epoch": 0.00962,
+      "grad_norm": 0.5336350485211147,
+      "learning_rate": 0.002886,
+      "loss": 4.4436,
+      "step": 962
+    },
+    {
+      "epoch": 0.00963,
+      "grad_norm": 0.5703472269796636,
+      "learning_rate": 0.002889,
+      "loss": 4.4467,
+      "step": 963
+    },
+    {
+      "epoch": 0.00964,
+      "grad_norm": 0.6524118600131552,
+      "learning_rate": 0.002892,
+      "loss": 4.4355,
+      "step": 964
+    },
+    {
+      "epoch": 0.00965,
+      "grad_norm": 0.7342343756321769,
+      "learning_rate": 0.002895,
+      "loss": 4.4144,
+      "step": 965
+    },
+    {
+      "epoch": 0.00966,
+      "grad_norm": 0.9544017629329996,
+      "learning_rate": 0.002898,
+      "loss": 4.4408,
+      "step": 966
+    },
+    {
+      "epoch": 0.00967,
+      "grad_norm": 1.1065817167809169,
+      "learning_rate": 0.002901,
+      "loss": 4.4927,
+      "step": 967
+    },
+    {
+      "epoch": 0.00968,
+      "grad_norm": 0.9829826223932284,
+      "learning_rate": 0.002904,
+      "loss": 4.4568,
+      "step": 968
+    },
+    {
+      "epoch": 0.00969,
+      "grad_norm": 0.8112557218556605,
+      "learning_rate": 0.002907,
+      "loss": 4.4805,
+      "step": 969
+    },
+    {
+      "epoch": 0.0097,
+      "grad_norm": 0.983264710703698,
+      "learning_rate": 0.00291,
+      "loss": 4.4488,
+      "step": 970
+    },
+    {
+      "epoch": 0.00971,
+      "grad_norm": 1.010591995610007,
+      "learning_rate": 0.002913,
+      "loss": 4.4796,
+      "step": 971
+    },
+    {
+      "epoch": 0.00972,
+      "grad_norm": 0.9930743768968635,
+      "learning_rate": 0.002916,
+      "loss": 4.4798,
+      "step": 972
+    },
+    {
+      "epoch": 0.00973,
+      "grad_norm": 1.1817662624154759,
+      "learning_rate": 0.002919,
+      "loss": 4.4931,
+      "step": 973
+    },
+    {
+      "epoch": 0.00974,
+      "grad_norm": 1.1915796929226945,
+      "learning_rate": 0.002922,
+      "loss": 4.4839,
+      "step": 974
+    },
+    {
+      "epoch": 0.00975,
+      "grad_norm": 0.9880254500237188,
+      "learning_rate": 0.002925,
+      "loss": 4.4969,
+      "step": 975
+    },
+    {
+      "epoch": 0.00976,
+      "grad_norm": 1.0394868289507098,
+      "learning_rate": 0.002928,
+      "loss": 4.4906,
+      "step": 976
+    },
+    {
+      "epoch": 0.00977,
+      "grad_norm": 1.1242485803004214,
+      "learning_rate": 0.002931,
+      "loss": 4.5346,
+      "step": 977
+    },
+    {
+      "epoch": 0.00978,
+      "grad_norm": 1.1287291198302327,
+      "learning_rate": 0.002934,
+      "loss": 4.5134,
+      "step": 978
+    },
+    {
+      "epoch": 0.00979,
+      "grad_norm": 0.937365777135907,
+      "learning_rate": 0.002937,
+      "loss": 4.4945,
+      "step": 979
+    },
+    {
+      "epoch": 0.0098,
+      "grad_norm": 0.9999316178288561,
+      "learning_rate": 0.00294,
+      "loss": 4.4906,
+      "step": 980
+    },
+    {
+      "epoch": 0.00981,
+      "grad_norm": 1.0692695747943242,
+      "learning_rate": 0.002943,
+      "loss": 4.5166,
+      "step": 981
+    },
+    {
+      "epoch": 0.00982,
+      "grad_norm": 0.855671043658802,
+      "learning_rate": 0.002946,
+      "loss": 4.4983,
+      "step": 982
+    },
+    {
+      "epoch": 0.00983,
+      "grad_norm": 0.8819131773225922,
+      "learning_rate": 0.0029490000000000002,
+      "loss": 4.4932,
+      "step": 983
+    },
+    {
+      "epoch": 0.00984,
+      "grad_norm": 0.8474323035176099,
+      "learning_rate": 0.002952,
+      "loss": 4.4794,
+      "step": 984
+    },
+    {
+      "epoch": 0.00985,
+      "grad_norm": 0.8010021292921615,
+      "learning_rate": 0.002955,
+      "loss": 4.4955,
+      "step": 985
+    },
+    {
+      "epoch": 0.00986,
+      "grad_norm": 0.6778548937329826,
+      "learning_rate": 0.002958,
+      "loss": 4.4543,
+      "step": 986
+    },
+    {
+      "epoch": 0.00987,
+      "grad_norm": 0.6335901643582099,
+      "learning_rate": 0.002961,
+      "loss": 4.4618,
+      "step": 987
+    },
+    {
+      "epoch": 0.00988,
+      "grad_norm": 0.6898288520048222,
+      "learning_rate": 0.002964,
+      "loss": 4.4934,
+      "step": 988
+    },
+    {
+      "epoch": 0.00989,
+      "grad_norm": 0.6968481266470886,
+      "learning_rate": 0.002967,
+      "loss": 4.4338,
+      "step": 989
+    },
+    {
+      "epoch": 0.0099,
+      "grad_norm": 0.7462093807424477,
+      "learning_rate": 0.00297,
+      "loss": 4.4706,
+      "step": 990
+    },
+    {
+      "epoch": 0.00991,
+      "grad_norm": 0.7437971839370325,
+      "learning_rate": 0.002973,
+      "loss": 4.4697,
+      "step": 991
+    },
+    {
+      "epoch": 0.00992,
+      "grad_norm": 0.6722774287362101,
+      "learning_rate": 0.002976,
+      "loss": 4.4576,
+      "step": 992
+    },
+    {
+      "epoch": 0.00993,
+      "grad_norm": 0.6583210228081078,
+      "learning_rate": 0.002979,
+      "loss": 4.4522,
+      "step": 993
+    },
+    {
+      "epoch": 0.00994,
+      "grad_norm": 0.6827391668775007,
+      "learning_rate": 0.002982,
+      "loss": 4.4553,
+      "step": 994
+    },
+    {
+      "epoch": 0.00995,
+      "grad_norm": 0.7054299500075728,
+      "learning_rate": 0.0029850000000000002,
+      "loss": 4.4538,
+      "step": 995
+    },
+    {
+      "epoch": 0.00996,
+      "grad_norm": 0.7382372248626952,
+      "learning_rate": 0.002988,
+      "loss": 4.4592,
+      "step": 996
+    },
+    {
+      "epoch": 0.00997,
+      "grad_norm": 0.6621181102222602,
+      "learning_rate": 0.002991,
+      "loss": 4.4437,
+      "step": 997
+    },
+    {
+      "epoch": 0.00998,
+      "grad_norm": 0.6532026522918166,
+      "learning_rate": 0.002994,
+      "loss": 4.4467,
+      "step": 998
+    },
+    {
+      "epoch": 0.00999,
+      "grad_norm": 0.6863970198472243,
+      "learning_rate": 0.002997,
+      "loss": 4.4317,
+      "step": 999
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6643394709769835,
+      "learning_rate": 0.003,
+      "loss": 4.4529,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01001,
+      "grad_norm": 0.5664327532419113,
+      "learning_rate": 0.003,
+      "loss": 4.4443,
+      "step": 1001
+    },
+    {
+      "epoch": 0.01002,
+      "grad_norm": 0.5614282124245983,
+      "learning_rate": 0.003,
+      "loss": 4.4107,
+      "step": 1002
+    },
+    {
+      "epoch": 0.01003,
+      "grad_norm": 0.5395347922472447,
+      "learning_rate": 0.003,
+      "loss": 4.4319,
+      "step": 1003
+    },
+    {
+      "epoch": 0.01004,
+      "grad_norm": 0.5638853817459814,
+      "learning_rate": 0.003,
+      "loss": 4.4194,
+      "step": 1004
+    },
+    {
+      "epoch": 0.01005,
+      "grad_norm": 0.5916080142589283,
+      "learning_rate": 0.003,
+      "loss": 4.3932,
+      "step": 1005
+    },
+    {
+      "epoch": 0.01006,
+      "grad_norm": 0.5700688347090367,
+      "learning_rate": 0.003,
+      "loss": 4.3913,
+      "step": 1006
+    },
+    {
+      "epoch": 0.01007,
+      "grad_norm": 0.5618557938751672,
+      "learning_rate": 0.003,
+      "loss": 4.4343,
+      "step": 1007
+    },
+    {
+      "epoch": 0.01008,
+      "grad_norm": 0.6176294799951538,
+      "learning_rate": 0.003,
+      "loss": 4.4449,
+      "step": 1008
+    },
+    {
+      "epoch": 0.01009,
+      "grad_norm": 0.6917308148051521,
+      "learning_rate": 0.003,
+      "loss": 4.4152,
+      "step": 1009
+    },
+    {
+      "epoch": 0.0101,
+      "grad_norm": 0.7355219940735465,
+      "learning_rate": 0.003,
+      "loss": 4.4242,
+      "step": 1010
+    },
+    {
+      "epoch": 0.01011,
+      "grad_norm": 0.8258393023556594,
+      "learning_rate": 0.003,
+      "loss": 4.437,
+      "step": 1011
+    },
+    {
+      "epoch": 0.01012,
+      "grad_norm": 0.7883368033913232,
+      "learning_rate": 0.003,
+      "loss": 4.4273,
+      "step": 1012
+    },
+    {
+      "epoch": 0.01013,
+      "grad_norm": 0.8421670780561094,
+      "learning_rate": 0.003,
+      "loss": 4.453,
+      "step": 1013
+    },
+    {
+      "epoch": 0.01014,
+      "grad_norm": 0.7611364620814401,
+      "learning_rate": 0.003,
+      "loss": 4.3906,
+      "step": 1014
+    },
+    {
+      "epoch": 0.01015,
+      "grad_norm": 0.8948224171004916,
+      "learning_rate": 0.003,
+      "loss": 4.4422,
+      "step": 1015
+    },
+    {
+      "epoch": 0.01016,
+      "grad_norm": 0.9330968648547454,
+      "learning_rate": 0.003,
+      "loss": 4.4537,
+      "step": 1016
+    },
+    {
+      "epoch": 0.01017,
+      "grad_norm": 0.9071988294298332,
+      "learning_rate": 0.003,
+      "loss": 4.4046,
+      "step": 1017
+    },
+    {
+      "epoch": 0.01018,
+      "grad_norm": 0.9028445315923417,
+      "learning_rate": 0.003,
+      "loss": 4.4536,
+      "step": 1018
+    },
+    {
+      "epoch": 0.01019,
+      "grad_norm": 0.701398424995911,
+      "learning_rate": 0.003,
+      "loss": 4.4297,
+      "step": 1019
+    },
+    {
+      "epoch": 0.0102,
+      "grad_norm": 0.7422831527736846,
+      "learning_rate": 0.003,
+      "loss": 4.428,
+      "step": 1020
+    },
+    {
+      "epoch": 0.01021,
+      "grad_norm": 0.7468650100074169,
+      "learning_rate": 0.003,
+      "loss": 4.4136,
+      "step": 1021
+    },
+    {
+      "epoch": 0.01022,
+      "grad_norm": 0.9325291879496833,
+      "learning_rate": 0.003,
+      "loss": 4.4294,
+      "step": 1022
+    },
+    {
+      "epoch": 0.01023,
+      "grad_norm": 0.8401676161943546,
+      "learning_rate": 0.003,
+      "loss": 4.4261,
+      "step": 1023
+    },
+    {
+      "epoch": 0.01024,
+      "grad_norm": 0.8270343729935282,
+      "learning_rate": 0.003,
+      "loss": 4.3898,
+      "step": 1024
+    },
+    {
+      "epoch": 0.01025,
+      "grad_norm": 0.7180913593704098,
+      "learning_rate": 0.003,
+      "loss": 4.4147,
+      "step": 1025
+    },
+    {
+      "epoch": 0.01026,
+      "grad_norm": 0.6277034329789435,
+      "learning_rate": 0.003,
+      "loss": 4.4039,
+      "step": 1026
+    },
+    {
+      "epoch": 0.01027,
+      "grad_norm": 0.5318737503801085,
+      "learning_rate": 0.003,
+      "loss": 4.4219,
+      "step": 1027
+    },
+    {
+      "epoch": 0.01028,
+      "grad_norm": 0.4999785226446009,
+      "learning_rate": 0.003,
+      "loss": 4.4015,
+      "step": 1028
+    },
+    {
+      "epoch": 0.01029,
+      "grad_norm": 0.48973961084120876,
+      "learning_rate": 0.003,
+      "loss": 4.3769,
+      "step": 1029
+    },
+    {
+      "epoch": 0.0103,
+      "grad_norm": 0.5942733082980836,
+      "learning_rate": 0.003,
+      "loss": 4.4125,
+      "step": 1030
+    },
+    {
+      "epoch": 0.01031,
+      "grad_norm": 0.7229752342980397,
+      "learning_rate": 0.003,
+      "loss": 4.4274,
+      "step": 1031
+    },
+    {
+      "epoch": 0.01032,
+      "grad_norm": 0.7373514084562369,
+      "learning_rate": 0.003,
+      "loss": 4.4145,
+      "step": 1032
+    },
+    {
+      "epoch": 0.01033,
+      "grad_norm": 0.6482800157489234,
+      "learning_rate": 0.003,
+      "loss": 4.4346,
+      "step": 1033
+    },
+    {
+      "epoch": 0.01034,
+      "grad_norm": 0.6524184922218951,
+      "learning_rate": 0.003,
+      "loss": 4.4051,
+      "step": 1034
+    },
+    {
+      "epoch": 0.01035,
+      "grad_norm": 0.705389079509784,
+      "learning_rate": 0.003,
+      "loss": 4.4073,
+      "step": 1035
+    },
+    {
+      "epoch": 0.01036,
+      "grad_norm": 0.7235541204605349,
+      "learning_rate": 0.003,
+      "loss": 4.3939,
+      "step": 1036
+    },
+    {
+      "epoch": 0.01037,
+      "grad_norm": 0.6796343657160102,
+      "learning_rate": 0.003,
+      "loss": 4.4161,
+      "step": 1037
+    },
+    {
+      "epoch": 0.01038,
+      "grad_norm": 0.5773141410598708,
+      "learning_rate": 0.003,
+      "loss": 4.4121,
+      "step": 1038
+    },
+    {
+      "epoch": 0.01039,
+      "grad_norm": 0.5565285106963757,
+      "learning_rate": 0.003,
+      "loss": 4.4014,
+      "step": 1039
+    },
+    {
+      "epoch": 0.0104,
+      "grad_norm": 0.526665138631767,
+      "learning_rate": 0.003,
+      "loss": 4.4124,
+      "step": 1040
+    },
+    {
+      "epoch": 0.01041,
+      "grad_norm": 0.5117793067919421,
+      "learning_rate": 0.003,
+      "loss": 4.3909,
+      "step": 1041
+    },
+    {
+      "epoch": 0.01042,
+      "grad_norm": 0.5092553549098504,
+      "learning_rate": 0.003,
+      "loss": 4.4167,
+      "step": 1042
+    },
+    {
+      "epoch": 0.01043,
+      "grad_norm": 0.5543198083705748,
+      "learning_rate": 0.003,
+      "loss": 4.3593,
+      "step": 1043
+    },
+    {
+      "epoch": 0.01044,
+      "grad_norm": 0.600481111898673,
+      "learning_rate": 0.003,
+      "loss": 4.3987,
+      "step": 1044
+    },
+    {
+      "epoch": 0.01045,
+      "grad_norm": 0.5965980004717368,
+      "learning_rate": 0.003,
+      "loss": 4.4162,
+      "step": 1045
+    },
+    {
+      "epoch": 0.01046,
+      "grad_norm": 0.6740252530686676,
+      "learning_rate": 0.003,
+      "loss": 4.3928,
+      "step": 1046
+    },
+    {
+      "epoch": 0.01047,
+      "grad_norm": 0.7361053030528937,
+      "learning_rate": 0.003,
+      "loss": 4.3997,
+      "step": 1047
+    },
+    {
+      "epoch": 0.01048,
+      "grad_norm": 0.767513361829787,
+      "learning_rate": 0.003,
+      "loss": 4.3936,
+      "step": 1048
+    },
+    {
+      "epoch": 0.01049,
+      "grad_norm": 0.6855190682446736,
+      "learning_rate": 0.003,
+      "loss": 4.3699,
+      "step": 1049
+    },
+    {
+      "epoch": 0.0105,
+      "grad_norm": 0.5741691502464937,
+      "learning_rate": 0.003,
+      "loss": 4.4062,
+      "step": 1050
+    },
+    {
+      "epoch": 0.01051,
+      "grad_norm": 0.6230464159511063,
+      "learning_rate": 0.003,
+      "loss": 4.3827,
+      "step": 1051
+    },
+    {
+      "epoch": 0.01052,
+      "grad_norm": 0.6366737529257143,
+      "learning_rate": 0.003,
+      "loss": 4.3967,
+      "step": 1052
+    },
+    {
+      "epoch": 0.01053,
+      "grad_norm": 0.6755427193415695,
+      "learning_rate": 0.003,
+      "loss": 4.4056,
+      "step": 1053
+    },
+    {
+      "epoch": 0.01054,
+      "grad_norm": 0.6231015147472246,
+      "learning_rate": 0.003,
+      "loss": 4.3965,
+      "step": 1054
+    },
+    {
+      "epoch": 0.01055,
+      "grad_norm": 0.5394645878748071,
+      "learning_rate": 0.003,
+      "loss": 4.3715,
+      "step": 1055
+    },
+    {
+      "epoch": 0.01056,
+      "grad_norm": 0.5755128064540977,
+      "learning_rate": 0.003,
+      "loss": 4.3719,
+      "step": 1056
+    },
+    {
+      "epoch": 0.01057,
+      "grad_norm": 0.5491694969265289,
+      "learning_rate": 0.003,
+      "loss": 4.3754,
+      "step": 1057
+    },
+    {
+      "epoch": 0.01058,
+      "grad_norm": 0.7343919109861163,
+      "learning_rate": 0.003,
+      "loss": 4.375,
+      "step": 1058
+    },
+    {
+      "epoch": 0.01059,
+      "grad_norm": 0.9361910412332192,
+      "learning_rate": 0.003,
+      "loss": 4.3892,
+      "step": 1059
+    },
+    {
+      "epoch": 0.0106,
+      "grad_norm": 0.9065515100292879,
+      "learning_rate": 0.003,
+      "loss": 4.3901,
+      "step": 1060
+    },
+    {
+      "epoch": 0.01061,
+      "grad_norm": 1.1211758916188983,
+      "learning_rate": 0.003,
+      "loss": 4.3972,
+      "step": 1061
+    },
+    {
+      "epoch": 0.01062,
+      "grad_norm": 0.7834596971762835,
+      "learning_rate": 0.003,
+      "loss": 4.3907,
+      "step": 1062
+    },
+    {
+      "epoch": 0.01063,
+      "grad_norm": 0.6963669031971059,
+      "learning_rate": 0.003,
+      "loss": 4.3832,
+      "step": 1063
+    },
+    {
+      "epoch": 0.01064,
+      "grad_norm": 0.6238185439704477,
+      "learning_rate": 0.003,
+      "loss": 4.3945,
+      "step": 1064
+    },
+    {
+      "epoch": 0.01065,
+      "grad_norm": 0.6958753605701185,
+      "learning_rate": 0.003,
+      "loss": 4.3993,
+      "step": 1065
+    },
+    {
+      "epoch": 0.01066,
+      "grad_norm": 0.7129893176363267,
+      "learning_rate": 0.003,
+      "loss": 4.4025,
+      "step": 1066
+    },
+    {
+      "epoch": 0.01067,
+      "grad_norm": 0.7862199002009355,
+      "learning_rate": 0.003,
+      "loss": 4.4178,
+      "step": 1067
+    },
+    {
+      "epoch": 0.01068,
+      "grad_norm": 0.7422673244500091,
+      "learning_rate": 0.003,
+      "loss": 4.3984,
+      "step": 1068
+    },
+    {
+      "epoch": 0.01069,
+      "grad_norm": 0.7997129592129516,
+      "learning_rate": 0.003,
+      "loss": 4.4072,
+      "step": 1069
+    },
+    {
+      "epoch": 0.0107,
+      "grad_norm": 0.945099613944759,
+      "learning_rate": 0.003,
+      "loss": 4.4674,
+      "step": 1070
+    },
+    {
+      "epoch": 0.01071,
+      "grad_norm": 0.8590095455876905,
+      "learning_rate": 0.003,
+      "loss": 4.387,
+      "step": 1071
+    },
+    {
+      "epoch": 0.01072,
+      "grad_norm": 0.8651521698594221,
+      "learning_rate": 0.003,
+      "loss": 4.4227,
+      "step": 1072
+    },
+    {
+      "epoch": 0.01073,
+      "grad_norm": 0.80387311045371,
+      "learning_rate": 0.003,
+      "loss": 4.4158,
+      "step": 1073
+    },
+    {
+      "epoch": 0.01074,
+      "grad_norm": 0.8137862583554676,
+      "learning_rate": 0.003,
+      "loss": 4.3801,
+      "step": 1074
+    },
+    {
+      "epoch": 0.01075,
+      "grad_norm": 0.8055768924194764,
+      "learning_rate": 0.003,
+      "loss": 4.4185,
+      "step": 1075
+    },
+    {
+      "epoch": 0.01076,
+      "grad_norm": 0.8990845414650448,
+      "learning_rate": 0.003,
+      "loss": 4.4196,
+      "step": 1076
+    },
+    {
+      "epoch": 0.01077,
+      "grad_norm": 0.9178768409478193,
+      "learning_rate": 0.003,
+      "loss": 4.4207,
+      "step": 1077
+    },
+    {
+      "epoch": 0.01078,
+      "grad_norm": 0.9083931280693812,
+      "learning_rate": 0.003,
+      "loss": 4.3748,
+      "step": 1078
+    },
+    {
+      "epoch": 0.01079,
+      "grad_norm": 0.861335694801563,
+      "learning_rate": 0.003,
+      "loss": 4.4104,
+      "step": 1079
+    },
+    {
+      "epoch": 0.0108,
+      "grad_norm": 0.6640978224934152,
+      "learning_rate": 0.003,
+      "loss": 4.3777,
+      "step": 1080
+    },
+    {
+      "epoch": 0.01081,
+      "grad_norm": 0.6684756830200824,
+      "learning_rate": 0.003,
+      "loss": 4.3836,
+      "step": 1081
+    },
+    {
+      "epoch": 0.01082,
+      "grad_norm": 0.7577090594132565,
+      "learning_rate": 0.003,
+      "loss": 4.4054,
+      "step": 1082
+    },
+    {
+      "epoch": 0.01083,
+      "grad_norm": 0.8160609336744046,
+      "learning_rate": 0.003,
+      "loss": 4.3924,
+      "step": 1083
+    },
+    {
+      "epoch": 0.01084,
+      "grad_norm": 0.7897303142858512,
+      "learning_rate": 0.003,
+      "loss": 4.4011,
+      "step": 1084
+    },
+    {
+      "epoch": 0.01085,
+      "grad_norm": 0.7064792552725174,
+      "learning_rate": 0.003,
+      "loss": 4.3862,
+      "step": 1085
+    },
+    {
+      "epoch": 0.01086,
+      "grad_norm": 0.6262724475441191,
+      "learning_rate": 0.003,
+      "loss": 4.3571,
+      "step": 1086
+    },
+    {
+      "epoch": 0.01087,
+      "grad_norm": 0.6391781173229293,
+      "learning_rate": 0.003,
+      "loss": 4.3729,
+      "step": 1087
+    },
+    {
+      "epoch": 0.01088,
+      "grad_norm": 0.6632808145329404,
+      "learning_rate": 0.003,
+      "loss": 4.4092,
+      "step": 1088
+    },
+    {
+      "epoch": 0.01089,
+      "grad_norm": 0.6686792714887265,
+      "learning_rate": 0.003,
+      "loss": 4.3783,
+      "step": 1089
+    },
+    {
+      "epoch": 0.0109,
+      "grad_norm": 0.5882408044346112,
+      "learning_rate": 0.003,
+      "loss": 4.3992,
+      "step": 1090
+    },
+    {
+      "epoch": 0.01091,
+      "grad_norm": 0.5565191786424966,
+      "learning_rate": 0.003,
+      "loss": 4.3662,
+      "step": 1091
+    },
+    {
+      "epoch": 0.01092,
+      "grad_norm": 0.571165259700944,
+      "learning_rate": 0.003,
+      "loss": 4.3627,
+      "step": 1092
+    },
+    {
+      "epoch": 0.01093,
+      "grad_norm": 0.5483152737763334,
+      "learning_rate": 0.003,
+      "loss": 4.379,
+      "step": 1093
+    },
+    {
+      "epoch": 0.01094,
+      "grad_norm": 0.4816093735859256,
+      "learning_rate": 0.003,
+      "loss": 4.3664,
+      "step": 1094
+    },
+    {
+      "epoch": 0.01095,
+      "grad_norm": 0.4607091969369753,
+      "learning_rate": 0.003,
+      "loss": 4.3913,
+      "step": 1095
+    },
+    {
+      "epoch": 0.01096,
+      "grad_norm": 0.5948450571982326,
+      "learning_rate": 0.003,
+      "loss": 4.3662,
+      "step": 1096
+    },
+    {
+      "epoch": 0.01097,
+      "grad_norm": 0.8991911041937081,
+      "learning_rate": 0.003,
+      "loss": 4.3648,
+      "step": 1097
+    },
+    {
+      "epoch": 0.01098,
+      "grad_norm": 1.00898451325596,
+      "learning_rate": 0.003,
+      "loss": 4.4173,
+      "step": 1098
+    },
+    {
+      "epoch": 0.01099,
+      "grad_norm": 0.837940726063793,
+      "learning_rate": 0.003,
+      "loss": 4.3702,
+      "step": 1099
+    },
+    {
+      "epoch": 0.011,
+      "grad_norm": 1.0135265311826804,
+      "learning_rate": 0.003,
+      "loss": 4.3911,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01101,
+      "grad_norm": 1.0865706640064363,
+      "learning_rate": 0.003,
+      "loss": 4.4129,
+      "step": 1101
+    },
+    {
+      "epoch": 0.01102,
+      "grad_norm": 0.9407674508321733,
+      "learning_rate": 0.003,
+      "loss": 4.4141,
+      "step": 1102
+    },
+    {
+      "epoch": 0.01103,
+      "grad_norm": 0.7826637713653176,
+      "learning_rate": 0.003,
+      "loss": 4.4004,
+      "step": 1103
+    },
+    {
+      "epoch": 0.01104,
+      "grad_norm": 0.6792138170427816,
+      "learning_rate": 0.003,
+      "loss": 4.3864,
+      "step": 1104
+    },
+    {
+      "epoch": 0.01105,
+      "grad_norm": 0.7494779633947588,
+      "learning_rate": 0.003,
+      "loss": 4.4098,
+      "step": 1105
+    },
+    {
+      "epoch": 0.01106,
+      "grad_norm": 0.7074202272250578,
+      "learning_rate": 0.003,
+      "loss": 4.3837,
+      "step": 1106
+    },
+    {
+      "epoch": 0.01107,
+      "grad_norm": 0.6111631455372859,
+      "learning_rate": 0.003,
+      "loss": 4.3903,
+      "step": 1107
+    },
+    {
+      "epoch": 0.01108,
+      "grad_norm": 0.5958647482211449,
+      "learning_rate": 0.003,
+      "loss": 4.3565,
+      "step": 1108
+    },
+    {
+      "epoch": 0.01109,
+      "grad_norm": 0.615214127583215,
+      "learning_rate": 0.003,
+      "loss": 4.3726,
+      "step": 1109
+    },
+    {
+      "epoch": 0.0111,
+      "grad_norm": 0.5327988413049348,
+      "learning_rate": 0.003,
+      "loss": 4.3878,
+      "step": 1110
+    },
+    {
+      "epoch": 0.01111,
+      "grad_norm": 0.47023927407071675,
+      "learning_rate": 0.003,
+      "loss": 4.3622,
+      "step": 1111
+    },
+    {
+      "epoch": 0.01112,
+      "grad_norm": 0.3776559700056012,
+      "learning_rate": 0.003,
+      "loss": 4.3759,
+      "step": 1112
+    },
+    {
+      "epoch": 0.01113,
+      "grad_norm": 0.3763967476164199,
+      "learning_rate": 0.003,
+      "loss": 4.3414,
+      "step": 1113
+    },
+    {
+      "epoch": 0.01114,
+      "grad_norm": 0.3564238802994098,
+      "learning_rate": 0.003,
+      "loss": 4.354,
+      "step": 1114
+    },
+    {
+      "epoch": 0.01115,
+      "grad_norm": 0.30679188146183123,
+      "learning_rate": 0.003,
+      "loss": 4.3753,
+      "step": 1115
+    },
+    {
+      "epoch": 0.01116,
+      "grad_norm": 0.33853412768388663,
+      "learning_rate": 0.003,
+      "loss": 4.3633,
+      "step": 1116
+    },
+    {
+      "epoch": 0.01117,
+      "grad_norm": 0.3107462307737106,
+      "learning_rate": 0.003,
+      "loss": 4.3492,
+      "step": 1117
+    },
+    {
+      "epoch": 0.01118,
+      "grad_norm": 0.32797709961339483,
+      "learning_rate": 0.003,
+      "loss": 4.352,
+      "step": 1118
+    },
+    {
+      "epoch": 0.01119,
+      "grad_norm": 0.3013823503547338,
+      "learning_rate": 0.003,
+      "loss": 4.3543,
+      "step": 1119
+    },
+    {
+      "epoch": 0.0112,
+      "grad_norm": 0.3198307315980734,
+      "learning_rate": 0.003,
+      "loss": 4.3378,
+      "step": 1120
+    },
+    {
+      "epoch": 0.01121,
+      "grad_norm": 0.33595869744145906,
+      "learning_rate": 0.003,
+      "loss": 4.3478,
+      "step": 1121
+    },
+    {
+      "epoch": 0.01122,
+      "grad_norm": 0.4236485736167927,
+      "learning_rate": 0.003,
+      "loss": 4.3259,
+      "step": 1122
+    },
+    {
+      "epoch": 0.01123,
+      "grad_norm": 0.5207438928810102,
+      "learning_rate": 0.003,
+      "loss": 4.3165,
+      "step": 1123
+    },
+    {
+      "epoch": 0.01124,
+      "grad_norm": 0.5897496309635056,
+      "learning_rate": 0.003,
+      "loss": 4.3639,
+      "step": 1124
+    },
+    {
+      "epoch": 0.01125,
+      "grad_norm": 0.6251226930864733,
+      "learning_rate": 0.003,
+      "loss": 4.3351,
+      "step": 1125
+    },
+    {
+      "epoch": 0.01126,
+      "grad_norm": 0.7904920259399445,
+      "learning_rate": 0.003,
+      "loss": 4.3485,
+      "step": 1126
+    },
+    {
+      "epoch": 0.01127,
+      "grad_norm": 0.8706338403024567,
+      "learning_rate": 0.003,
+      "loss": 4.3878,
+      "step": 1127
+    },
+    {
+      "epoch": 0.01128,
+      "grad_norm": 0.7520394017234492,
+      "learning_rate": 0.003,
+      "loss": 4.3618,
+      "step": 1128
+    },
+    {
+      "epoch": 0.01129,
+      "grad_norm": 0.9434772506575799,
+      "learning_rate": 0.003,
+      "loss": 4.3617,
+      "step": 1129
+    },
+    {
+      "epoch": 0.0113,
+      "grad_norm": 0.9515853773144796,
+      "learning_rate": 0.003,
+      "loss": 4.3566,
+      "step": 1130
+    },
+    {
+      "epoch": 0.01131,
+      "grad_norm": 1.1049329865545139,
+      "learning_rate": 0.003,
+      "loss": 4.3775,
+      "step": 1131
+    },
+    {
+      "epoch": 0.01132,
+      "grad_norm": 0.869968096155467,
+      "learning_rate": 0.003,
+      "loss": 4.3991,
+      "step": 1132
+    },
+    {
+      "epoch": 0.01133,
+      "grad_norm": 0.7539769321914868,
+      "learning_rate": 0.003,
+      "loss": 4.3532,
+      "step": 1133
+    },
+    {
+      "epoch": 0.01134,
+      "grad_norm": 0.8275263694532653,
+      "learning_rate": 0.003,
+      "loss": 4.3686,
+      "step": 1134
+    },
+    {
+      "epoch": 0.01135,
+      "grad_norm": 0.8617681055772174,
+      "learning_rate": 0.003,
+      "loss": 4.402,
+      "step": 1135
+    },
+    {
+      "epoch": 0.01136,
+      "grad_norm": 0.7602421394000872,
+      "learning_rate": 0.003,
+      "loss": 4.3721,
+      "step": 1136
+    },
+    {
+      "epoch": 0.01137,
+      "grad_norm": 0.7778258812220352,
+      "learning_rate": 0.003,
+      "loss": 4.3963,
+      "step": 1137
+    },
+    {
+      "epoch": 0.01138,
+      "grad_norm": 0.7594339489071814,
+      "learning_rate": 0.003,
+      "loss": 4.3534,
+      "step": 1138
+    },
+    {
+      "epoch": 0.01139,
+      "grad_norm": 0.7671596268779747,
+      "learning_rate": 0.003,
+      "loss": 4.3643,
+      "step": 1139
+    },
+    {
+      "epoch": 0.0114,
+      "grad_norm": 0.8353688690035959,
+      "learning_rate": 0.003,
+      "loss": 4.3455,
+      "step": 1140
+    },
+    {
+      "epoch": 0.01141,
+      "grad_norm": 0.9712894348927099,
+      "learning_rate": 0.003,
+      "loss": 4.3818,
+      "step": 1141
+    },
+    {
+      "epoch": 0.01142,
+      "grad_norm": 0.8605012156156574,
+      "learning_rate": 0.003,
+      "loss": 4.355,
+      "step": 1142
+    },
+    {
+      "epoch": 0.01143,
+      "grad_norm": 0.7447870609491655,
+      "learning_rate": 0.003,
+      "loss": 4.3674,
+      "step": 1143
+    },
+    {
+      "epoch": 0.01144,
+      "grad_norm": 0.8620443864386994,
+      "learning_rate": 0.003,
+      "loss": 4.339,
+      "step": 1144
+    },
+    {
+      "epoch": 0.01145,
+      "grad_norm": 0.9495119843583,
+      "learning_rate": 0.003,
+      "loss": 4.3837,
+      "step": 1145
+    },
+    {
+      "epoch": 0.01146,
+      "grad_norm": 0.8719913828537735,
+      "learning_rate": 0.003,
+      "loss": 4.3682,
+      "step": 1146
+    },
+    {
+      "epoch": 0.01147,
+      "grad_norm": 0.8406491643071615,
+      "learning_rate": 0.003,
+      "loss": 4.3599,
+      "step": 1147
+    },
+    {
+      "epoch": 0.01148,
+      "grad_norm": 0.7299655375485881,
+      "learning_rate": 0.003,
+      "loss": 4.3549,
+      "step": 1148
+    },
+    {
+      "epoch": 0.01149,
+      "grad_norm": 0.7470410614101604,
+      "learning_rate": 0.003,
+      "loss": 4.3727,
+      "step": 1149
+    },
+    {
+      "epoch": 0.0115,
+      "grad_norm": 0.7232429634523785,
+      "learning_rate": 0.003,
+      "loss": 4.364,
+      "step": 1150
+    },
+    {
+      "epoch": 0.01151,
+      "grad_norm": 0.6295150712699281,
+      "learning_rate": 0.003,
+      "loss": 4.3586,
+      "step": 1151
+    },
+    {
+      "epoch": 0.01152,
+      "grad_norm": 0.652544707149974,
+      "learning_rate": 0.003,
+      "loss": 4.3661,
+      "step": 1152
+    },
+    {
+      "epoch": 0.01153,
+      "grad_norm": 0.5683271703897544,
+      "learning_rate": 0.003,
+      "loss": 4.3775,
+      "step": 1153
+    },
+    {
+      "epoch": 0.01154,
+      "grad_norm": 0.535300431807245,
+      "learning_rate": 0.003,
+      "loss": 4.3515,
+      "step": 1154
+    },
+    {
+      "epoch": 0.01155,
+      "grad_norm": 0.4766589820931375,
+      "learning_rate": 0.003,
+      "loss": 4.3377,
+      "step": 1155
+    },
+    {
+      "epoch": 0.01156,
+      "grad_norm": 0.5530856703722951,
+      "learning_rate": 0.003,
+      "loss": 4.342,
+      "step": 1156
+    },
+    {
+      "epoch": 0.01157,
+      "grad_norm": 0.6154599631908627,
+      "learning_rate": 0.003,
+      "loss": 4.374,
+      "step": 1157
+    },
+    {
+      "epoch": 0.01158,
+      "grad_norm": 0.693370036269422,
+      "learning_rate": 0.003,
+      "loss": 4.3417,
+      "step": 1158
+    },
+    {
+      "epoch": 0.01159,
+      "grad_norm": 0.7413452595288457,
+      "learning_rate": 0.003,
+      "loss": 4.3455,
+      "step": 1159
+    },
+    {
+      "epoch": 0.0116,
+      "grad_norm": 0.7301771780442605,
+      "learning_rate": 0.003,
+      "loss": 4.3545,
+      "step": 1160
+    },
+    {
+      "epoch": 0.01161,
+      "grad_norm": 0.8159054162369931,
+      "learning_rate": 0.003,
+      "loss": 4.3334,
+      "step": 1161
+    },
+    {
+      "epoch": 0.01162,
+      "grad_norm": 0.7760523716092679,
+      "learning_rate": 0.003,
+      "loss": 4.3486,
+      "step": 1162
+    },
+    {
+      "epoch": 0.01163,
+      "grad_norm": 0.6309863142169387,
+      "learning_rate": 0.003,
+      "loss": 4.36,
+      "step": 1163
+    },
+    {
+      "epoch": 0.01164,
+      "grad_norm": 0.559610446270928,
+      "learning_rate": 0.003,
+      "loss": 4.3468,
+      "step": 1164
+    },
+    {
+      "epoch": 0.01165,
+      "grad_norm": 0.5932728337340286,
+      "learning_rate": 0.003,
+      "loss": 4.3443,
+      "step": 1165
+    },
+    {
+      "epoch": 0.01166,
+      "grad_norm": 0.47536439541847164,
+      "learning_rate": 0.003,
+      "loss": 4.3373,
+      "step": 1166
+    },
+    {
+      "epoch": 0.01167,
+      "grad_norm": 0.4935599339579411,
+      "learning_rate": 0.003,
+      "loss": 4.3054,
+      "step": 1167
+    },
+    {
+      "epoch": 0.01168,
+      "grad_norm": 0.530157986351205,
+      "learning_rate": 0.003,
+      "loss": 4.3136,
+      "step": 1168
+    },
+    {
+      "epoch": 0.01169,
+      "grad_norm": 0.4039160553718301,
+      "learning_rate": 0.003,
+      "loss": 4.3497,
+      "step": 1169
+    },
+    {
+      "epoch": 0.0117,
+      "grad_norm": 0.3887795730112425,
+      "learning_rate": 0.003,
+      "loss": 4.3331,
+      "step": 1170
+    },
+    {
+      "epoch": 0.01171,
+      "grad_norm": 0.44146129518485994,
+      "learning_rate": 0.003,
+      "loss": 4.3186,
+      "step": 1171
+    },
+    {
+      "epoch": 0.01172,
+      "grad_norm": 0.3842074037767514,
+      "learning_rate": 0.003,
+      "loss": 4.3032,
+      "step": 1172
+    },
+    {
+      "epoch": 0.01173,
+      "grad_norm": 0.3741626855737148,
+      "learning_rate": 0.003,
+      "loss": 4.3368,
+      "step": 1173
+    },
+    {
+      "epoch": 0.01174,
+      "grad_norm": 0.41728859284602393,
+      "learning_rate": 0.003,
+      "loss": 4.3131,
+      "step": 1174
+    },
+    {
+      "epoch": 0.01175,
+      "grad_norm": 0.41981830032719913,
+      "learning_rate": 0.003,
+      "loss": 4.2965,
+      "step": 1175
+    },
+    {
+      "epoch": 0.01176,
+      "grad_norm": 0.37957919503111737,
+      "learning_rate": 0.003,
+      "loss": 4.3296,
+      "step": 1176
+    },
+    {
+      "epoch": 0.01177,
+      "grad_norm": 0.3992995566049025,
+      "learning_rate": 0.003,
+      "loss": 4.2992,
+      "step": 1177
+    },
+    {
+      "epoch": 0.01178,
+      "grad_norm": 0.4902741144335086,
+      "learning_rate": 0.003,
+      "loss": 4.3211,
+      "step": 1178
+    },
+    {
+      "epoch": 0.01179,
+      "grad_norm": 0.6458507017424667,
+      "learning_rate": 0.003,
+      "loss": 4.3023,
+      "step": 1179
+    },
+    {
+      "epoch": 0.0118,
+      "grad_norm": 0.7925849610797445,
+      "learning_rate": 0.003,
+      "loss": 4.3757,
+      "step": 1180
+    },
+    {
+      "epoch": 0.01181,
+      "grad_norm": 0.8606498974404746,
+      "learning_rate": 0.003,
+      "loss": 4.3669,
+      "step": 1181
+    },
+    {
+      "epoch": 0.01182,
+      "grad_norm": 0.7553862661379784,
+      "learning_rate": 0.003,
+      "loss": 4.3259,
+      "step": 1182
+    },
+    {
+      "epoch": 0.01183,
+      "grad_norm": 0.7241455767877882,
+      "learning_rate": 0.003,
+      "loss": 4.3159,
+      "step": 1183
+    },
+    {
+      "epoch": 0.01184,
+      "grad_norm": 0.6945084653960217,
+      "learning_rate": 0.003,
+      "loss": 4.3593,
+      "step": 1184
+    },
+    {
+      "epoch": 0.01185,
+      "grad_norm": 0.5809073742888989,
+      "learning_rate": 0.003,
+      "loss": 4.3238,
+      "step": 1185
+    },
+    {
+      "epoch": 0.01186,
+      "grad_norm": 0.556060401057652,
+      "learning_rate": 0.003,
+      "loss": 4.3461,
+      "step": 1186
+    },
+    {
+      "epoch": 0.01187,
+      "grad_norm": 0.6174052114657497,
+      "learning_rate": 0.003,
+      "loss": 4.3108,
+      "step": 1187
+    },
+    {
+      "epoch": 0.01188,
+      "grad_norm": 0.721410101268809,
+      "learning_rate": 0.003,
+      "loss": 4.3288,
+      "step": 1188
+    },
+    {
+      "epoch": 0.01189,
+      "grad_norm": 0.7947776232413513,
+      "learning_rate": 0.003,
+      "loss": 4.3281,
+      "step": 1189
+    },
+    {
+      "epoch": 0.0119,
+      "grad_norm": 0.882572116743252,
+      "learning_rate": 0.003,
+      "loss": 4.3193,
+      "step": 1190
+    },
+    {
+      "epoch": 0.01191,
+      "grad_norm": 1.0146227461981676,
+      "learning_rate": 0.003,
+      "loss": 4.307,
+      "step": 1191
+    },
+    {
+      "epoch": 0.01192,
+      "grad_norm": 1.112972484482218,
+      "learning_rate": 0.003,
+      "loss": 4.3512,
+      "step": 1192
+    },
+    {
+      "epoch": 0.01193,
+      "grad_norm": 0.9615449743933697,
+      "learning_rate": 0.003,
+      "loss": 4.3541,
+      "step": 1193
+    },
+    {
+      "epoch": 0.01194,
+      "grad_norm": 0.8644326468862364,
+      "learning_rate": 0.003,
+      "loss": 4.3507,
+      "step": 1194
+    },
+    {
+      "epoch": 0.01195,
+      "grad_norm": 0.8480336292760144,
+      "learning_rate": 0.003,
+      "loss": 4.3317,
+      "step": 1195
+    },
+    {
+      "epoch": 0.01196,
+      "grad_norm": 0.8784381650624298,
+      "learning_rate": 0.003,
+      "loss": 4.3413,
+      "step": 1196
+    },
+    {
+      "epoch": 0.01197,
+      "grad_norm": 0.9498997917047696,
+      "learning_rate": 0.003,
+      "loss": 4.3186,
+      "step": 1197
+    },
+    {
+      "epoch": 0.01198,
+      "grad_norm": 0.8920952388458802,
+      "learning_rate": 0.003,
+      "loss": 4.388,
+      "step": 1198
+    },
+    {
+      "epoch": 0.01199,
+      "grad_norm": 1.0871047324767358,
+      "learning_rate": 0.003,
+      "loss": 4.3742,
+      "step": 1199
+    },
+    {
+      "epoch": 0.012,
+      "grad_norm": 0.9019122174295907,
+      "learning_rate": 0.003,
+      "loss": 4.3579,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01201,
+      "grad_norm": 0.85321834247841,
+      "learning_rate": 0.003,
+      "loss": 4.3858,
+      "step": 1201
+    },
+    {
+      "epoch": 0.01202,
+      "grad_norm": 0.7349368268324239,
+      "learning_rate": 0.003,
+      "loss": 4.3505,
+      "step": 1202
+    },
+    {
+      "epoch": 0.01203,
+      "grad_norm": 0.6774471285416843,
+      "learning_rate": 0.003,
+      "loss": 4.357,
+      "step": 1203
+    },
+    {
+      "epoch": 0.01204,
+      "grad_norm": 0.6937988439444178,
+      "learning_rate": 0.003,
+      "loss": 4.3513,
+      "step": 1204
+    },
+    {
+      "epoch": 0.01205,
+      "grad_norm": 0.6364064699328508,
+      "learning_rate": 0.003,
+      "loss": 4.3406,
+      "step": 1205
+    },
+    {
+      "epoch": 0.01206,
+      "grad_norm": 0.522276421270424,
+      "learning_rate": 0.003,
+      "loss": 4.3798,
+      "step": 1206
+    },
+    {
+      "epoch": 0.01207,
+      "grad_norm": 0.5076942128127889,
+      "learning_rate": 0.003,
+      "loss": 4.2994,
+      "step": 1207
+    },
+    {
+      "epoch": 0.01208,
+      "grad_norm": 0.47287636588909837,
+      "learning_rate": 0.003,
+      "loss": 4.3565,
+      "step": 1208
+    },
+    {
+      "epoch": 0.01209,
+      "grad_norm": 0.5041956058683869,
+      "learning_rate": 0.003,
+      "loss": 4.3379,
+      "step": 1209
+    },
+    {
+      "epoch": 0.0121,
+      "grad_norm": 0.6167670848439076,
+      "learning_rate": 0.003,
+      "loss": 4.341,
+      "step": 1210
+    },
+    {
+      "epoch": 0.01211,
+      "grad_norm": 0.7242996454774496,
+      "learning_rate": 0.003,
+      "loss": 4.3379,
+      "step": 1211
+    },
+    {
+      "epoch": 0.01212,
+      "grad_norm": 0.8213908511974465,
+      "learning_rate": 0.003,
+      "loss": 4.3624,
+      "step": 1212
+    },
+    {
+      "epoch": 0.01213,
+      "grad_norm": 0.6880698765717139,
+      "learning_rate": 0.003,
+      "loss": 4.3342,
+      "step": 1213
+    },
+    {
+      "epoch": 0.01214,
+      "grad_norm": 0.5407616002767914,
+      "learning_rate": 0.003,
+      "loss": 4.311,
+      "step": 1214
+    },
+    {
+      "epoch": 0.01215,
+      "grad_norm": 0.6707389680302347,
+      "learning_rate": 0.003,
+      "loss": 4.3237,
+      "step": 1215
+    },
+    {
+      "epoch": 0.01216,
+      "grad_norm": 0.715155480014794,
+      "learning_rate": 0.003,
+      "loss": 4.3206,
+      "step": 1216
+    },
+    {
+      "epoch": 0.01217,
+      "grad_norm": 0.6669611805526361,
+      "learning_rate": 0.003,
+      "loss": 4.3242,
+      "step": 1217
+    },
+    {
+      "epoch": 0.01218,
+      "grad_norm": 0.6194536332440173,
+      "learning_rate": 0.003,
+      "loss": 4.3183,
+      "step": 1218
+    },
+    {
+      "epoch": 0.01219,
+      "grad_norm": 0.5624325800012369,
+      "learning_rate": 0.003,
+      "loss": 4.3151,
+      "step": 1219
+    },
+    {
+      "epoch": 0.0122,
+      "grad_norm": 0.5994752597130036,
+      "learning_rate": 0.003,
+      "loss": 4.3082,
+      "step": 1220
+    },
+    {
+      "epoch": 0.01221,
+      "grad_norm": 0.5929964237485116,
+      "learning_rate": 0.003,
+      "loss": 4.3217,
+      "step": 1221
+    },
+    {
+      "epoch": 0.01222,
+      "grad_norm": 0.5364789869662485,
+      "learning_rate": 0.003,
+      "loss": 4.3034,
+      "step": 1222
+    },
+    {
+      "epoch": 0.01223,
+      "grad_norm": 0.4883935817063237,
+      "learning_rate": 0.003,
+      "loss": 4.2989,
+      "step": 1223
+    },
+    {
+      "epoch": 0.01224,
+      "grad_norm": 0.5365327424367556,
+      "learning_rate": 0.003,
+      "loss": 4.2844,
+      "step": 1224
+    },
+    {
+      "epoch": 0.01225,
+      "grad_norm": 0.5060580384449018,
+      "learning_rate": 0.003,
+      "loss": 4.3078,
+      "step": 1225
+    },
+    {
+      "epoch": 0.01226,
+      "grad_norm": 0.4316062800492973,
+      "learning_rate": 0.003,
+      "loss": 4.297,
+      "step": 1226
+    },
+    {
+      "epoch": 0.01227,
+      "grad_norm": 0.41324487305946883,
+      "learning_rate": 0.003,
+      "loss": 4.3243,
+      "step": 1227
+    },
+    {
+      "epoch": 0.01228,
+      "grad_norm": 0.4686302520601058,
+      "learning_rate": 0.003,
+      "loss": 4.3085,
+      "step": 1228
+    },
+    {
+      "epoch": 0.01229,
+      "grad_norm": 0.5830158408620507,
+      "learning_rate": 0.003,
+      "loss": 4.2989,
+      "step": 1229
+    },
+    {
+      "epoch": 0.0123,
+      "grad_norm": 0.7143816453247004,
+      "learning_rate": 0.003,
+      "loss": 4.3393,
+      "step": 1230
+    },
+    {
+      "epoch": 0.01231,
+      "grad_norm": 0.8518087843200909,
+      "learning_rate": 0.003,
+      "loss": 4.3134,
+      "step": 1231
+    },
+    {
+      "epoch": 0.01232,
+      "grad_norm": 0.8467846529950723,
+      "learning_rate": 0.003,
+      "loss": 4.3028,
+      "step": 1232
+    },
+    {
+      "epoch": 0.01233,
+      "grad_norm": 0.7843462879273329,
+      "learning_rate": 0.003,
+      "loss": 4.3052,
+      "step": 1233
+    },
+    {
+      "epoch": 0.01234,
+      "grad_norm": 0.8956533482473612,
+      "learning_rate": 0.003,
+      "loss": 4.3298,
+      "step": 1234
+    },
+    {
+      "epoch": 0.01235,
+      "grad_norm": 0.9162978759093423,
+      "learning_rate": 0.003,
+      "loss": 4.3266,
+      "step": 1235
+    },
+    {
+      "epoch": 0.01236,
+      "grad_norm": 0.860682215697324,
+      "learning_rate": 0.003,
+      "loss": 4.3299,
+      "step": 1236
+    },
+    {
+      "epoch": 0.01237,
+      "grad_norm": 0.851315880486755,
+      "learning_rate": 0.003,
+      "loss": 4.3393,
+      "step": 1237
+    },
+    {
+      "epoch": 0.01238,
+      "grad_norm": 0.8294971616591855,
+      "learning_rate": 0.003,
+      "loss": 4.3358,
+      "step": 1238
+    },
+    {
+      "epoch": 0.01239,
+      "grad_norm": 0.9437199934028814,
+      "learning_rate": 0.003,
+      "loss": 4.3282,
+      "step": 1239
+    },
+    {
+      "epoch": 0.0124,
+      "grad_norm": 0.9106084651998833,
+      "learning_rate": 0.003,
+      "loss": 4.3137,
+      "step": 1240
+    },
+    {
+      "epoch": 0.01241,
+      "grad_norm": 0.8351496076806352,
+      "learning_rate": 0.003,
+      "loss": 4.3273,
+      "step": 1241
+    },
+    {
+      "epoch": 0.01242,
+      "grad_norm": 0.6918670619381783,
+      "learning_rate": 0.003,
+      "loss": 4.3257,
+      "step": 1242
+    },
+    {
+      "epoch": 0.01243,
+      "grad_norm": 0.732924073441867,
+      "learning_rate": 0.003,
+      "loss": 4.3077,
+      "step": 1243
+    },
+    {
+      "epoch": 0.01244,
+      "grad_norm": 0.7515551346375118,
+      "learning_rate": 0.003,
+      "loss": 4.3219,
+      "step": 1244
+    },
+    {
+      "epoch": 0.01245,
+      "grad_norm": 0.635919850996669,
+      "learning_rate": 0.003,
+      "loss": 4.3108,
+      "step": 1245
+    },
+    {
+      "epoch": 0.01246,
+      "grad_norm": 0.5245627080366145,
+      "learning_rate": 0.003,
+      "loss": 4.2806,
+      "step": 1246
+    },
+    {
+      "epoch": 0.01247,
+      "grad_norm": 0.5964323822303592,
+      "learning_rate": 0.003,
+      "loss": 4.3378,
+      "step": 1247
+    },
+    {
+      "epoch": 0.01248,
+      "grad_norm": 0.7016619320460757,
+      "learning_rate": 0.003,
+      "loss": 4.3217,
+      "step": 1248
+    },
+    {
+      "epoch": 0.01249,
+      "grad_norm": 0.7231849999092693,
+      "learning_rate": 0.003,
+      "loss": 4.3154,
+      "step": 1249
+    },
+    {
+      "epoch": 0.0125,
+      "grad_norm": 0.6036388669727554,
+      "learning_rate": 0.003,
+      "loss": 4.2869,
+      "step": 1250
+    },
+    {
+      "epoch": 0.01251,
+      "grad_norm": 0.5253366712272899,
+      "learning_rate": 0.003,
+      "loss": 4.316,
+      "step": 1251
+    },
+    {
+      "epoch": 0.01252,
+      "grad_norm": 0.46320485886810614,
+      "learning_rate": 0.003,
+      "loss": 4.2912,
+      "step": 1252
+    },
+    {
+      "epoch": 0.01253,
+      "grad_norm": 0.4963582371471178,
+      "learning_rate": 0.003,
+      "loss": 4.2694,
+      "step": 1253
+    },
+    {
+      "epoch": 0.01254,
+      "grad_norm": 0.5081054207596929,
+      "learning_rate": 0.003,
+      "loss": 4.3157,
+      "step": 1254
+    },
+    {
+      "epoch": 0.01255,
+      "grad_norm": 0.4761071055102706,
+      "learning_rate": 0.003,
+      "loss": 4.2641,
+      "step": 1255
+    },
+    {
+      "epoch": 0.01256,
+      "grad_norm": 0.5232678892632563,
+      "learning_rate": 0.003,
+      "loss": 4.2916,
+      "step": 1256
+    },
+    {
+      "epoch": 0.01257,
+      "grad_norm": 0.5344373546996498,
+      "learning_rate": 0.003,
+      "loss": 4.2768,
+      "step": 1257
+    },
+    {
+      "epoch": 0.01258,
+      "grad_norm": 0.5773429526145795,
+      "learning_rate": 0.003,
+      "loss": 4.2867,
+      "step": 1258
+    },
+    {
+      "epoch": 0.01259,
+      "grad_norm": 0.6584062985843472,
+      "learning_rate": 0.003,
+      "loss": 4.2918,
+      "step": 1259
+    },
+    {
+      "epoch": 0.0126,
+      "grad_norm": 0.6804145752695209,
+      "learning_rate": 0.003,
+      "loss": 4.3214,
+      "step": 1260
+    },
+    {
+      "epoch": 0.01261,
+      "grad_norm": 0.6981998021120679,
+      "learning_rate": 0.003,
+      "loss": 4.3161,
+      "step": 1261
+    },
+    {
+      "epoch": 0.01262,
+      "grad_norm": 0.6112143338587285,
+      "learning_rate": 0.003,
+      "loss": 4.304,
+      "step": 1262
+    },
+    {
+      "epoch": 0.01263,
+      "grad_norm": 0.5927786151181257,
+      "learning_rate": 0.003,
+      "loss": 4.298,
+      "step": 1263
+    },
+    {
+      "epoch": 0.01264,
+      "grad_norm": 0.6346760589695591,
+      "learning_rate": 0.003,
+      "loss": 4.2851,
+      "step": 1264
+    },
+    {
+      "epoch": 0.01265,
+      "grad_norm": 0.6927824976704505,
+      "learning_rate": 0.003,
+      "loss": 4.3011,
+      "step": 1265
+    },
+    {
+      "epoch": 0.01266,
+      "grad_norm": 0.7328319392289776,
+      "learning_rate": 0.003,
+      "loss": 4.3026,
+      "step": 1266
+    },
+    {
+      "epoch": 0.01267,
+      "grad_norm": 0.8489062737553393,
+      "learning_rate": 0.003,
+      "loss": 4.325,
+      "step": 1267
+    },
+    {
+      "epoch": 0.01268,
+      "grad_norm": 0.8961579939511539,
+      "learning_rate": 0.003,
+      "loss": 4.3257,
+      "step": 1268
+    },
+    {
+      "epoch": 0.01269,
+      "grad_norm": 0.821810765966473,
+      "learning_rate": 0.003,
+      "loss": 4.2839,
+      "step": 1269
+    },
+    {
+      "epoch": 0.0127,
+      "grad_norm": 0.8357784457532415,
+      "learning_rate": 0.003,
+      "loss": 4.3228,
+      "step": 1270
+    },
+    {
+      "epoch": 0.01271,
+      "grad_norm": 0.7922720998753914,
+      "learning_rate": 0.003,
+      "loss": 4.3037,
+      "step": 1271
+    },
+    {
+      "epoch": 0.01272,
+      "grad_norm": 0.7964991232532974,
+      "learning_rate": 0.003,
+      "loss": 4.3039,
+      "step": 1272
+    },
+    {
+      "epoch": 0.01273,
+      "grad_norm": 0.7778280400168709,
+      "learning_rate": 0.003,
+      "loss": 4.3248,
+      "step": 1273
+    },
+    {
+      "epoch": 0.01274,
+      "grad_norm": 0.9153345412853071,
+      "learning_rate": 0.003,
+      "loss": 4.3241,
+      "step": 1274
+    },
+    {
+      "epoch": 0.01275,
+      "grad_norm": 0.8587735334056731,
+      "learning_rate": 0.003,
+      "loss": 4.294,
+      "step": 1275
+    },
+    {
+      "epoch": 0.01276,
+      "grad_norm": 0.7198180114679126,
+      "learning_rate": 0.003,
+      "loss": 4.2973,
+      "step": 1276
+    },
+    {
+      "epoch": 0.01277,
+      "grad_norm": 0.7510923187630398,
+      "learning_rate": 0.003,
+      "loss": 4.3066,
+      "step": 1277
+    },
+    {
+      "epoch": 0.01278,
+      "grad_norm": 0.8054501129326099,
+      "learning_rate": 0.003,
+      "loss": 4.3179,
+      "step": 1278
+    },
+    {
+      "epoch": 0.01279,
+      "grad_norm": 0.7824154610465377,
+      "learning_rate": 0.003,
+      "loss": 4.2997,
+      "step": 1279
+    },
+    {
+      "epoch": 0.0128,
+      "grad_norm": 0.7320824999573285,
+      "learning_rate": 0.003,
+      "loss": 4.2956,
+      "step": 1280
+    },
+    {
+      "epoch": 0.01281,
+      "grad_norm": 0.6960943615110375,
+      "learning_rate": 0.003,
+      "loss": 4.2806,
+      "step": 1281
+    },
+    {
+      "epoch": 0.01282,
+      "grad_norm": 0.5992318655887974,
+      "learning_rate": 0.003,
+      "loss": 4.3205,
+      "step": 1282
+    },
+    {
+      "epoch": 0.01283,
+      "grad_norm": 0.6115671413316082,
+      "learning_rate": 0.003,
+      "loss": 4.2898,
+      "step": 1283
+    },
+    {
+      "epoch": 0.01284,
+      "grad_norm": 0.5685168810152473,
+      "learning_rate": 0.003,
+      "loss": 4.2893,
+      "step": 1284
+    },
+    {
+      "epoch": 0.01285,
+      "grad_norm": 0.6350368869509793,
+      "learning_rate": 0.003,
+      "loss": 4.2923,
+      "step": 1285
+    },
+    {
+      "epoch": 0.01286,
+      "grad_norm": 0.6769385854863653,
+      "learning_rate": 0.003,
+      "loss": 4.3189,
+      "step": 1286
+    },
+    {
+      "epoch": 0.01287,
+      "grad_norm": 0.5997038647975563,
+      "learning_rate": 0.003,
+      "loss": 4.2792,
+      "step": 1287
+    },
+    {
+      "epoch": 0.01288,
+      "grad_norm": 0.6318526861304625,
+      "learning_rate": 0.003,
+      "loss": 4.2923,
+      "step": 1288
+    },
+    {
+      "epoch": 0.01289,
+      "grad_norm": 0.6619003980657734,
+      "learning_rate": 0.003,
+      "loss": 4.2877,
+      "step": 1289
+    },
+    {
+      "epoch": 0.0129,
+      "grad_norm": 0.7141266768182228,
+      "learning_rate": 0.003,
+      "loss": 4.3104,
+      "step": 1290
+    },
+    {
+      "epoch": 0.01291,
+      "grad_norm": 0.7063834987249501,
+      "learning_rate": 0.003,
+      "loss": 4.2921,
+      "step": 1291
+    },
+    {
+      "epoch": 0.01292,
+      "grad_norm": 0.6423787800334868,
+      "learning_rate": 0.003,
+      "loss": 4.3242,
+      "step": 1292
+    },
+    {
+      "epoch": 0.01293,
+      "grad_norm": 0.5344471091426154,
+      "learning_rate": 0.003,
+      "loss": 4.3062,
+      "step": 1293
+    },
+    {
+      "epoch": 0.01294,
+      "grad_norm": 0.4523910877254715,
+      "learning_rate": 0.003,
+      "loss": 4.2707,
+      "step": 1294
+    },
+    {
+      "epoch": 0.01295,
+      "grad_norm": 0.5040604805545252,
+      "learning_rate": 0.003,
+      "loss": 4.2993,
+      "step": 1295
+    },
+    {
+      "epoch": 0.01296,
+      "grad_norm": 0.5193740138388458,
+      "learning_rate": 0.003,
+      "loss": 4.2656,
+      "step": 1296
+    },
+    {
+      "epoch": 0.01297,
+      "grad_norm": 0.5346893098464323,
+      "learning_rate": 0.003,
+      "loss": 4.2788,
+      "step": 1297
+    },
+    {
+      "epoch": 0.01298,
+      "grad_norm": 0.5373039682823991,
+      "learning_rate": 0.003,
+      "loss": 4.2592,
+      "step": 1298
+    },
+    {
+      "epoch": 0.01299,
+      "grad_norm": 0.5661548725913075,
+      "learning_rate": 0.003,
+      "loss": 4.2709,
+      "step": 1299
+    },
+    {
+      "epoch": 0.013,
+      "grad_norm": 0.7446298448290246,
+      "learning_rate": 0.003,
+      "loss": 4.2678,
+      "step": 1300
+    },
+    {
+      "epoch": 0.01301,
+      "grad_norm": 0.8230150822772082,
+      "learning_rate": 0.003,
+      "loss": 4.2955,
+      "step": 1301
+    },
+    {
+      "epoch": 0.01302,
+      "grad_norm": 0.7744841619477817,
+      "learning_rate": 0.003,
+      "loss": 4.3361,
+      "step": 1302
+    },
+    {
+      "epoch": 0.01303,
+      "grad_norm": 0.6454504186361478,
+      "learning_rate": 0.003,
+      "loss": 4.2765,
+      "step": 1303
+    },
+    {
+      "epoch": 0.01304,
+      "grad_norm": 0.7188710255689805,
+      "learning_rate": 0.003,
+      "loss": 4.3063,
+      "step": 1304
+    },
+    {
+      "epoch": 0.01305,
+      "grad_norm": 0.757402061055745,
+      "learning_rate": 0.003,
+      "loss": 4.2938,
+      "step": 1305
+    },
+    {
+      "epoch": 0.01306,
+      "grad_norm": 0.691272304301314,
+      "learning_rate": 0.003,
+      "loss": 4.2864,
+      "step": 1306
+    },
+    {
+      "epoch": 0.01307,
+      "grad_norm": 0.706053416610055,
+      "learning_rate": 0.003,
+      "loss": 4.3074,
+      "step": 1307
+    },
+    {
+      "epoch": 0.01308,
+      "grad_norm": 0.6221880747565551,
+      "learning_rate": 0.003,
+      "loss": 4.284,
+      "step": 1308
+    },
+    {
+      "epoch": 0.01309,
+      "grad_norm": 0.5791770340268083,
+      "learning_rate": 0.003,
+      "loss": 4.3174,
+      "step": 1309
+    },
+    {
+      "epoch": 0.0131,
+      "grad_norm": 0.6695730992498613,
+      "learning_rate": 0.003,
+      "loss": 4.2672,
+      "step": 1310
+    },
+    {
+      "epoch": 0.01311,
+      "grad_norm": 0.6922028781625468,
+      "learning_rate": 0.003,
+      "loss": 4.3073,
+      "step": 1311
+    },
+    {
+      "epoch": 0.01312,
+      "grad_norm": 0.8287735129364245,
+      "learning_rate": 0.003,
+      "loss": 4.2832,
+      "step": 1312
+    },
+    {
+      "epoch": 0.01313,
+      "grad_norm": 0.7976431888385418,
+      "learning_rate": 0.003,
+      "loss": 4.2757,
+      "step": 1313
+    },
+    {
+      "epoch": 0.01314,
+      "grad_norm": 0.7220518823683293,
+      "learning_rate": 0.003,
+      "loss": 4.2709,
+      "step": 1314
+    },
+    {
+      "epoch": 0.01315,
+      "grad_norm": 0.6095265673664224,
+      "learning_rate": 0.003,
+      "loss": 4.2914,
+      "step": 1315
+    },
+    {
+      "epoch": 0.01316,
+      "grad_norm": 0.7674466841738673,
+      "learning_rate": 0.003,
+      "loss": 4.2855,
+      "step": 1316
+    },
+    {
+      "epoch": 0.01317,
+      "grad_norm": 0.9059745922778419,
+      "learning_rate": 0.003,
+      "loss": 4.3151,
+      "step": 1317
+    },
+    {
+      "epoch": 0.01318,
+      "grad_norm": 0.9427302613153309,
+      "learning_rate": 0.003,
+      "loss": 4.3056,
+      "step": 1318
+    },
+    {
+      "epoch": 0.01319,
+      "grad_norm": 0.73054171411267,
+      "learning_rate": 0.003,
+      "loss": 4.3015,
+      "step": 1319
+    },
+    {
+      "epoch": 0.0132,
+      "grad_norm": 0.7111720819817223,
+      "learning_rate": 0.003,
+      "loss": 4.3007,
+      "step": 1320
+    },
+    {
+      "epoch": 0.01321,
+      "grad_norm": 0.7456349593949091,
+      "learning_rate": 0.003,
+      "loss": 4.2768,
+      "step": 1321
+    },
+    {
+      "epoch": 0.01322,
+      "grad_norm": 0.7807986643070355,
+      "learning_rate": 0.003,
+      "loss": 4.3113,
+      "step": 1322
+    },
+    {
+      "epoch": 0.01323,
+      "grad_norm": 0.8224389080614981,
+      "learning_rate": 0.003,
+      "loss": 4.3093,
+      "step": 1323
+    },
+    {
+      "epoch": 0.01324,
+      "grad_norm": 1.035353941682105,
+      "learning_rate": 0.003,
+      "loss": 4.3086,
+      "step": 1324
+    },
+    {
+      "epoch": 0.01325,
+      "grad_norm": 0.8946051092509474,
+      "learning_rate": 0.003,
+      "loss": 4.3063,
+      "step": 1325
+    },
+    {
+      "epoch": 0.01326,
+      "grad_norm": 1.0201089618117705,
+      "learning_rate": 0.003,
+      "loss": 4.2973,
+      "step": 1326
+    },
+    {
+      "epoch": 0.01327,
+      "grad_norm": 0.864867336973391,
+      "learning_rate": 0.003,
+      "loss": 4.3115,
+      "step": 1327
+    },
+    {
+      "epoch": 0.01328,
+      "grad_norm": 0.8025779319028616,
+      "learning_rate": 0.003,
+      "loss": 4.3034,
+      "step": 1328
+    },
+    {
+      "epoch": 0.01329,
+      "grad_norm": 0.7807799111244325,
+      "learning_rate": 0.003,
+      "loss": 4.2681,
+      "step": 1329
+    },
+    {
+      "epoch": 0.0133,
+      "grad_norm": 0.8681894612953762,
+      "learning_rate": 0.003,
+      "loss": 4.3154,
+      "step": 1330
+    },
+    {
+      "epoch": 0.01331,
+      "grad_norm": 1.0494856392086225,
+      "learning_rate": 0.003,
+      "loss": 4.3161,
+      "step": 1331
+    },
+    {
+      "epoch": 0.01332,
+      "grad_norm": 0.9504785674947578,
+      "learning_rate": 0.003,
+      "loss": 4.3364,
+      "step": 1332
+    },
+    {
+      "epoch": 0.01333,
+      "grad_norm": 0.9898834831271559,
+      "learning_rate": 0.003,
+      "loss": 4.3084,
+      "step": 1333
+    },
+    {
+      "epoch": 0.01334,
+      "grad_norm": 1.0019143105022317,
+      "learning_rate": 0.003,
+      "loss": 4.3012,
+      "step": 1334
+    },
+    {
+      "epoch": 0.01335,
+      "grad_norm": 0.8650334140122408,
+      "learning_rate": 0.003,
+      "loss": 4.3214,
+      "step": 1335
+    },
+    {
+      "epoch": 0.01336,
+      "grad_norm": 0.8920064299561147,
+      "learning_rate": 0.003,
+      "loss": 4.317,
+      "step": 1336
+    },
+    {
+      "epoch": 0.01337,
+      "grad_norm": 0.7707840908187834,
+      "learning_rate": 0.003,
+      "loss": 4.3286,
+      "step": 1337
+    },
+    {
+      "epoch": 0.01338,
+      "grad_norm": 0.7632598332017204,
+      "learning_rate": 0.003,
+      "loss": 4.2919,
+      "step": 1338
+    },
+    {
+      "epoch": 0.01339,
+      "grad_norm": 0.7360297385886754,
+      "learning_rate": 0.003,
+      "loss": 4.3155,
+      "step": 1339
+    },
+    {
+      "epoch": 0.0134,
+      "grad_norm": 0.6003622832803963,
+      "learning_rate": 0.003,
+      "loss": 4.2916,
+      "step": 1340
+    },
+    {
+      "epoch": 0.01341,
+      "grad_norm": 0.6955896552188154,
+      "learning_rate": 0.003,
+      "loss": 4.3105,
+      "step": 1341
+    },
+    {
+      "epoch": 0.01342,
+      "grad_norm": 0.6883046873976916,
+      "learning_rate": 0.003,
+      "loss": 4.2818,
+      "step": 1342
+    },
+    {
+      "epoch": 0.01343,
+      "grad_norm": 0.7284558823650665,
+      "learning_rate": 0.003,
+      "loss": 4.2883,
+      "step": 1343
+    },
+    {
+      "epoch": 0.01344,
+      "grad_norm": 0.6360126374822531,
+      "learning_rate": 0.003,
+      "loss": 4.2837,
+      "step": 1344
+    },
+    {
+      "epoch": 0.01345,
+      "grad_norm": 0.5334484786655853,
+      "learning_rate": 0.003,
+      "loss": 4.3176,
+      "step": 1345
+    },
+    {
+      "epoch": 0.01346,
+      "grad_norm": 0.5681507381379014,
+      "learning_rate": 0.003,
+      "loss": 4.2885,
+      "step": 1346
+    },
+    {
+      "epoch": 0.01347,
+      "grad_norm": 0.5788404528903132,
+      "learning_rate": 0.003,
+      "loss": 4.2816,
+      "step": 1347
+    },
+    {
+      "epoch": 0.01348,
+      "grad_norm": 0.5418128304616573,
+      "learning_rate": 0.003,
+      "loss": 4.2755,
+      "step": 1348
+    },
+    {
+      "epoch": 0.01349,
+      "grad_norm": 0.525198484250559,
+      "learning_rate": 0.003,
+      "loss": 4.2954,
+      "step": 1349
+    },
+    {
+      "epoch": 0.0135,
+      "grad_norm": 0.46326210720927236,
+      "learning_rate": 0.003,
+      "loss": 4.2853,
+      "step": 1350
+    },
+    {
+      "epoch": 0.01351,
+      "grad_norm": 0.4585121858352072,
+      "learning_rate": 0.003,
+      "loss": 4.313,
+      "step": 1351
+    },
+    {
+      "epoch": 0.01352,
+      "grad_norm": 0.4283890963121618,
+      "learning_rate": 0.003,
+      "loss": 4.286,
+      "step": 1352
+    },
+    {
+      "epoch": 0.01353,
+      "grad_norm": 0.42199420882026434,
+      "learning_rate": 0.003,
+      "loss": 4.2702,
+      "step": 1353
+    },
+    {
+      "epoch": 0.01354,
+      "grad_norm": 0.4262744973588792,
+      "learning_rate": 0.003,
+      "loss": 4.2568,
+      "step": 1354
+    },
+    {
+      "epoch": 0.01355,
+      "grad_norm": 0.40850743293980507,
+      "learning_rate": 0.003,
+      "loss": 4.268,
+      "step": 1355
+    },
+    {
+      "epoch": 0.01356,
+      "grad_norm": 0.41818294724190413,
+      "learning_rate": 0.003,
+      "loss": 4.2623,
+      "step": 1356
+    },
+    {
+      "epoch": 0.01357,
+      "grad_norm": 0.43655941768902196,
+      "learning_rate": 0.003,
+      "loss": 4.2503,
+      "step": 1357
+    },
+    {
+      "epoch": 0.01358,
+      "grad_norm": 0.49395890148457533,
+      "learning_rate": 0.003,
+      "loss": 4.2463,
+      "step": 1358
+    },
+    {
+      "epoch": 0.01359,
+      "grad_norm": 0.616978373290081,
+      "learning_rate": 0.003,
+      "loss": 4.2894,
+      "step": 1359
+    },
+    {
+      "epoch": 0.0136,
+      "grad_norm": 0.735374833430507,
+      "learning_rate": 0.003,
+      "loss": 4.2639,
+      "step": 1360
+    },
+    {
+      "epoch": 0.01361,
+      "grad_norm": 0.7389266502027623,
+      "learning_rate": 0.003,
+      "loss": 4.2792,
+      "step": 1361
+    },
+    {
+      "epoch": 0.01362,
+      "grad_norm": 0.65177426874564,
+      "learning_rate": 0.003,
+      "loss": 4.2865,
+      "step": 1362
+    },
+    {
+      "epoch": 0.01363,
+      "grad_norm": 0.6753488054632895,
+      "learning_rate": 0.003,
+      "loss": 4.2893,
+      "step": 1363
+    },
+    {
+      "epoch": 0.01364,
+      "grad_norm": 0.8240837082437891,
+      "learning_rate": 0.003,
+      "loss": 4.2868,
+      "step": 1364
+    },
+    {
+      "epoch": 0.01365,
+      "grad_norm": 0.7993265568553571,
+      "learning_rate": 0.003,
+      "loss": 4.306,
+      "step": 1365
+    },
+    {
+      "epoch": 0.01366,
+      "grad_norm": 0.6564545276995603,
+      "learning_rate": 0.003,
+      "loss": 4.2558,
+      "step": 1366
+    },
+    {
+      "epoch": 0.01367,
+      "grad_norm": 0.5493737530398193,
+      "learning_rate": 0.003,
+      "loss": 4.3003,
+      "step": 1367
+    },
+    {
+      "epoch": 0.01368,
+      "grad_norm": 0.5650653386363494,
+      "learning_rate": 0.003,
+      "loss": 4.2739,
+      "step": 1368
+    },
+    {
+      "epoch": 0.01369,
+      "grad_norm": 0.46328750272215335,
+      "learning_rate": 0.003,
+      "loss": 4.262,
+      "step": 1369
+    },
+    {
+      "epoch": 0.0137,
+      "grad_norm": 0.5045609042462222,
+      "learning_rate": 0.003,
+      "loss": 4.2894,
+      "step": 1370
+    },
+    {
+      "epoch": 0.01371,
+      "grad_norm": 0.5910891998788811,
+      "learning_rate": 0.003,
+      "loss": 4.2822,
+      "step": 1371
+    },
+    {
+      "epoch": 0.01372,
+      "grad_norm": 0.6815189005820225,
+      "learning_rate": 0.003,
+      "loss": 4.2381,
+      "step": 1372
+    },
+    {
+      "epoch": 0.01373,
+      "grad_norm": 0.842580155535958,
+      "learning_rate": 0.003,
+      "loss": 4.2808,
+      "step": 1373
+    },
+    {
+      "epoch": 0.01374,
+      "grad_norm": 0.907805736475784,
+      "learning_rate": 0.003,
+      "loss": 4.2961,
+      "step": 1374
+    },
+    {
+      "epoch": 0.01375,
+      "grad_norm": 0.8312986426917952,
+      "learning_rate": 0.003,
+      "loss": 4.2534,
+      "step": 1375
+    },
+    {
+      "epoch": 0.01376,
+      "grad_norm": 0.8401650089095936,
+      "learning_rate": 0.003,
+      "loss": 4.2829,
+      "step": 1376
+    },
+    {
+      "epoch": 0.01377,
+      "grad_norm": 0.7706178388137035,
+      "learning_rate": 0.003,
+      "loss": 4.2816,
+      "step": 1377
+    },
+    {
+      "epoch": 0.01378,
+      "grad_norm": 0.6957781157082459,
+      "learning_rate": 0.003,
+      "loss": 4.3087,
+      "step": 1378
+    },
+    {
+      "epoch": 0.01379,
+      "grad_norm": 0.6831406156786277,
+      "learning_rate": 0.003,
+      "loss": 4.2792,
+      "step": 1379
+    },
+    {
+      "epoch": 0.0138,
+      "grad_norm": 0.6334610013963772,
+      "learning_rate": 0.003,
+      "loss": 4.2739,
+      "step": 1380
+    },
+    {
+      "epoch": 0.01381,
+      "grad_norm": 0.5316939383075773,
+      "learning_rate": 0.003,
+      "loss": 4.2776,
+      "step": 1381
+    },
+    {
+      "epoch": 0.01382,
+      "grad_norm": 0.516573615018121,
+      "learning_rate": 0.003,
+      "loss": 4.2735,
+      "step": 1382
+    },
+    {
+      "epoch": 0.01383,
+      "grad_norm": 0.5599212453399529,
+      "learning_rate": 0.003,
+      "loss": 4.2692,
+      "step": 1383
+    },
+    {
+      "epoch": 0.01384,
+      "grad_norm": 0.6188520650392525,
+      "learning_rate": 0.003,
+      "loss": 4.3,
+      "step": 1384
+    },
+    {
+      "epoch": 0.01385,
+      "grad_norm": 0.6353409140192112,
+      "learning_rate": 0.003,
+      "loss": 4.2877,
+      "step": 1385
+    },
+    {
+      "epoch": 0.01386,
+      "grad_norm": 0.6356880916095538,
+      "learning_rate": 0.003,
+      "loss": 4.2638,
+      "step": 1386
+    },
+    {
+      "epoch": 0.01387,
+      "grad_norm": 0.7257450023455765,
+      "learning_rate": 0.003,
+      "loss": 4.2537,
+      "step": 1387
+    },
+    {
+      "epoch": 0.01388,
+      "grad_norm": 0.7014765592615619,
+      "learning_rate": 0.003,
+      "loss": 4.2914,
+      "step": 1388
+    },
+    {
+      "epoch": 0.01389,
+      "grad_norm": 0.6616359972565842,
+      "learning_rate": 0.003,
+      "loss": 4.2702,
+      "step": 1389
+    },
+    {
+      "epoch": 0.0139,
+      "grad_norm": 0.6527699070789852,
+      "learning_rate": 0.003,
+      "loss": 4.2735,
+      "step": 1390
+    },
+    {
+      "epoch": 0.01391,
+      "grad_norm": 0.7095964540484884,
+      "learning_rate": 0.003,
+      "loss": 4.2741,
+      "step": 1391
+    },
+    {
+      "epoch": 0.01392,
+      "grad_norm": 0.7501817269895354,
+      "learning_rate": 0.003,
+      "loss": 4.2866,
+      "step": 1392
+    },
+    {
+      "epoch": 0.01393,
+      "grad_norm": 0.6714529168989051,
+      "learning_rate": 0.003,
+      "loss": 4.2555,
+      "step": 1393
+    },
+    {
+      "epoch": 0.01394,
+      "grad_norm": 0.5871737080135582,
+      "learning_rate": 0.003,
+      "loss": 4.2392,
+      "step": 1394
+    },
+    {
+      "epoch": 0.01395,
+      "grad_norm": 0.5621623507740485,
+      "learning_rate": 0.003,
+      "loss": 4.2747,
+      "step": 1395
+    },
+    {
+      "epoch": 0.01396,
+      "grad_norm": 0.664320619088776,
+      "learning_rate": 0.003,
+      "loss": 4.2655,
+      "step": 1396
+    },
+    {
+      "epoch": 0.01397,
+      "grad_norm": 0.7274104246733604,
+      "learning_rate": 0.003,
+      "loss": 4.2817,
+      "step": 1397
+    },
+    {
+      "epoch": 0.01398,
+      "grad_norm": 0.6424657996436192,
+      "learning_rate": 0.003,
+      "loss": 4.2594,
+      "step": 1398
+    },
+    {
+      "epoch": 0.01399,
+      "grad_norm": 0.6842824738024023,
+      "learning_rate": 0.003,
+      "loss": 4.2829,
+      "step": 1399
+    },
+    {
+      "epoch": 0.014,
+      "grad_norm": 0.8261095933984154,
+      "learning_rate": 0.003,
+      "loss": 4.2493,
+      "step": 1400
+    },
+    {
+      "epoch": 0.01401,
+      "grad_norm": 0.7965575457737929,
+      "learning_rate": 0.003,
+      "loss": 4.2708,
+      "step": 1401
+    },
+    {
+      "epoch": 0.01402,
+      "grad_norm": 0.8026102237656736,
+      "learning_rate": 0.003,
+      "loss": 4.2842,
+      "step": 1402
+    },
+    {
+      "epoch": 0.01403,
+      "grad_norm": 0.816192334350639,
+      "learning_rate": 0.003,
+      "loss": 4.2628,
+      "step": 1403
+    },
+    {
+      "epoch": 0.01404,
+      "grad_norm": 0.9312442614599608,
+      "learning_rate": 0.003,
+      "loss": 4.2897,
+      "step": 1404
+    },
+    {
+      "epoch": 0.01405,
+      "grad_norm": 1.0392313893683864,
+      "learning_rate": 0.003,
+      "loss": 4.319,
+      "step": 1405
+    },
+    {
+      "epoch": 0.01406,
+      "grad_norm": 0.9393883719208338,
+      "learning_rate": 0.003,
+      "loss": 4.3,
+      "step": 1406
+    },
+    {
+      "epoch": 0.01407,
+      "grad_norm": 0.9978258942739143,
+      "learning_rate": 0.003,
+      "loss": 4.3193,
+      "step": 1407
+    },
+    {
+      "epoch": 0.01408,
+      "grad_norm": 1.1500652805383735,
+      "learning_rate": 0.003,
+      "loss": 4.2914,
+      "step": 1408
+    },
+    {
+      "epoch": 0.01409,
+      "grad_norm": 1.0900231626852614,
+      "learning_rate": 0.003,
+      "loss": 4.3253,
+      "step": 1409
+    },
+    {
+      "epoch": 0.0141,
+      "grad_norm": 1.0202054584780536,
+      "learning_rate": 0.003,
+      "loss": 4.3009,
+      "step": 1410
+    },
+    {
+      "epoch": 0.01411,
+      "grad_norm": 0.898714028166491,
+      "learning_rate": 0.003,
+      "loss": 4.2744,
+      "step": 1411
+    },
+    {
+      "epoch": 0.01412,
+      "grad_norm": 0.7435025676246113,
+      "learning_rate": 0.003,
+      "loss": 4.3061,
+      "step": 1412
+    },
+    {
+      "epoch": 0.01413,
+      "grad_norm": 0.5873930770772416,
+      "learning_rate": 0.003,
+      "loss": 4.2941,
+      "step": 1413
+    },
+    {
+      "epoch": 0.01414,
+      "grad_norm": 0.6016255593464158,
+      "learning_rate": 0.003,
+      "loss": 4.2747,
+      "step": 1414
+    },
+    {
+      "epoch": 0.01415,
+      "grad_norm": 0.6099426868013017,
+      "learning_rate": 0.003,
+      "loss": 4.2659,
+      "step": 1415
+    },
+    {
+      "epoch": 0.01416,
+      "grad_norm": 0.5575813775095378,
+      "learning_rate": 0.003,
+      "loss": 4.2911,
+      "step": 1416
+    },
+    {
+      "epoch": 0.01417,
+      "grad_norm": 0.5917623365800833,
+      "learning_rate": 0.003,
+      "loss": 4.2803,
+      "step": 1417
+    },
+    {
+      "epoch": 0.01418,
+      "grad_norm": 0.6068864838347083,
+      "learning_rate": 0.003,
+      "loss": 4.2958,
+      "step": 1418
+    },
+    {
+      "epoch": 0.01419,
+      "grad_norm": 0.6824165624145411,
+      "learning_rate": 0.003,
+      "loss": 4.2858,
+      "step": 1419
+    },
+    {
+      "epoch": 0.0142,
+      "grad_norm": 0.6199413471532862,
+      "learning_rate": 0.003,
+      "loss": 4.291,
+      "step": 1420
+    },
+    {
+      "epoch": 0.01421,
+      "grad_norm": 0.5408211469998263,
+      "learning_rate": 0.003,
+      "loss": 4.2544,
+      "step": 1421
+    },
+    {
+      "epoch": 0.01422,
+      "grad_norm": 0.5533737134579111,
+      "learning_rate": 0.003,
+      "loss": 4.2598,
+      "step": 1422
+    },
+    {
+      "epoch": 0.01423,
+      "grad_norm": 0.5685842699001438,
+      "learning_rate": 0.003,
+      "loss": 4.2645,
+      "step": 1423
+    },
+    {
+      "epoch": 0.01424,
+      "grad_norm": 0.6304522666735822,
+      "learning_rate": 0.003,
+      "loss": 4.2833,
+      "step": 1424
+    },
+    {
+      "epoch": 0.01425,
+      "grad_norm": 0.6515865244102063,
+      "learning_rate": 0.003,
+      "loss": 4.2865,
+      "step": 1425
+    },
+    {
+      "epoch": 0.01426,
+      "grad_norm": 0.7444754270308325,
+      "learning_rate": 0.003,
+      "loss": 4.2655,
+      "step": 1426
+    },
+    {
+      "epoch": 0.01427,
+      "grad_norm": 0.7497304563870452,
+      "learning_rate": 0.003,
+      "loss": 4.263,
+      "step": 1427
+    },
+    {
+      "epoch": 0.01428,
+      "grad_norm": 0.6556010515324793,
+      "learning_rate": 0.003,
+      "loss": 4.2594,
+      "step": 1428
+    },
+    {
+      "epoch": 0.01429,
+      "grad_norm": 0.6636248907749417,
+      "learning_rate": 0.003,
+      "loss": 4.265,
+      "step": 1429
+    },
+    {
+      "epoch": 0.0143,
+      "grad_norm": 0.7211105035251077,
+      "learning_rate": 0.003,
+      "loss": 4.28,
+      "step": 1430
+    },
+    {
+      "epoch": 0.01431,
+      "grad_norm": 0.7061472665079888,
+      "learning_rate": 0.003,
+      "loss": 4.2684,
+      "step": 1431
+    },
+    {
+      "epoch": 0.01432,
+      "grad_norm": 0.6456431394575474,
+      "learning_rate": 0.003,
+      "loss": 4.2556,
+      "step": 1432
+    },
+    {
+      "epoch": 0.01433,
+      "grad_norm": 0.7570512378286692,
+      "learning_rate": 0.003,
+      "loss": 4.2568,
+      "step": 1433
+    },
+    {
+      "epoch": 0.01434,
+      "grad_norm": 0.7648500642290317,
+      "learning_rate": 0.003,
+      "loss": 4.2443,
+      "step": 1434
+    },
+    {
+      "epoch": 0.01435,
+      "grad_norm": 0.7689156199684593,
+      "learning_rate": 0.003,
+      "loss": 4.2531,
+      "step": 1435
+    },
+    {
+      "epoch": 0.01436,
+      "grad_norm": 0.7360234649948437,
+      "learning_rate": 0.003,
+      "loss": 4.274,
+      "step": 1436
+    },
+    {
+      "epoch": 0.01437,
+      "grad_norm": 0.7697472519317102,
+      "learning_rate": 0.003,
+      "loss": 4.2879,
+      "step": 1437
+    },
+    {
+      "epoch": 0.01438,
+      "grad_norm": 0.7889422694170278,
+      "learning_rate": 0.003,
+      "loss": 4.2805,
+      "step": 1438
+    },
+    {
+      "epoch": 0.01439,
+      "grad_norm": 0.6891331546216416,
+      "learning_rate": 0.003,
+      "loss": 4.2417,
+      "step": 1439
+    },
+    {
+      "epoch": 0.0144,
+      "grad_norm": 0.560460778999846,
+      "learning_rate": 0.003,
+      "loss": 4.2508,
+      "step": 1440
+    },
+    {
+      "epoch": 0.01441,
+      "grad_norm": 0.5400270352929513,
+      "learning_rate": 0.003,
+      "loss": 4.2578,
+      "step": 1441
+    },
+    {
+      "epoch": 0.01442,
+      "grad_norm": 0.4911140953080599,
+      "learning_rate": 0.003,
+      "loss": 4.2391,
+      "step": 1442
+    },
+    {
+      "epoch": 0.01443,
+      "grad_norm": 0.45294215274078276,
+      "learning_rate": 0.003,
+      "loss": 4.2993,
+      "step": 1443
+    },
+    {
+      "epoch": 0.01444,
+      "grad_norm": 0.4224101633170518,
+      "learning_rate": 0.003,
+      "loss": 4.2618,
+      "step": 1444
+    },
+    {
+      "epoch": 0.01445,
+      "grad_norm": 0.3961185894381269,
+      "learning_rate": 0.003,
+      "loss": 4.2206,
+      "step": 1445
+    },
+    {
+      "epoch": 0.01446,
+      "grad_norm": 0.33118667900370186,
+      "learning_rate": 0.003,
+      "loss": 4.2343,
+      "step": 1446
+    },
+    {
+      "epoch": 0.01447,
+      "grad_norm": 0.3556733761768107,
+      "learning_rate": 0.003,
+      "loss": 4.2433,
+      "step": 1447
+    },
+    {
+      "epoch": 0.01448,
+      "grad_norm": 0.36561455130036685,
+      "learning_rate": 0.003,
+      "loss": 4.2519,
+      "step": 1448
+    },
+    {
+      "epoch": 0.01449,
+      "grad_norm": 0.46874354720560246,
+      "learning_rate": 0.003,
+      "loss": 4.2461,
+      "step": 1449
+    },
+    {
+      "epoch": 0.0145,
+      "grad_norm": 0.60149549500153,
+      "learning_rate": 0.003,
+      "loss": 4.266,
+      "step": 1450
+    },
+    {
+      "epoch": 0.01451,
+      "grad_norm": 0.6696836890558183,
+      "learning_rate": 0.003,
+      "loss": 4.2469,
+      "step": 1451
+    },
+    {
+      "epoch": 0.01452,
+      "grad_norm": 0.6515350232296636,
+      "learning_rate": 0.003,
+      "loss": 4.2476,
+      "step": 1452
+    },
+    {
+      "epoch": 0.01453,
+      "grad_norm": 0.662277322649583,
+      "learning_rate": 0.003,
+      "loss": 4.2608,
+      "step": 1453
+    },
+    {
+      "epoch": 0.01454,
+      "grad_norm": 0.8972969416580211,
+      "learning_rate": 0.003,
+      "loss": 4.2374,
+      "step": 1454
+    },
+    {
+      "epoch": 0.01455,
+      "grad_norm": 1.030023068993243,
+      "learning_rate": 0.003,
+      "loss": 4.2974,
+      "step": 1455
+    },
+    {
+      "epoch": 0.01456,
+      "grad_norm": 1.0186167234807815,
+      "learning_rate": 0.003,
+      "loss": 4.2685,
+      "step": 1456
+    },
+    {
+      "epoch": 0.01457,
+      "grad_norm": 0.9067819891416041,
+      "learning_rate": 0.003,
+      "loss": 4.2556,
+      "step": 1457
+    },
+    {
+      "epoch": 0.01458,
+      "grad_norm": 0.8145250303169672,
+      "learning_rate": 0.003,
+      "loss": 4.2507,
+      "step": 1458
+    },
+    {
+      "epoch": 0.01459,
+      "grad_norm": 0.8089455723824179,
+      "learning_rate": 0.003,
+      "loss": 4.2807,
+      "step": 1459
+    },
+    {
+      "epoch": 0.0146,
+      "grad_norm": 0.7818105608289873,
+      "learning_rate": 0.003,
+      "loss": 4.2825,
+      "step": 1460
+    },
+    {
+      "epoch": 0.01461,
+      "grad_norm": 0.8037592863895617,
+      "learning_rate": 0.003,
+      "loss": 4.2628,
+      "step": 1461
+    },
+    {
+      "epoch": 0.01462,
+      "grad_norm": 0.7476671417509403,
+      "learning_rate": 0.003,
+      "loss": 4.2637,
+      "step": 1462
+    },
+    {
+      "epoch": 0.01463,
+      "grad_norm": 0.7531868665794315,
+      "learning_rate": 0.003,
+      "loss": 4.2649,
+      "step": 1463
+    },
+    {
+      "epoch": 0.01464,
+      "grad_norm": 0.7893765528220285,
+      "learning_rate": 0.003,
+      "loss": 4.2417,
+      "step": 1464
+    },
+    {
+      "epoch": 0.01465,
+      "grad_norm": 0.6572421930862311,
+      "learning_rate": 0.003,
+      "loss": 4.2823,
+      "step": 1465
+    },
+    {
+      "epoch": 0.01466,
+      "grad_norm": 0.5506048327632588,
+      "learning_rate": 0.003,
+      "loss": 4.2484,
+      "step": 1466
+    },
+    {
+      "epoch": 0.01467,
+      "grad_norm": 0.611053596724818,
+      "learning_rate": 0.003,
+      "loss": 4.2828,
+      "step": 1467
+    },
+    {
+      "epoch": 0.01468,
+      "grad_norm": 0.68776369659849,
+      "learning_rate": 0.003,
+      "loss": 4.2597,
+      "step": 1468
+    },
+    {
+      "epoch": 0.01469,
+      "grad_norm": 0.6702695663961474,
+      "learning_rate": 0.003,
+      "loss": 4.2631,
+      "step": 1469
+    },
+    {
+      "epoch": 0.0147,
+      "grad_norm": 0.6961169309212991,
+      "learning_rate": 0.003,
+      "loss": 4.2544,
+      "step": 1470
+    },
+    {
+      "epoch": 0.01471,
+      "grad_norm": 0.760155282734985,
+      "learning_rate": 0.003,
+      "loss": 4.2691,
+      "step": 1471
+    },
+    {
+      "epoch": 0.01472,
+      "grad_norm": 0.7721176095708484,
+      "learning_rate": 0.003,
+      "loss": 4.2542,
+      "step": 1472
+    },
+    {
+      "epoch": 0.01473,
+      "grad_norm": 0.7877033148700755,
+      "learning_rate": 0.003,
+      "loss": 4.2676,
+      "step": 1473
+    },
+    {
+      "epoch": 0.01474,
+      "grad_norm": 0.7464730946193943,
+      "learning_rate": 0.003,
+      "loss": 4.2539,
+      "step": 1474
+    },
+    {
+      "epoch": 0.01475,
+      "grad_norm": 0.6862863699060794,
+      "learning_rate": 0.003,
+      "loss": 4.2711,
+      "step": 1475
+    },
+    {
+      "epoch": 0.01476,
+      "grad_norm": 0.6949444506676095,
+      "learning_rate": 0.003,
+      "loss": 4.2569,
+      "step": 1476
+    },
+    {
+      "epoch": 0.01477,
+      "grad_norm": 0.6485108315252704,
+      "learning_rate": 0.003,
+      "loss": 4.2593,
+      "step": 1477
+    },
+    {
+      "epoch": 0.01478,
+      "grad_norm": 0.7249395862424638,
+      "learning_rate": 0.003,
+      "loss": 4.2474,
+      "step": 1478
+    },
+    {
+      "epoch": 0.01479,
+      "grad_norm": 0.6746802016444003,
+      "learning_rate": 0.003,
+      "loss": 4.28,
+      "step": 1479
+    },
+    {
+      "epoch": 0.0148,
+      "grad_norm": 0.6561363565990774,
+      "learning_rate": 0.003,
+      "loss": 4.2473,
+      "step": 1480
+    },
+    {
+      "epoch": 0.01481,
+      "grad_norm": 0.7984072714276804,
+      "learning_rate": 0.003,
+      "loss": 4.2564,
+      "step": 1481
+    },
+    {
+      "epoch": 0.01482,
+      "grad_norm": 0.8535873443126762,
+      "learning_rate": 0.003,
+      "loss": 4.2436,
+      "step": 1482
+    },
+    {
+      "epoch": 0.01483,
+      "grad_norm": 0.8309335851279978,
+      "learning_rate": 0.003,
+      "loss": 4.2469,
+      "step": 1483
+    },
+    {
+      "epoch": 0.01484,
+      "grad_norm": 0.7153917621202344,
+      "learning_rate": 0.003,
+      "loss": 4.2896,
+      "step": 1484
+    },
+    {
+      "epoch": 0.01485,
+      "grad_norm": 0.7644465422516641,
+      "learning_rate": 0.003,
+      "loss": 4.2586,
+      "step": 1485
+    },
+    {
+      "epoch": 0.01486,
+      "grad_norm": 0.6575547181769861,
+      "learning_rate": 0.003,
+      "loss": 4.2505,
+      "step": 1486
+    },
+    {
+      "epoch": 0.01487,
+      "grad_norm": 0.616480892995353,
+      "learning_rate": 0.003,
+      "loss": 4.2525,
+      "step": 1487
+    },
+    {
+      "epoch": 0.01488,
+      "grad_norm": 0.5676164891655556,
+      "learning_rate": 0.003,
+      "loss": 4.259,
+      "step": 1488
+    },
+    {
+      "epoch": 0.01489,
+      "grad_norm": 0.5087336805317257,
+      "learning_rate": 0.003,
+      "loss": 4.252,
+      "step": 1489
+    },
+    {
+      "epoch": 0.0149,
+      "grad_norm": 0.5711204937411694,
+      "learning_rate": 0.003,
+      "loss": 4.2449,
+      "step": 1490
+    },
+    {
+      "epoch": 0.01491,
+      "grad_norm": 0.5555383511180694,
+      "learning_rate": 0.003,
+      "loss": 4.2594,
+      "step": 1491
+    },
+    {
+      "epoch": 0.01492,
+      "grad_norm": 0.5910867309514242,
+      "learning_rate": 0.003,
+      "loss": 4.2249,
+      "step": 1492
+    },
+    {
+      "epoch": 0.01493,
+      "grad_norm": 0.5552788367316016,
+      "learning_rate": 0.003,
+      "loss": 4.2566,
+      "step": 1493
+    },
+    {
+      "epoch": 0.01494,
+      "grad_norm": 0.597504731138664,
+      "learning_rate": 0.003,
+      "loss": 4.236,
+      "step": 1494
+    },
+    {
+      "epoch": 0.01495,
+      "grad_norm": 0.7040905961754581,
+      "learning_rate": 0.003,
+      "loss": 4.2494,
+      "step": 1495
+    },
+    {
+      "epoch": 0.01496,
+      "grad_norm": 0.6504979761973873,
+      "learning_rate": 0.003,
+      "loss": 4.2253,
+      "step": 1496
+    },
+    {
+      "epoch": 0.01497,
+      "grad_norm": 0.7479904857429174,
+      "learning_rate": 0.003,
+      "loss": 4.2013,
+      "step": 1497
+    },
+    {
+      "epoch": 0.01498,
+      "grad_norm": 0.8354013055859276,
+      "learning_rate": 0.003,
+      "loss": 4.2649,
+      "step": 1498
+    },
+    {
+      "epoch": 0.01499,
+      "grad_norm": 0.9228916776476623,
+      "learning_rate": 0.003,
+      "loss": 4.2868,
+      "step": 1499
+    },
+    {
+      "epoch": 0.015,
+      "grad_norm": 0.9718172707744857,
+      "learning_rate": 0.003,
+      "loss": 4.25,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01501,
+      "grad_norm": 1.0136419160882046,
+      "learning_rate": 0.003,
+      "loss": 4.2532,
+      "step": 1501
+    },
+    {
+      "epoch": 0.01502,
+      "grad_norm": 0.9325478476141171,
+      "learning_rate": 0.003,
+      "loss": 4.2999,
+      "step": 1502
+    },
+    {
+      "epoch": 0.01503,
+      "grad_norm": 1.045999048375591,
+      "learning_rate": 0.003,
+      "loss": 4.2898,
+      "step": 1503
+    },
+    {
+      "epoch": 0.01504,
+      "grad_norm": 0.9914485824003936,
+      "learning_rate": 0.003,
+      "loss": 4.2936,
+      "step": 1504
+    },
+    {
+      "epoch": 0.01505,
+      "grad_norm": 0.915373742082242,
+      "learning_rate": 0.003,
+      "loss": 4.2776,
+      "step": 1505
+    },
+    {
+      "epoch": 0.01506,
+      "grad_norm": 0.8272684821900249,
+      "learning_rate": 0.003,
+      "loss": 4.287,
+      "step": 1506
+    },
+    {
+      "epoch": 0.01507,
+      "grad_norm": 0.8618015345967726,
+      "learning_rate": 0.003,
+      "loss": 4.305,
+      "step": 1507
+    },
+    {
+      "epoch": 0.01508,
+      "grad_norm": 0.9316744896757736,
+      "learning_rate": 0.003,
+      "loss": 4.2861,
+      "step": 1508
+    },
+    {
+      "epoch": 0.01509,
+      "grad_norm": 0.8440466174421261,
+      "learning_rate": 0.003,
+      "loss": 4.2981,
+      "step": 1509
+    },
+    {
+      "epoch": 0.0151,
+      "grad_norm": 0.7063003162706939,
+      "learning_rate": 0.003,
+      "loss": 4.2679,
+      "step": 1510
+    },
+    {
+      "epoch": 0.01511,
+      "grad_norm": 0.6428638550974116,
+      "learning_rate": 0.003,
+      "loss": 4.2518,
+      "step": 1511
+    },
+    {
+      "epoch": 0.01512,
+      "grad_norm": 0.6112249954297037,
+      "learning_rate": 0.003,
+      "loss": 4.2613,
+      "step": 1512
+    },
+    {
+      "epoch": 0.01513,
+      "grad_norm": 0.5280577640396528,
+      "learning_rate": 0.003,
+      "loss": 4.2622,
+      "step": 1513
+    },
+    {
+      "epoch": 0.01514,
+      "grad_norm": 0.5269385985194992,
+      "learning_rate": 0.003,
+      "loss": 4.2672,
+      "step": 1514
+    },
+    {
+      "epoch": 0.01515,
+      "grad_norm": 0.5487987330042349,
+      "learning_rate": 0.003,
+      "loss": 4.2562,
+      "step": 1515
+    },
+    {
+      "epoch": 0.01516,
+      "grad_norm": 0.5680391134156543,
+      "learning_rate": 0.003,
+      "loss": 4.2622,
+      "step": 1516
+    },
+    {
+      "epoch": 0.01517,
+      "grad_norm": 0.6062107860159427,
+      "learning_rate": 0.003,
+      "loss": 4.2129,
+      "step": 1517
+    },
+    {
+      "epoch": 0.01518,
+      "grad_norm": 0.6323154755891632,
+      "learning_rate": 0.003,
+      "loss": 4.2497,
+      "step": 1518
+    },
+    {
+      "epoch": 0.01519,
+      "grad_norm": 0.5798448883069626,
+      "learning_rate": 0.003,
+      "loss": 4.2808,
+      "step": 1519
+    },
+    {
+      "epoch": 0.0152,
+      "grad_norm": 0.5997731039690386,
+      "learning_rate": 0.003,
+      "loss": 4.2326,
+      "step": 1520
+    },
+    {
+      "epoch": 0.01521,
+      "grad_norm": 0.6222254423875409,
+      "learning_rate": 0.003,
+      "loss": 4.248,
+      "step": 1521
+    },
+    {
+      "epoch": 0.01522,
+      "grad_norm": 0.5290334858261236,
+      "learning_rate": 0.003,
+      "loss": 4.2398,
+      "step": 1522
+    },
+    {
+      "epoch": 0.01523,
+      "grad_norm": 0.5493394171973806,
+      "learning_rate": 0.003,
+      "loss": 4.2321,
+      "step": 1523
+    },
+    {
+      "epoch": 0.01524,
+      "grad_norm": 0.6415152277388406,
+      "learning_rate": 0.003,
+      "loss": 4.269,
+      "step": 1524
+    },
+    {
+      "epoch": 0.01525,
+      "grad_norm": 0.7027827268185114,
+      "learning_rate": 0.003,
+      "loss": 4.2694,
+      "step": 1525
+    },
+    {
+      "epoch": 0.01526,
+      "grad_norm": 0.6962686558421027,
+      "learning_rate": 0.003,
+      "loss": 4.259,
+      "step": 1526
+    },
+    {
+      "epoch": 0.01527,
+      "grad_norm": 0.6968664170247966,
+      "learning_rate": 0.003,
+      "loss": 4.2293,
+      "step": 1527
+    },
+    {
+      "epoch": 0.01528,
+      "grad_norm": 0.6258396233188714,
+      "learning_rate": 0.003,
+      "loss": 4.2456,
+      "step": 1528
+    },
+    {
+      "epoch": 0.01529,
+      "grad_norm": 0.6831052588682119,
+      "learning_rate": 0.003,
+      "loss": 4.2426,
+      "step": 1529
+    },
+    {
+      "epoch": 0.0153,
+      "grad_norm": 0.5524942784192245,
+      "learning_rate": 0.003,
+      "loss": 4.2472,
+      "step": 1530
+    },
+    {
+      "epoch": 0.01531,
+      "grad_norm": 0.5976116227060053,
+      "learning_rate": 0.003,
+      "loss": 4.241,
+      "step": 1531
+    },
+    {
+      "epoch": 0.01532,
+      "grad_norm": 0.5212214454213059,
+      "learning_rate": 0.003,
+      "loss": 4.2166,
+      "step": 1532
+    },
+    {
+      "epoch": 0.01533,
+      "grad_norm": 0.5878087168177306,
+      "learning_rate": 0.003,
+      "loss": 4.2439,
+      "step": 1533
+    },
+    {
+      "epoch": 0.01534,
+      "grad_norm": 0.8062347592746759,
+      "learning_rate": 0.003,
+      "loss": 4.2507,
+      "step": 1534
+    },
+    {
+      "epoch": 0.01535,
+      "grad_norm": 1.1825381040517529,
+      "learning_rate": 0.003,
+      "loss": 4.2824,
+      "step": 1535
+    },
+    {
+      "epoch": 0.01536,
+      "grad_norm": 0.7819176506012738,
+      "learning_rate": 0.003,
+      "loss": 4.2554,
+      "step": 1536
+    },
+    {
+      "epoch": 0.01537,
+      "grad_norm": 0.7933291638944598,
+      "learning_rate": 0.003,
+      "loss": 4.2465,
+      "step": 1537
+    },
+    {
+      "epoch": 0.01538,
+      "grad_norm": 0.7224294373390021,
+      "learning_rate": 0.003,
+      "loss": 4.2237,
+      "step": 1538
+    },
+    {
+      "epoch": 0.01539,
+      "grad_norm": 0.7725279995753968,
+      "learning_rate": 0.003,
+      "loss": 4.2589,
+      "step": 1539
+    },
+    {
+      "epoch": 0.0154,
+      "grad_norm": 0.6639445396789071,
+      "learning_rate": 0.003,
+      "loss": 4.2578,
+      "step": 1540
+    },
+    {
+      "epoch": 0.01541,
+      "grad_norm": 0.5655107354833333,
+      "learning_rate": 0.003,
+      "loss": 4.2359,
+      "step": 1541
+    },
+    {
+      "epoch": 0.01542,
+      "grad_norm": 0.4939945119276493,
+      "learning_rate": 0.003,
+      "loss": 4.2606,
+      "step": 1542
+    },
+    {
+      "epoch": 0.01543,
+      "grad_norm": 0.6098938664838743,
+      "learning_rate": 0.003,
+      "loss": 4.2469,
+      "step": 1543
+    },
+    {
+      "epoch": 0.01544,
+      "grad_norm": 0.58488864006147,
+      "learning_rate": 0.003,
+      "loss": 4.2344,
+      "step": 1544
+    },
+    {
+      "epoch": 0.01545,
+      "grad_norm": 0.5313644010442865,
+      "learning_rate": 0.003,
+      "loss": 4.2981,
+      "step": 1545
+    },
+    {
+      "epoch": 0.01546,
+      "grad_norm": 0.45959833989927285,
+      "learning_rate": 0.003,
+      "loss": 4.263,
+      "step": 1546
+    },
+    {
+      "epoch": 0.01547,
+      "grad_norm": 0.41872682800093836,
+      "learning_rate": 0.003,
+      "loss": 4.2626,
+      "step": 1547
+    },
+    {
+      "epoch": 0.01548,
+      "grad_norm": 0.4070005827014792,
+      "learning_rate": 0.003,
+      "loss": 4.2237,
+      "step": 1548
+    },
+    {
+      "epoch": 0.01549,
+      "grad_norm": 0.40332574830782875,
+      "learning_rate": 0.003,
+      "loss": 4.2125,
+      "step": 1549
+    },
+    {
+      "epoch": 0.0155,
+      "grad_norm": 0.4273425707216713,
+      "learning_rate": 0.003,
+      "loss": 4.2323,
+      "step": 1550
+    },
+    {
+      "epoch": 0.01551,
+      "grad_norm": 0.44340108675114914,
+      "learning_rate": 0.003,
+      "loss": 4.2095,
+      "step": 1551
+    },
+    {
+      "epoch": 0.01552,
+      "grad_norm": 0.4283402844721516,
+      "learning_rate": 0.003,
+      "loss": 4.2297,
+      "step": 1552
+    },
+    {
+      "epoch": 0.01553,
+      "grad_norm": 0.39909368113710497,
+      "learning_rate": 0.003,
+      "loss": 4.2548,
+      "step": 1553
+    },
+    {
+      "epoch": 0.01554,
+      "grad_norm": 0.4303761621068326,
+      "learning_rate": 0.003,
+      "loss": 4.2356,
+      "step": 1554
+    },
+    {
+      "epoch": 0.01555,
+      "grad_norm": 0.5073116437720543,
+      "learning_rate": 0.003,
+      "loss": 4.2089,
+      "step": 1555
+    },
+    {
+      "epoch": 0.01556,
+      "grad_norm": 0.6171177000269511,
+      "learning_rate": 0.003,
+      "loss": 4.2523,
+      "step": 1556
+    },
+    {
+      "epoch": 0.01557,
+      "grad_norm": 0.678464738142453,
+      "learning_rate": 0.003,
+      "loss": 4.229,
+      "step": 1557
+    },
+    {
+      "epoch": 0.01558,
+      "grad_norm": 0.6779041786729552,
+      "learning_rate": 0.003,
+      "loss": 4.22,
+      "step": 1558
+    },
+    {
+      "epoch": 0.01559,
+      "grad_norm": 0.7160170404134826,
+      "learning_rate": 0.003,
+      "loss": 4.2601,
+      "step": 1559
+    },
+    {
+      "epoch": 0.0156,
+      "grad_norm": 0.646776348851303,
+      "learning_rate": 0.003,
+      "loss": 4.2484,
+      "step": 1560
+    },
+    {
+      "epoch": 0.01561,
+      "grad_norm": 0.7362963973677504,
+      "learning_rate": 0.003,
+      "loss": 4.232,
+      "step": 1561
+    },
+    {
+      "epoch": 0.01562,
+      "grad_norm": 0.8173738140090338,
+      "learning_rate": 0.003,
+      "loss": 4.2318,
+      "step": 1562
+    },
+    {
+      "epoch": 0.01563,
+      "grad_norm": 0.6949580272753917,
+      "learning_rate": 0.003,
+      "loss": 4.2334,
+      "step": 1563
+    },
+    {
+      "epoch": 0.01564,
+      "grad_norm": 0.8704354124122435,
+      "learning_rate": 0.003,
+      "loss": 4.2705,
+      "step": 1564
+    },
+    {
+      "epoch": 0.01565,
+      "grad_norm": 1.0516715274291202,
+      "learning_rate": 0.003,
+      "loss": 4.2561,
+      "step": 1565
+    },
+    {
+      "epoch": 0.01566,
+      "grad_norm": 0.9138795915356578,
+      "learning_rate": 0.003,
+      "loss": 4.2729,
+      "step": 1566
+    },
+    {
+      "epoch": 0.01567,
+      "grad_norm": 0.7933215729025597,
+      "learning_rate": 0.003,
+      "loss": 4.2643,
+      "step": 1567
+    },
+    {
+      "epoch": 0.01568,
+      "grad_norm": 0.7524451299585192,
+      "learning_rate": 0.003,
+      "loss": 4.2455,
+      "step": 1568
+    },
+    {
+      "epoch": 0.01569,
+      "grad_norm": 0.8688441972266158,
+      "learning_rate": 0.003,
+      "loss": 4.2557,
+      "step": 1569
+    },
+    {
+      "epoch": 0.0157,
+      "grad_norm": 1.0213964087576262,
+      "learning_rate": 0.003,
+      "loss": 4.2839,
+      "step": 1570
+    },
+    {
+      "epoch": 0.01571,
+      "grad_norm": 0.9658714210834742,
+      "learning_rate": 0.003,
+      "loss": 4.2787,
+      "step": 1571
+    },
+    {
+      "epoch": 0.01572,
+      "grad_norm": 0.8962942943425795,
+      "learning_rate": 0.003,
+      "loss": 4.28,
+      "step": 1572
+    },
+    {
+      "epoch": 0.01573,
+      "grad_norm": 0.936508363757706,
+      "learning_rate": 0.003,
+      "loss": 4.2727,
+      "step": 1573
+    },
+    {
+      "epoch": 0.01574,
+      "grad_norm": 0.8583102869770243,
+      "learning_rate": 0.003,
+      "loss": 4.279,
+      "step": 1574
+    },
+    {
+      "epoch": 0.01575,
+      "grad_norm": 0.8543483448302793,
+      "learning_rate": 0.003,
+      "loss": 4.2483,
+      "step": 1575
+    },
+    {
+      "epoch": 0.01576,
+      "grad_norm": 0.6740643949431316,
+      "learning_rate": 0.003,
+      "loss": 4.2592,
+      "step": 1576
+    },
+    {
+      "epoch": 0.01577,
+      "grad_norm": 0.7068574162099283,
+      "learning_rate": 0.003,
+      "loss": 4.2234,
+      "step": 1577
+    },
+    {
+      "epoch": 0.01578,
+      "grad_norm": 0.679751091133263,
+      "learning_rate": 0.003,
+      "loss": 4.2501,
+      "step": 1578
+    },
+    {
+      "epoch": 0.01579,
+      "grad_norm": 0.6207998089836335,
+      "learning_rate": 0.003,
+      "loss": 4.2428,
+      "step": 1579
+    },
+    {
+      "epoch": 0.0158,
+      "grad_norm": 0.546999417238052,
+      "learning_rate": 0.003,
+      "loss": 4.2499,
+      "step": 1580
+    },
+    {
+      "epoch": 0.01581,
+      "grad_norm": 0.5890586432360596,
+      "learning_rate": 0.003,
+      "loss": 4.2581,
+      "step": 1581
+    },
+    {
+      "epoch": 0.01582,
+      "grad_norm": 0.6738822960276017,
+      "learning_rate": 0.003,
+      "loss": 4.2412,
+      "step": 1582
+    },
+    {
+      "epoch": 0.01583,
+      "grad_norm": 0.8449645723590827,
+      "learning_rate": 0.003,
+      "loss": 4.235,
+      "step": 1583
+    },
+    {
+      "epoch": 0.01584,
+      "grad_norm": 0.8760719049873085,
+      "learning_rate": 0.003,
+      "loss": 4.2744,
+      "step": 1584
+    },
+    {
+      "epoch": 0.01585,
+      "grad_norm": 0.8749576730978259,
+      "learning_rate": 0.003,
+      "loss": 4.2653,
+      "step": 1585
+    },
+    {
+      "epoch": 0.01586,
+      "grad_norm": 0.9817764014825603,
+      "learning_rate": 0.003,
+      "loss": 4.2423,
+      "step": 1586
+    },
+    {
+      "epoch": 0.01587,
+      "grad_norm": 0.916565337571196,
+      "learning_rate": 0.003,
+      "loss": 4.2657,
+      "step": 1587
+    },
+    {
+      "epoch": 0.01588,
+      "grad_norm": 0.5958323489289675,
+      "learning_rate": 0.003,
+      "loss": 4.2761,
+      "step": 1588
+    },
+    {
+      "epoch": 0.01589,
+      "grad_norm": 0.6040948221660047,
+      "learning_rate": 0.003,
+      "loss": 4.2694,
+      "step": 1589
+    },
+    {
+      "epoch": 0.0159,
+      "grad_norm": 0.5541532228259608,
+      "learning_rate": 0.003,
+      "loss": 4.2601,
+      "step": 1590
+    },
+    {
+      "epoch": 0.01591,
+      "grad_norm": 0.5732941807574177,
+      "learning_rate": 0.003,
+      "loss": 4.2648,
+      "step": 1591
+    },
+    {
+      "epoch": 0.01592,
+      "grad_norm": 0.6004965962049119,
+      "learning_rate": 0.003,
+      "loss": 4.2552,
+      "step": 1592
+    },
+    {
+      "epoch": 0.01593,
+      "grad_norm": 0.6750400712704605,
+      "learning_rate": 0.003,
+      "loss": 4.2608,
+      "step": 1593
+    },
+    {
+      "epoch": 0.01594,
+      "grad_norm": 0.6685590290117844,
+      "learning_rate": 0.003,
+      "loss": 4.2444,
+      "step": 1594
+    },
+    {
+      "epoch": 0.01595,
+      "grad_norm": 0.6394603041102755,
+      "learning_rate": 0.003,
+      "loss": 4.27,
+      "step": 1595
+    },
+    {
+      "epoch": 0.01596,
+      "grad_norm": 0.5525501078218852,
+      "learning_rate": 0.003,
+      "loss": 4.2582,
+      "step": 1596
+    },
+    {
+      "epoch": 0.01597,
+      "grad_norm": 0.5429531891975062,
+      "learning_rate": 0.003,
+      "loss": 4.2192,
+      "step": 1597
+    },
+    {
+      "epoch": 0.01598,
+      "grad_norm": 0.5494281969618091,
+      "learning_rate": 0.003,
+      "loss": 4.2123,
+      "step": 1598
+    },
+    {
+      "epoch": 0.01599,
+      "grad_norm": 0.48563975518620667,
+      "learning_rate": 0.003,
+      "loss": 4.2331,
+      "step": 1599
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 0.4293130690644179,
+      "learning_rate": 0.003,
+      "loss": 4.1975,
+      "step": 1600
+    },
+    {
+      "epoch": 0.01601,
+      "grad_norm": 0.4385393368882689,
+      "learning_rate": 0.003,
+      "loss": 4.2457,
+      "step": 1601
+    },
+    {
+      "epoch": 0.01602,
+      "grad_norm": 0.6149074026063737,
+      "learning_rate": 0.003,
+      "loss": 4.2243,
+      "step": 1602
+    },
+    {
+      "epoch": 0.01603,
+      "grad_norm": 0.7618943354871598,
+      "learning_rate": 0.003,
+      "loss": 4.2459,
+      "step": 1603
+    },
+    {
+      "epoch": 0.01604,
+      "grad_norm": 0.8231087133002348,
+      "learning_rate": 0.003,
+      "loss": 4.2421,
+      "step": 1604
+    },
+    {
+      "epoch": 0.01605,
+      "grad_norm": 0.792939816240345,
+      "learning_rate": 0.003,
+      "loss": 4.2225,
+      "step": 1605
+    },
+    {
+      "epoch": 0.01606,
+      "grad_norm": 0.8283386222980521,
+      "learning_rate": 0.003,
+      "loss": 4.2264,
+      "step": 1606
+    },
+    {
+      "epoch": 0.01607,
+      "grad_norm": 0.7929167281458711,
+      "learning_rate": 0.003,
+      "loss": 4.2525,
+      "step": 1607
+    },
+    {
+      "epoch": 0.01608,
+      "grad_norm": 0.8238114241929788,
+      "learning_rate": 0.003,
+      "loss": 4.2293,
+      "step": 1608
+    },
+    {
+      "epoch": 0.01609,
+      "grad_norm": 0.8482072969488592,
+      "learning_rate": 0.003,
+      "loss": 4.2448,
+      "step": 1609
+    },
+    {
+      "epoch": 0.0161,
+      "grad_norm": 0.9731734226169589,
+      "learning_rate": 0.003,
+      "loss": 4.2157,
+      "step": 1610
+    },
+    {
+      "epoch": 0.01611,
+      "grad_norm": 0.8477820620248173,
+      "learning_rate": 0.003,
+      "loss": 4.2328,
+      "step": 1611
+    },
+    {
+      "epoch": 0.01612,
+      "grad_norm": 0.8003056368656658,
+      "learning_rate": 0.003,
+      "loss": 4.2224,
+      "step": 1612
+    },
+    {
+      "epoch": 0.01613,
+      "grad_norm": 0.9384869181655229,
+      "learning_rate": 0.003,
+      "loss": 4.2759,
+      "step": 1613
+    },
+    {
+      "epoch": 0.01614,
+      "grad_norm": 1.094095683053054,
+      "learning_rate": 0.003,
+      "loss": 4.305,
+      "step": 1614
+    },
+    {
+      "epoch": 0.01615,
+      "grad_norm": 1.0454054633439653,
+      "learning_rate": 0.003,
+      "loss": 4.284,
+      "step": 1615
+    },
+    {
+      "epoch": 0.01616,
+      "grad_norm": 0.728394920780058,
+      "learning_rate": 0.003,
+      "loss": 4.2459,
+      "step": 1616
+    },
+    {
+      "epoch": 0.01617,
+      "grad_norm": 0.7455143593181587,
+      "learning_rate": 0.003,
+      "loss": 4.2373,
+      "step": 1617
+    },
+    {
+      "epoch": 0.01618,
+      "grad_norm": 0.6872936552392757,
+      "learning_rate": 0.003,
+      "loss": 4.2451,
+      "step": 1618
+    },
+    {
+      "epoch": 0.01619,
+      "grad_norm": 0.5634358260711833,
+      "learning_rate": 0.003,
+      "loss": 4.2274,
+      "step": 1619
+    },
+    {
+      "epoch": 0.0162,
+      "grad_norm": 0.5499920921999998,
+      "learning_rate": 0.003,
+      "loss": 4.231,
+      "step": 1620
+    },
+    {
+      "epoch": 0.01621,
+      "grad_norm": 0.4948415173461953,
+      "learning_rate": 0.003,
+      "loss": 4.226,
+      "step": 1621
+    },
+    {
+      "epoch": 0.01622,
+      "grad_norm": 0.45485020618430855,
+      "learning_rate": 0.003,
+      "loss": 4.2573,
+      "step": 1622
+    },
+    {
+      "epoch": 0.01623,
+      "grad_norm": 0.4128235375978694,
+      "learning_rate": 0.003,
+      "loss": 4.2306,
+      "step": 1623
+    },
+    {
+      "epoch": 0.01624,
+      "grad_norm": 0.35723113909525034,
+      "learning_rate": 0.003,
+      "loss": 4.2422,
+      "step": 1624
+    },
+    {
+      "epoch": 0.01625,
+      "grad_norm": 0.3930391853636471,
+      "learning_rate": 0.003,
+      "loss": 4.2175,
+      "step": 1625
+    },
+    {
+      "epoch": 0.01626,
+      "grad_norm": 0.3980902109371374,
+      "learning_rate": 0.003,
+      "loss": 4.2372,
+      "step": 1626
+    },
+    {
+      "epoch": 0.01627,
+      "grad_norm": 0.4192958733549867,
+      "learning_rate": 0.003,
+      "loss": 4.2057,
+      "step": 1627
+    },
+    {
+      "epoch": 0.01628,
+      "grad_norm": 0.4456736368427327,
+      "learning_rate": 0.003,
+      "loss": 4.2132,
+      "step": 1628
+    },
+    {
+      "epoch": 0.01629,
+      "grad_norm": 0.46084356058160403,
+      "learning_rate": 0.003,
+      "loss": 4.2138,
+      "step": 1629
+    },
+    {
+      "epoch": 0.0163,
+      "grad_norm": 0.4494988261781255,
+      "learning_rate": 0.003,
+      "loss": 4.2025,
+      "step": 1630
+    },
+    {
+      "epoch": 0.01631,
+      "grad_norm": 0.49049706860110326,
+      "learning_rate": 0.003,
+      "loss": 4.2248,
+      "step": 1631
+    },
+    {
+      "epoch": 0.01632,
+      "grad_norm": 0.4941434115670244,
+      "learning_rate": 0.003,
+      "loss": 4.234,
+      "step": 1632
+    },
+    {
+      "epoch": 0.01633,
+      "grad_norm": 0.5015101022129131,
+      "learning_rate": 0.003,
+      "loss": 4.2072,
+      "step": 1633
+    },
+    {
+      "epoch": 0.01634,
+      "grad_norm": 0.5355988850166129,
+      "learning_rate": 0.003,
+      "loss": 4.2264,
+      "step": 1634
+    },
+    {
+      "epoch": 0.01635,
+      "grad_norm": 0.6114378514764486,
+      "learning_rate": 0.003,
+      "loss": 4.2077,
+      "step": 1635
+    },
+    {
+      "epoch": 0.01636,
+      "grad_norm": 0.659185840591732,
+      "learning_rate": 0.003,
+      "loss": 4.2261,
+      "step": 1636
+    },
+    {
+      "epoch": 0.01637,
+      "grad_norm": 0.6890891308278232,
+      "learning_rate": 0.003,
+      "loss": 4.2122,
+      "step": 1637
+    },
+    {
+      "epoch": 0.01638,
+      "grad_norm": 0.7917384547475219,
+      "learning_rate": 0.003,
+      "loss": 4.2254,
+      "step": 1638
+    },
+    {
+      "epoch": 0.01639,
+      "grad_norm": 0.8310494258877913,
+      "learning_rate": 0.003,
+      "loss": 4.2496,
+      "step": 1639
+    },
+    {
+      "epoch": 0.0164,
+      "grad_norm": 0.6945218446546085,
+      "learning_rate": 0.003,
+      "loss": 4.2286,
+      "step": 1640
+    },
+    {
+      "epoch": 0.01641,
+      "grad_norm": 0.643774136427719,
+      "learning_rate": 0.003,
+      "loss": 4.2028,
+      "step": 1641
+    },
+    {
+      "epoch": 0.01642,
+      "grad_norm": 0.5630395282689907,
+      "learning_rate": 0.003,
+      "loss": 4.2418,
+      "step": 1642
+    },
+    {
+      "epoch": 0.01643,
+      "grad_norm": 0.6319019547985867,
+      "learning_rate": 0.003,
+      "loss": 4.23,
+      "step": 1643
+    },
+    {
+      "epoch": 0.01644,
+      "grad_norm": 0.6923812193124619,
+      "learning_rate": 0.003,
+      "loss": 4.2497,
+      "step": 1644
+    },
+    {
+      "epoch": 0.01645,
+      "grad_norm": 0.7835455696779042,
+      "learning_rate": 0.003,
+      "loss": 4.2006,
+      "step": 1645
+    },
+    {
+      "epoch": 0.01646,
+      "grad_norm": 0.9362858528690088,
+      "learning_rate": 0.003,
+      "loss": 4.2431,
+      "step": 1646
+    },
+    {
+      "epoch": 0.01647,
+      "grad_norm": 1.2289381301321516,
+      "learning_rate": 0.003,
+      "loss": 4.255,
+      "step": 1647
+    },
+    {
+      "epoch": 0.01648,
+      "grad_norm": 0.8093728261853784,
+      "learning_rate": 0.003,
+      "loss": 4.2206,
+      "step": 1648
+    },
+    {
+      "epoch": 0.01649,
+      "grad_norm": 0.7780860637025212,
+      "learning_rate": 0.003,
+      "loss": 4.2535,
+      "step": 1649
+    },
+    {
+      "epoch": 0.0165,
+      "grad_norm": 0.8631062415473619,
+      "learning_rate": 0.003,
+      "loss": 4.2375,
+      "step": 1650
+    },
+    {
+      "epoch": 0.01651,
+      "grad_norm": 0.9142840285195195,
+      "learning_rate": 0.003,
+      "loss": 4.2523,
+      "step": 1651
+    },
+    {
+      "epoch": 0.01652,
+      "grad_norm": 0.8336324309893678,
+      "learning_rate": 0.003,
+      "loss": 4.2094,
+      "step": 1652
+    },
+    {
+      "epoch": 0.01653,
+      "grad_norm": 0.856140708339795,
+      "learning_rate": 0.003,
+      "loss": 4.2552,
+      "step": 1653
+    },
+    {
+      "epoch": 0.01654,
+      "grad_norm": 0.9035014747718915,
+      "learning_rate": 0.003,
+      "loss": 4.2428,
+      "step": 1654
+    },
+    {
+      "epoch": 0.01655,
+      "grad_norm": 0.8007220508865799,
+      "learning_rate": 0.003,
+      "loss": 4.2521,
+      "step": 1655
+    },
+    {
+      "epoch": 0.01656,
+      "grad_norm": 0.8140605513593369,
+      "learning_rate": 0.003,
+      "loss": 4.2202,
+      "step": 1656
+    },
+    {
+      "epoch": 0.01657,
+      "grad_norm": 0.752017726600531,
+      "learning_rate": 0.003,
+      "loss": 4.2402,
+      "step": 1657
+    },
+    {
+      "epoch": 0.01658,
+      "grad_norm": 0.8846697411940587,
+      "learning_rate": 0.003,
+      "loss": 4.2338,
+      "step": 1658
+    },
+    {
+      "epoch": 0.01659,
+      "grad_norm": 0.8979228683366066,
+      "learning_rate": 0.003,
+      "loss": 4.2417,
+      "step": 1659
+    },
+    {
+      "epoch": 0.0166,
+      "grad_norm": 0.73714542235099,
+      "learning_rate": 0.003,
+      "loss": 4.2224,
+      "step": 1660
+    },
+    {
+      "epoch": 0.01661,
+      "grad_norm": 0.6566854277166256,
+      "learning_rate": 0.003,
+      "loss": 4.2212,
+      "step": 1661
+    },
+    {
+      "epoch": 0.01662,
+      "grad_norm": 0.5217494005958339,
+      "learning_rate": 0.003,
+      "loss": 4.2208,
+      "step": 1662
+    },
+    {
+      "epoch": 0.01663,
+      "grad_norm": 0.4794160667989573,
+      "learning_rate": 0.003,
+      "loss": 4.2395,
+      "step": 1663
+    },
+    {
+      "epoch": 0.01664,
+      "grad_norm": 0.5082285006914213,
+      "learning_rate": 0.003,
+      "loss": 4.2563,
+      "step": 1664
+    },
+    {
+      "epoch": 0.01665,
+      "grad_norm": 0.5164271930929455,
+      "learning_rate": 0.003,
+      "loss": 4.2188,
+      "step": 1665
+    },
+    {
+      "epoch": 0.01666,
+      "grad_norm": 0.6299180885641164,
+      "learning_rate": 0.003,
+      "loss": 4.2234,
+      "step": 1666
+    },
+    {
+      "epoch": 0.01667,
+      "grad_norm": 0.6409430167727818,
+      "learning_rate": 0.003,
+      "loss": 4.2382,
+      "step": 1667
+    },
+    {
+      "epoch": 0.01668,
+      "grad_norm": 0.6129313286956495,
+      "learning_rate": 0.003,
+      "loss": 4.2415,
+      "step": 1668
+    },
+    {
+      "epoch": 0.01669,
+      "grad_norm": 0.5586023730002122,
+      "learning_rate": 0.003,
+      "loss": 4.223,
+      "step": 1669
+    },
+    {
+      "epoch": 0.0167,
+      "grad_norm": 0.6736111064245857,
+      "learning_rate": 0.003,
+      "loss": 4.2169,
+      "step": 1670
+    },
+    {
+      "epoch": 0.01671,
+      "grad_norm": 0.7441939131840011,
+      "learning_rate": 0.003,
+      "loss": 4.2115,
+      "step": 1671
+    },
+    {
+      "epoch": 0.01672,
+      "grad_norm": 0.8159394752701435,
+      "learning_rate": 0.003,
+      "loss": 4.2214,
+      "step": 1672
+    },
+    {
+      "epoch": 0.01673,
+      "grad_norm": 0.7952390142878619,
+      "learning_rate": 0.003,
+      "loss": 4.2239,
+      "step": 1673
+    },
+    {
+      "epoch": 0.01674,
+      "grad_norm": 0.7199816431285693,
+      "learning_rate": 0.003,
+      "loss": 4.2398,
+      "step": 1674
+    },
+    {
+      "epoch": 0.01675,
+      "grad_norm": 0.7812011305713198,
+      "learning_rate": 0.003,
+      "loss": 4.2244,
+      "step": 1675
+    },
+    {
+      "epoch": 0.01676,
+      "grad_norm": 0.7469280078216791,
+      "learning_rate": 0.003,
+      "loss": 4.2309,
+      "step": 1676
+    },
+    {
+      "epoch": 0.01677,
+      "grad_norm": 0.7271162906259458,
+      "learning_rate": 0.003,
+      "loss": 4.2196,
+      "step": 1677
+    },
+    {
+      "epoch": 0.01678,
+      "grad_norm": 0.8881037129933885,
+      "learning_rate": 0.003,
+      "loss": 4.2406,
+      "step": 1678
+    },
+    {
+      "epoch": 0.01679,
+      "grad_norm": 0.7948545784278963,
+      "learning_rate": 0.003,
+      "loss": 4.2229,
+      "step": 1679
+    },
+    {
+      "epoch": 0.0168,
+      "grad_norm": 0.7189446111250655,
+      "learning_rate": 0.003,
+      "loss": 4.2374,
+      "step": 1680
+    },
+    {
+      "epoch": 0.01681,
+      "grad_norm": 0.7033815727690789,
+      "learning_rate": 0.003,
+      "loss": 4.2384,
+      "step": 1681
+    },
+    {
+      "epoch": 0.01682,
+      "grad_norm": 0.6685587436768649,
+      "learning_rate": 0.003,
+      "loss": 4.231,
+      "step": 1682
+    },
+    {
+      "epoch": 0.01683,
+      "grad_norm": 0.637929998688511,
+      "learning_rate": 0.003,
+      "loss": 4.241,
+      "step": 1683
+    },
+    {
+      "epoch": 0.01684,
+      "grad_norm": 0.628959376324859,
+      "learning_rate": 0.003,
+      "loss": 4.2154,
+      "step": 1684
+    },
+    {
+      "epoch": 0.01685,
+      "grad_norm": 0.6231765249837203,
+      "learning_rate": 0.003,
+      "loss": 4.231,
+      "step": 1685
+    },
+    {
+      "epoch": 0.01686,
+      "grad_norm": 0.5820543845215673,
+      "learning_rate": 0.003,
+      "loss": 4.2226,
+      "step": 1686
+    },
+    {
+      "epoch": 0.01687,
+      "grad_norm": 0.5908668431711505,
+      "learning_rate": 0.003,
+      "loss": 4.2167,
+      "step": 1687
+    },
+    {
+      "epoch": 0.01688,
+      "grad_norm": 0.6244492735925097,
+      "learning_rate": 0.003,
+      "loss": 4.2038,
+      "step": 1688
+    },
+    {
+      "epoch": 0.01689,
+      "grad_norm": 0.601372798813749,
+      "learning_rate": 0.003,
+      "loss": 4.2192,
+      "step": 1689
+    },
+    {
+      "epoch": 0.0169,
+      "grad_norm": 0.5539912290934804,
+      "learning_rate": 0.003,
+      "loss": 4.2069,
+      "step": 1690
+    },
+    {
+      "epoch": 0.01691,
+      "grad_norm": 0.6406257238232621,
+      "learning_rate": 0.003,
+      "loss": 4.2196,
+      "step": 1691
+    },
+    {
+      "epoch": 0.01692,
+      "grad_norm": 0.6715764673442781,
+      "learning_rate": 0.003,
+      "loss": 4.2214,
+      "step": 1692
+    },
+    {
+      "epoch": 0.01693,
+      "grad_norm": 0.594965855834695,
+      "learning_rate": 0.003,
+      "loss": 4.2461,
+      "step": 1693
+    },
+    {
+      "epoch": 0.01694,
+      "grad_norm": 0.5534146564551462,
+      "learning_rate": 0.003,
+      "loss": 4.1943,
+      "step": 1694
+    },
+    {
+      "epoch": 0.01695,
+      "grad_norm": 0.6691590053804509,
+      "learning_rate": 0.003,
+      "loss": 4.2273,
+      "step": 1695
+    },
+    {
+      "epoch": 0.01696,
+      "grad_norm": 0.7988479106506707,
+      "learning_rate": 0.003,
+      "loss": 4.2265,
+      "step": 1696
+    },
+    {
+      "epoch": 0.01697,
+      "grad_norm": 0.9378853835820298,
+      "learning_rate": 0.003,
+      "loss": 4.2129,
+      "step": 1697
+    },
+    {
+      "epoch": 0.01698,
+      "grad_norm": 0.8351693624536569,
+      "learning_rate": 0.003,
+      "loss": 4.2098,
+      "step": 1698
+    },
+    {
+      "epoch": 0.01699,
+      "grad_norm": 0.6689256544140322,
+      "learning_rate": 0.003,
+      "loss": 4.2116,
+      "step": 1699
+    },
+    {
+      "epoch": 0.017,
+      "grad_norm": 0.718191815671613,
+      "learning_rate": 0.003,
+      "loss": 4.2331,
+      "step": 1700
+    },
+    {
+      "epoch": 0.01701,
+      "grad_norm": 0.6236805335228212,
+      "learning_rate": 0.003,
+      "loss": 4.2114,
+      "step": 1701
+    },
+    {
+      "epoch": 0.01702,
+      "grad_norm": 0.6198051313671201,
+      "learning_rate": 0.003,
+      "loss": 4.235,
+      "step": 1702
+    },
+    {
+      "epoch": 0.01703,
+      "grad_norm": 0.6124735544668184,
+      "learning_rate": 0.003,
+      "loss": 4.2192,
+      "step": 1703
+    },
+    {
+      "epoch": 0.01704,
+      "grad_norm": 0.5788411907923544,
+      "learning_rate": 0.003,
+      "loss": 4.2046,
+      "step": 1704
+    },
+    {
+      "epoch": 0.01705,
+      "grad_norm": 0.528954361543718,
+      "learning_rate": 0.003,
+      "loss": 4.2048,
+      "step": 1705
+    },
+    {
+      "epoch": 0.01706,
+      "grad_norm": 0.5207913048275148,
+      "learning_rate": 0.003,
+      "loss": 4.203,
+      "step": 1706
+    },
+    {
+      "epoch": 0.01707,
+      "grad_norm": 0.49956087831547324,
+      "learning_rate": 0.003,
+      "loss": 4.2234,
+      "step": 1707
+    },
+    {
+      "epoch": 0.01708,
+      "grad_norm": 0.5265055046037553,
+      "learning_rate": 0.003,
+      "loss": 4.2005,
+      "step": 1708
+    },
+    {
+      "epoch": 0.01709,
+      "grad_norm": 0.601511173671749,
+      "learning_rate": 0.003,
+      "loss": 4.2245,
+      "step": 1709
+    },
+    {
+      "epoch": 0.0171,
+      "grad_norm": 0.6457072596477392,
+      "learning_rate": 0.003,
+      "loss": 4.1751,
+      "step": 1710
+    },
+    {
+      "epoch": 0.01711,
+      "grad_norm": 0.6931354519414087,
+      "learning_rate": 0.003,
+      "loss": 4.206,
+      "step": 1711
+    },
+    {
+      "epoch": 0.01712,
+      "grad_norm": 0.7843884994764989,
+      "learning_rate": 0.003,
+      "loss": 4.2207,
+      "step": 1712
+    },
+    {
+      "epoch": 0.01713,
+      "grad_norm": 0.8034349282711927,
+      "learning_rate": 0.003,
+      "loss": 4.1971,
+      "step": 1713
+    },
+    {
+      "epoch": 0.01714,
+      "grad_norm": 0.8979650985806761,
+      "learning_rate": 0.003,
+      "loss": 4.2412,
+      "step": 1714
+    },
+    {
+      "epoch": 0.01715,
+      "grad_norm": 0.9199811287834594,
+      "learning_rate": 0.003,
+      "loss": 4.2496,
+      "step": 1715
+    },
+    {
+      "epoch": 0.01716,
+      "grad_norm": 0.8606879750752993,
+      "learning_rate": 0.003,
+      "loss": 4.2283,
+      "step": 1716
+    },
+    {
+      "epoch": 0.01717,
+      "grad_norm": 0.9133738914940963,
+      "learning_rate": 0.003,
+      "loss": 4.2649,
+      "step": 1717
+    },
+    {
+      "epoch": 0.01718,
+      "grad_norm": 0.9751937973978706,
+      "learning_rate": 0.003,
+      "loss": 4.2385,
+      "step": 1718
+    },
+    {
+      "epoch": 0.01719,
+      "grad_norm": 0.9249947621239035,
+      "learning_rate": 0.003,
+      "loss": 4.2632,
+      "step": 1719
+    },
+    {
+      "epoch": 0.0172,
+      "grad_norm": 0.9364522247503431,
+      "learning_rate": 0.003,
+      "loss": 4.2208,
+      "step": 1720
+    },
+    {
+      "epoch": 0.01721,
+      "grad_norm": 0.8201093609998811,
+      "learning_rate": 0.003,
+      "loss": 4.2353,
+      "step": 1721
+    },
+    {
+      "epoch": 0.01722,
+      "grad_norm": 0.8214210127594672,
+      "learning_rate": 0.003,
+      "loss": 4.2288,
+      "step": 1722
+    },
+    {
+      "epoch": 0.01723,
+      "grad_norm": 0.8471893471073342,
+      "learning_rate": 0.003,
+      "loss": 4.2399,
+      "step": 1723
+    },
+    {
+      "epoch": 0.01724,
+      "grad_norm": 0.7683690946961591,
+      "learning_rate": 0.003,
+      "loss": 4.2434,
+      "step": 1724
+    },
+    {
+      "epoch": 0.01725,
+      "grad_norm": 0.8115047065658012,
+      "learning_rate": 0.003,
+      "loss": 4.2272,
+      "step": 1725
+    },
+    {
+      "epoch": 0.01726,
+      "grad_norm": 0.8092210144832711,
+      "learning_rate": 0.003,
+      "loss": 4.2174,
+      "step": 1726
+    },
+    {
+      "epoch": 0.01727,
+      "grad_norm": 0.8550625584545292,
+      "learning_rate": 0.003,
+      "loss": 4.226,
+      "step": 1727
+    },
+    {
+      "epoch": 0.01728,
+      "grad_norm": 0.8021206512196133,
+      "learning_rate": 0.003,
+      "loss": 4.2298,
+      "step": 1728
+    },
+    {
+      "epoch": 0.01729,
+      "grad_norm": 0.7363522392419929,
+      "learning_rate": 0.003,
+      "loss": 4.2336,
+      "step": 1729
+    },
+    {
+      "epoch": 0.0173,
+      "grad_norm": 0.6975599031052584,
+      "learning_rate": 0.003,
+      "loss": 4.2051,
+      "step": 1730
+    },
+    {
+      "epoch": 0.01731,
+      "grad_norm": 0.6221588088033019,
+      "learning_rate": 0.003,
+      "loss": 4.2243,
+      "step": 1731
+    },
+    {
+      "epoch": 0.01732,
+      "grad_norm": 0.612618335472448,
+      "learning_rate": 0.003,
+      "loss": 4.2315,
+      "step": 1732
+    },
+    {
+      "epoch": 0.01733,
+      "grad_norm": 0.6541617874624247,
+      "learning_rate": 0.003,
+      "loss": 4.2125,
+      "step": 1733
+    },
+    {
+      "epoch": 0.01734,
+      "grad_norm": 0.5863531048899692,
+      "learning_rate": 0.003,
+      "loss": 4.2205,
+      "step": 1734
+    },
+    {
+      "epoch": 0.01735,
+      "grad_norm": 0.5699608850382641,
+      "learning_rate": 0.003,
+      "loss": 4.1915,
+      "step": 1735
+    },
+    {
+      "epoch": 0.01736,
+      "grad_norm": 0.5871625865225932,
+      "learning_rate": 0.003,
+      "loss": 4.1898,
+      "step": 1736
+    },
+    {
+      "epoch": 0.01737,
+      "grad_norm": 0.5870381595261084,
+      "learning_rate": 0.003,
+      "loss": 4.2212,
+      "step": 1737
+    },
+    {
+      "epoch": 0.01738,
+      "grad_norm": 0.5498612952733963,
+      "learning_rate": 0.003,
+      "loss": 4.2039,
+      "step": 1738
+    },
+    {
+      "epoch": 0.01739,
+      "grad_norm": 0.6117901077151079,
+      "learning_rate": 0.003,
+      "loss": 4.2218,
+      "step": 1739
+    },
+    {
+      "epoch": 0.0174,
+      "grad_norm": 0.6161361385546128,
+      "learning_rate": 0.003,
+      "loss": 4.2194,
+      "step": 1740
+    },
+    {
+      "epoch": 0.01741,
+      "grad_norm": 0.6356544908412293,
+      "learning_rate": 0.003,
+      "loss": 4.2288,
+      "step": 1741
+    },
+    {
+      "epoch": 0.01742,
+      "grad_norm": 0.6972113147848942,
+      "learning_rate": 0.003,
+      "loss": 4.2174,
+      "step": 1742
+    },
+    {
+      "epoch": 0.01743,
+      "grad_norm": 0.76783001607299,
+      "learning_rate": 0.003,
+      "loss": 4.2327,
+      "step": 1743
+    },
+    {
+      "epoch": 0.01744,
+      "grad_norm": 0.7771532804911003,
+      "learning_rate": 0.003,
+      "loss": 4.2177,
+      "step": 1744
+    },
+    {
+      "epoch": 0.01745,
+      "grad_norm": 0.61106822860021,
+      "learning_rate": 0.003,
+      "loss": 4.1781,
+      "step": 1745
+    },
+    {
+      "epoch": 0.01746,
+      "grad_norm": 0.6089323419748714,
+      "learning_rate": 0.003,
+      "loss": 4.2031,
+      "step": 1746
+    },
+    {
+      "epoch": 0.01747,
+      "grad_norm": 0.6068526864216675,
+      "learning_rate": 0.003,
+      "loss": 4.2005,
+      "step": 1747
+    },
+    {
+      "epoch": 0.01748,
+      "grad_norm": 0.6143873963395499,
+      "learning_rate": 0.003,
+      "loss": 4.194,
+      "step": 1748
+    },
+    {
+      "epoch": 0.01749,
+      "grad_norm": 0.505834715053384,
+      "learning_rate": 0.003,
+      "loss": 4.1866,
+      "step": 1749
+    },
+    {
+      "epoch": 0.0175,
+      "grad_norm": 0.4183071099190176,
+      "learning_rate": 0.003,
+      "loss": 4.1986,
+      "step": 1750
+    },
+    {
+      "epoch": 0.01751,
+      "grad_norm": 0.4200651633429046,
+      "learning_rate": 0.003,
+      "loss": 4.1916,
+      "step": 1751
+    },
+    {
+      "epoch": 0.01752,
+      "grad_norm": 0.42294753534060125,
+      "learning_rate": 0.003,
+      "loss": 4.2318,
+      "step": 1752
+    },
+    {
+      "epoch": 0.01753,
+      "grad_norm": 0.4040582052822703,
+      "learning_rate": 0.003,
+      "loss": 4.1786,
+      "step": 1753
+    },
+    {
+      "epoch": 0.01754,
+      "grad_norm": 0.39927824671480916,
+      "learning_rate": 0.003,
+      "loss": 4.2108,
+      "step": 1754
+    },
+    {
+      "epoch": 0.01755,
+      "grad_norm": 0.39324726452057945,
+      "learning_rate": 0.003,
+      "loss": 4.1497,
+      "step": 1755
+    },
+    {
+      "epoch": 0.01756,
+      "grad_norm": 0.40168143250637856,
+      "learning_rate": 0.003,
+      "loss": 4.227,
+      "step": 1756
+    },
+    {
+      "epoch": 0.01757,
+      "grad_norm": 0.4561700566055392,
+      "learning_rate": 0.003,
+      "loss": 4.2081,
+      "step": 1757
+    },
+    {
+      "epoch": 0.01758,
+      "grad_norm": 0.5466402873638766,
+      "learning_rate": 0.003,
+      "loss": 4.2157,
+      "step": 1758
+    },
+    {
+      "epoch": 0.01759,
+      "grad_norm": 0.6515131335626271,
+      "learning_rate": 0.003,
+      "loss": 4.1877,
+      "step": 1759
+    },
+    {
+      "epoch": 0.0176,
+      "grad_norm": 0.7221307724120931,
+      "learning_rate": 0.003,
+      "loss": 4.2153,
+      "step": 1760
+    },
+    {
+      "epoch": 0.01761,
+      "grad_norm": 0.7514386530231217,
+      "learning_rate": 0.003,
+      "loss": 4.2014,
+      "step": 1761
+    },
+    {
+      "epoch": 0.01762,
+      "grad_norm": 0.8563467654669285,
+      "learning_rate": 0.003,
+      "loss": 4.2266,
+      "step": 1762
+    },
+    {
+      "epoch": 0.01763,
+      "grad_norm": 1.0251781966749487,
+      "learning_rate": 0.003,
+      "loss": 4.2304,
+      "step": 1763
+    },
+    {
+      "epoch": 0.01764,
+      "grad_norm": 1.1022130069351062,
+      "learning_rate": 0.003,
+      "loss": 4.204,
+      "step": 1764
+    },
+    {
+      "epoch": 0.01765,
+      "grad_norm": 0.9424736955387317,
+      "learning_rate": 0.003,
+      "loss": 4.2315,
+      "step": 1765
+    },
+    {
+      "epoch": 0.01766,
+      "grad_norm": 0.9227888573535311,
+      "learning_rate": 0.003,
+      "loss": 4.2455,
+      "step": 1766
+    },
+    {
+      "epoch": 0.01767,
+      "grad_norm": 0.8152567408138972,
+      "learning_rate": 0.003,
+      "loss": 4.2043,
+      "step": 1767
+    },
+    {
+      "epoch": 0.01768,
+      "grad_norm": 0.8952526282025846,
+      "learning_rate": 0.003,
+      "loss": 4.2178,
+      "step": 1768
+    },
+    {
+      "epoch": 0.01769,
+      "grad_norm": 0.9903660885622939,
+      "learning_rate": 0.003,
+      "loss": 4.221,
+      "step": 1769
+    },
+    {
+      "epoch": 0.0177,
+      "grad_norm": 1.1434768458973932,
+      "learning_rate": 0.003,
+      "loss": 4.2196,
+      "step": 1770
+    },
+    {
+      "epoch": 0.01771,
+      "grad_norm": 1.1463152027154602,
+      "learning_rate": 0.003,
+      "loss": 4.2541,
+      "step": 1771
+    },
+    {
+      "epoch": 0.01772,
+      "grad_norm": 0.9320499048871318,
+      "learning_rate": 0.003,
+      "loss": 4.2202,
+      "step": 1772
+    },
+    {
+      "epoch": 0.01773,
+      "grad_norm": 0.9158947764816099,
+      "learning_rate": 0.003,
+      "loss": 4.2126,
+      "step": 1773
+    },
+    {
+      "epoch": 0.01774,
+      "grad_norm": 1.0234123818722582,
+      "learning_rate": 0.003,
+      "loss": 4.25,
+      "step": 1774
+    },
+    {
+      "epoch": 0.01775,
+      "grad_norm": 1.0396053755051766,
+      "learning_rate": 0.003,
+      "loss": 4.2387,
+      "step": 1775
+    },
+    {
+      "epoch": 0.01776,
+      "grad_norm": 0.8699023391257188,
+      "learning_rate": 0.003,
+      "loss": 4.2279,
+      "step": 1776
+    },
+    {
+      "epoch": 0.01777,
+      "grad_norm": 0.7504359108002385,
+      "learning_rate": 0.003,
+      "loss": 4.2409,
+      "step": 1777
+    },
+    {
+      "epoch": 0.01778,
+      "grad_norm": 0.6686463646026519,
+      "learning_rate": 0.003,
+      "loss": 4.2296,
+      "step": 1778
+    },
+    {
+      "epoch": 0.01779,
+      "grad_norm": 0.627501081983313,
+      "learning_rate": 0.003,
+      "loss": 4.2441,
+      "step": 1779
+    },
+    {
+      "epoch": 0.0178,
+      "grad_norm": 0.6074782341836313,
+      "learning_rate": 0.003,
+      "loss": 4.213,
+      "step": 1780
+    },
+    {
+      "epoch": 0.01781,
+      "grad_norm": 0.6632306275046532,
+      "learning_rate": 0.003,
+      "loss": 4.2271,
+      "step": 1781
+    },
+    {
+      "epoch": 0.01782,
+      "grad_norm": 0.6482376732581775,
+      "learning_rate": 0.003,
+      "loss": 4.2498,
+      "step": 1782
+    },
+    {
+      "epoch": 0.01783,
+      "grad_norm": 0.6647693064019089,
+      "learning_rate": 0.003,
+      "loss": 4.229,
+      "step": 1783
+    },
+    {
+      "epoch": 0.01784,
+      "grad_norm": 0.7433144987650033,
+      "learning_rate": 0.003,
+      "loss": 4.2289,
+      "step": 1784
+    },
+    {
+      "epoch": 0.01785,
+      "grad_norm": 0.8023042672883854,
+      "learning_rate": 0.003,
+      "loss": 4.1929,
+      "step": 1785
+    },
+    {
+      "epoch": 0.01786,
+      "grad_norm": 0.7590350402092263,
+      "learning_rate": 0.003,
+      "loss": 4.2189,
+      "step": 1786
+    },
+    {
+      "epoch": 0.01787,
+      "grad_norm": 0.6044391074634635,
+      "learning_rate": 0.003,
+      "loss": 4.223,
+      "step": 1787
+    },
+    {
+      "epoch": 0.01788,
+      "grad_norm": 0.5734225415780418,
+      "learning_rate": 0.003,
+      "loss": 4.2291,
+      "step": 1788
+    },
+    {
+      "epoch": 0.01789,
+      "grad_norm": 0.5619094879913861,
+      "learning_rate": 0.003,
+      "loss": 4.1969,
+      "step": 1789
+    },
+    {
+      "epoch": 0.0179,
+      "grad_norm": 0.5903172908647163,
+      "learning_rate": 0.003,
+      "loss": 4.2152,
+      "step": 1790
+    },
+    {
+      "epoch": 0.01791,
+      "grad_norm": 0.48196952921006103,
+      "learning_rate": 0.003,
+      "loss": 4.2224,
+      "step": 1791
+    },
+    {
+      "epoch": 0.01792,
+      "grad_norm": 0.451545250518652,
+      "learning_rate": 0.003,
+      "loss": 4.2239,
+      "step": 1792
+    },
+    {
+      "epoch": 0.01793,
+      "grad_norm": 0.521884061957321,
+      "learning_rate": 0.003,
+      "loss": 4.1852,
+      "step": 1793
+    },
+    {
+      "epoch": 0.01794,
+      "grad_norm": 0.5052209264531972,
+      "learning_rate": 0.003,
+      "loss": 4.1997,
+      "step": 1794
+    },
+    {
+      "epoch": 0.01795,
+      "grad_norm": 0.4700227936734472,
+      "learning_rate": 0.003,
+      "loss": 4.1984,
+      "step": 1795
+    },
+    {
+      "epoch": 0.01796,
+      "grad_norm": 0.48384581004820526,
+      "learning_rate": 0.003,
+      "loss": 4.2111,
+      "step": 1796
+    },
+    {
+      "epoch": 0.01797,
+      "grad_norm": 0.5674789805804968,
+      "learning_rate": 0.003,
+      "loss": 4.1791,
+      "step": 1797
+    },
+    {
+      "epoch": 0.01798,
+      "grad_norm": 0.5733510313406954,
+      "learning_rate": 0.003,
+      "loss": 4.1913,
+      "step": 1798
+    },
+    {
+      "epoch": 0.01799,
+      "grad_norm": 0.6643001766771917,
+      "learning_rate": 0.003,
+      "loss": 4.2106,
+      "step": 1799
+    },
+    {
+      "epoch": 0.018,
+      "grad_norm": 0.7693742824268138,
+      "learning_rate": 0.003,
+      "loss": 4.2019,
+      "step": 1800
+    },
+    {
+      "epoch": 0.01801,
+      "grad_norm": 0.7789742970956256,
+      "learning_rate": 0.003,
+      "loss": 4.2028,
+      "step": 1801
+    },
+    {
+      "epoch": 0.01802,
+      "grad_norm": 0.7483411999928193,
+      "learning_rate": 0.003,
+      "loss": 4.2174,
+      "step": 1802
+    },
+    {
+      "epoch": 0.01803,
+      "grad_norm": 0.6912125156211221,
+      "learning_rate": 0.003,
+      "loss": 4.2209,
+      "step": 1803
+    },
+    {
+      "epoch": 0.01804,
+      "grad_norm": 0.6110578547720885,
+      "learning_rate": 0.003,
+      "loss": 4.1902,
+      "step": 1804
+    },
+    {
+      "epoch": 0.01805,
+      "grad_norm": 0.5506478742000366,
+      "learning_rate": 0.003,
+      "loss": 4.2069,
+      "step": 1805
+    },
+    {
+      "epoch": 0.01806,
+      "grad_norm": 0.6061937579522265,
+      "learning_rate": 0.003,
+      "loss": 4.1674,
+      "step": 1806
+    },
+    {
+      "epoch": 0.01807,
+      "grad_norm": 0.5932129898926418,
+      "learning_rate": 0.003,
+      "loss": 4.2038,
+      "step": 1807
+    },
+    {
+      "epoch": 0.01808,
+      "grad_norm": 0.5444224258882361,
+      "learning_rate": 0.003,
+      "loss": 4.1921,
+      "step": 1808
+    },
+    {
+      "epoch": 0.01809,
+      "grad_norm": 0.5399756231214402,
+      "learning_rate": 0.003,
+      "loss": 4.2143,
+      "step": 1809
+    },
+    {
+      "epoch": 0.0181,
+      "grad_norm": 0.4832914161347141,
+      "learning_rate": 0.003,
+      "loss": 4.2061,
+      "step": 1810
+    },
+    {
+      "epoch": 0.01811,
+      "grad_norm": 0.5091858383706224,
+      "learning_rate": 0.003,
+      "loss": 4.1923,
+      "step": 1811
+    },
+    {
+      "epoch": 0.01812,
+      "grad_norm": 0.4960424818906239,
+      "learning_rate": 0.003,
+      "loss": 4.1761,
+      "step": 1812
+    },
+    {
+      "epoch": 0.01813,
+      "grad_norm": 0.4858161245409072,
+      "learning_rate": 0.003,
+      "loss": 4.2194,
+      "step": 1813
+    },
+    {
+      "epoch": 0.01814,
+      "grad_norm": 0.5128609502675707,
+      "learning_rate": 0.003,
+      "loss": 4.1853,
+      "step": 1814
+    },
+    {
+      "epoch": 0.01815,
+      "grad_norm": 0.4696917504376051,
+      "learning_rate": 0.003,
+      "loss": 4.1713,
+      "step": 1815
+    },
+    {
+      "epoch": 0.01816,
+      "grad_norm": 0.4276704853681577,
+      "learning_rate": 0.003,
+      "loss": 4.1969,
+      "step": 1816
+    },
+    {
+      "epoch": 0.01817,
+      "grad_norm": 0.49742714220067663,
+      "learning_rate": 0.003,
+      "loss": 4.1904,
+      "step": 1817
+    },
+    {
+      "epoch": 0.01818,
+      "grad_norm": 0.6269759675631359,
+      "learning_rate": 0.003,
+      "loss": 4.1931,
+      "step": 1818
+    },
+    {
+      "epoch": 0.01819,
+      "grad_norm": 0.7520796750217182,
+      "learning_rate": 0.003,
+      "loss": 4.1832,
+      "step": 1819
+    },
+    {
+      "epoch": 0.0182,
+      "grad_norm": 0.9014481818810978,
+      "learning_rate": 0.003,
+      "loss": 4.2101,
+      "step": 1820
+    },
+    {
+      "epoch": 0.01821,
+      "grad_norm": 0.7861765129999728,
+      "learning_rate": 0.003,
+      "loss": 4.1954,
+      "step": 1821
+    },
+    {
+      "epoch": 0.01822,
+      "grad_norm": 0.6190127192025973,
+      "learning_rate": 0.003,
+      "loss": 4.2,
+      "step": 1822
+    },
+    {
+      "epoch": 0.01823,
+      "grad_norm": 0.7778520136299403,
+      "learning_rate": 0.003,
+      "loss": 4.2211,
+      "step": 1823
+    },
+    {
+      "epoch": 0.01824,
+      "grad_norm": 0.7843362143306014,
+      "learning_rate": 0.003,
+      "loss": 4.2055,
+      "step": 1824
+    },
+    {
+      "epoch": 0.01825,
+      "grad_norm": 0.7380149583575544,
+      "learning_rate": 0.003,
+      "loss": 4.2108,
+      "step": 1825
+    },
+    {
+      "epoch": 0.01826,
+      "grad_norm": 0.7040620507144565,
+      "learning_rate": 0.003,
+      "loss": 4.2188,
+      "step": 1826
+    },
+    {
+      "epoch": 0.01827,
+      "grad_norm": 0.8207902713006542,
+      "learning_rate": 0.003,
+      "loss": 4.1955,
+      "step": 1827
+    },
+    {
+      "epoch": 0.01828,
+      "grad_norm": 0.8054878482534694,
+      "learning_rate": 0.003,
+      "loss": 4.2093,
+      "step": 1828
+    },
+    {
+      "epoch": 0.01829,
+      "grad_norm": 0.8245790564775142,
+      "learning_rate": 0.003,
+      "loss": 4.2301,
+      "step": 1829
+    },
+    {
+      "epoch": 0.0183,
+      "grad_norm": 0.8846767870010911,
+      "learning_rate": 0.003,
+      "loss": 4.2277,
+      "step": 1830
+    },
+    {
+      "epoch": 0.01831,
+      "grad_norm": 0.8289617540456092,
+      "learning_rate": 0.003,
+      "loss": 4.2216,
+      "step": 1831
+    },
+    {
+      "epoch": 0.01832,
+      "grad_norm": 0.731274205528004,
+      "learning_rate": 0.003,
+      "loss": 4.2108,
+      "step": 1832
+    },
+    {
+      "epoch": 0.01833,
+      "grad_norm": 0.7928566591291573,
+      "learning_rate": 0.003,
+      "loss": 4.2459,
+      "step": 1833
+    },
+    {
+      "epoch": 0.01834,
+      "grad_norm": 0.7576155251040918,
+      "learning_rate": 0.003,
+      "loss": 4.2123,
+      "step": 1834
+    },
+    {
+      "epoch": 0.01835,
+      "grad_norm": 0.7931842778411774,
+      "learning_rate": 0.003,
+      "loss": 4.214,
+      "step": 1835
+    },
+    {
+      "epoch": 0.01836,
+      "grad_norm": 0.6445999325297602,
+      "learning_rate": 0.003,
+      "loss": 4.2036,
+      "step": 1836
+    },
+    {
+      "epoch": 0.01837,
+      "grad_norm": 0.680643728796186,
+      "learning_rate": 0.003,
+      "loss": 4.208,
+      "step": 1837
+    },
+    {
+      "epoch": 0.01838,
+      "grad_norm": 0.6239843687734086,
+      "learning_rate": 0.003,
+      "loss": 4.2079,
+      "step": 1838
+    },
+    {
+      "epoch": 0.01839,
+      "grad_norm": 0.6182658027468049,
+      "learning_rate": 0.003,
+      "loss": 4.2277,
+      "step": 1839
+    },
+    {
+      "epoch": 0.0184,
+      "grad_norm": 0.5890463339530954,
+      "learning_rate": 0.003,
+      "loss": 4.2247,
+      "step": 1840
+    },
+    {
+      "epoch": 0.01841,
+      "grad_norm": 0.5863839914398512,
+      "learning_rate": 0.003,
+      "loss": 4.1619,
+      "step": 1841
+    },
+    {
+      "epoch": 0.01842,
+      "grad_norm": 0.5858091006393259,
+      "learning_rate": 0.003,
+      "loss": 4.1952,
+      "step": 1842
+    },
+    {
+      "epoch": 0.01843,
+      "grad_norm": 0.5624576041074113,
+      "learning_rate": 0.003,
+      "loss": 4.1755,
+      "step": 1843
+    },
+    {
+      "epoch": 0.01844,
+      "grad_norm": 0.5465927738839299,
+      "learning_rate": 0.003,
+      "loss": 4.1732,
+      "step": 1844
+    },
+    {
+      "epoch": 0.01845,
+      "grad_norm": 0.6027290912828838,
+      "learning_rate": 0.003,
+      "loss": 4.1768,
+      "step": 1845
+    },
+    {
+      "epoch": 0.01846,
+      "grad_norm": 0.7027351856984548,
+      "learning_rate": 0.003,
+      "loss": 4.2065,
+      "step": 1846
+    },
+    {
+      "epoch": 0.01847,
+      "grad_norm": 0.8199216861922868,
+      "learning_rate": 0.003,
+      "loss": 4.2072,
+      "step": 1847
+    },
+    {
+      "epoch": 0.01848,
+      "grad_norm": 0.7370764094470499,
+      "learning_rate": 0.003,
+      "loss": 4.2221,
+      "step": 1848
+    },
+    {
+      "epoch": 0.01849,
+      "grad_norm": 0.6895980947043083,
+      "learning_rate": 0.003,
+      "loss": 4.2176,
+      "step": 1849
+    },
+    {
+      "epoch": 0.0185,
+      "grad_norm": 0.6744601685013915,
+      "learning_rate": 0.003,
+      "loss": 4.2071,
+      "step": 1850
+    },
+    {
+      "epoch": 0.01851,
+      "grad_norm": 0.7592763226230413,
+      "learning_rate": 0.003,
+      "loss": 4.1894,
+      "step": 1851
+    },
+    {
+      "epoch": 0.01852,
+      "grad_norm": 0.7512384494982312,
+      "learning_rate": 0.003,
+      "loss": 4.1913,
+      "step": 1852
+    },
+    {
+      "epoch": 0.01853,
+      "grad_norm": 0.642959606761384,
+      "learning_rate": 0.003,
+      "loss": 4.1897,
+      "step": 1853
+    },
+    {
+      "epoch": 0.01854,
+      "grad_norm": 0.6302719538000866,
+      "learning_rate": 0.003,
+      "loss": 4.2198,
+      "step": 1854
+    },
+    {
+      "epoch": 0.01855,
+      "grad_norm": 0.5865658475812112,
+      "learning_rate": 0.003,
+      "loss": 4.2046,
+      "step": 1855
+    },
+    {
+      "epoch": 0.01856,
+      "grad_norm": 0.6166190320727692,
+      "learning_rate": 0.003,
+      "loss": 4.189,
+      "step": 1856
+    },
+    {
+      "epoch": 0.01857,
+      "grad_norm": 0.7181557293855338,
+      "learning_rate": 0.003,
+      "loss": 4.1771,
+      "step": 1857
+    },
+    {
+      "epoch": 0.01858,
+      "grad_norm": 0.8412048999626534,
+      "learning_rate": 0.003,
+      "loss": 4.1934,
+      "step": 1858
+    },
+    {
+      "epoch": 0.01859,
+      "grad_norm": 0.9023360494815195,
+      "learning_rate": 0.003,
+      "loss": 4.1912,
+      "step": 1859
+    },
+    {
+      "epoch": 0.0186,
+      "grad_norm": 0.7898380300240098,
+      "learning_rate": 0.003,
+      "loss": 4.1718,
+      "step": 1860
+    },
+    {
+      "epoch": 0.01861,
+      "grad_norm": 0.7208167303209774,
+      "learning_rate": 0.003,
+      "loss": 4.2166,
+      "step": 1861
+    },
+    {
+      "epoch": 0.01862,
+      "grad_norm": 0.7104369497590945,
+      "learning_rate": 0.003,
+      "loss": 4.179,
+      "step": 1862
+    },
+    {
+      "epoch": 0.01863,
+      "grad_norm": 0.7138975727375058,
+      "learning_rate": 0.003,
+      "loss": 4.1882,
+      "step": 1863
+    },
+    {
+      "epoch": 0.01864,
+      "grad_norm": 0.6519153740387305,
+      "learning_rate": 0.003,
+      "loss": 4.2193,
+      "step": 1864
+    },
+    {
+      "epoch": 0.01865,
+      "grad_norm": 0.7057872052617501,
+      "learning_rate": 0.003,
+      "loss": 4.1967,
+      "step": 1865
+    },
+    {
+      "epoch": 0.01866,
+      "grad_norm": 0.7579696750373845,
+      "learning_rate": 0.003,
+      "loss": 4.203,
+      "step": 1866
+    },
+    {
+      "epoch": 0.01867,
+      "grad_norm": 0.7437399572447418,
+      "learning_rate": 0.003,
+      "loss": 4.1772,
+      "step": 1867
+    },
+    {
+      "epoch": 0.01868,
+      "grad_norm": 0.7143070391507745,
+      "learning_rate": 0.003,
+      "loss": 4.1947,
+      "step": 1868
+    },
+    {
+      "epoch": 0.01869,
+      "grad_norm": 0.7078429320794878,
+      "learning_rate": 0.003,
+      "loss": 4.2102,
+      "step": 1869
+    },
+    {
+      "epoch": 0.0187,
+      "grad_norm": 0.7908878960855552,
+      "learning_rate": 0.003,
+      "loss": 4.2031,
+      "step": 1870
+    },
+    {
+      "epoch": 0.01871,
+      "grad_norm": 0.9307560381931724,
+      "learning_rate": 0.003,
+      "loss": 4.1837,
+      "step": 1871
+    },
+    {
+      "epoch": 0.01872,
+      "grad_norm": 0.937846873989918,
+      "learning_rate": 0.003,
+      "loss": 4.2367,
+      "step": 1872
+    },
+    {
+      "epoch": 0.01873,
+      "grad_norm": 0.7588221299073229,
+      "learning_rate": 0.003,
+      "loss": 4.1923,
+      "step": 1873
+    },
+    {
+      "epoch": 0.01874,
+      "grad_norm": 0.6728512390013429,
+      "learning_rate": 0.003,
+      "loss": 4.1913,
+      "step": 1874
+    },
+    {
+      "epoch": 0.01875,
+      "grad_norm": 0.6514462862930852,
+      "learning_rate": 0.003,
+      "loss": 4.21,
+      "step": 1875
+    },
+    {
+      "epoch": 0.01876,
+      "grad_norm": 0.703173016887,
+      "learning_rate": 0.003,
+      "loss": 4.2153,
+      "step": 1876
+    },
+    {
+      "epoch": 0.01877,
+      "grad_norm": 0.73028206380223,
+      "learning_rate": 0.003,
+      "loss": 4.2227,
+      "step": 1877
+    },
+    {
+      "epoch": 0.01878,
+      "grad_norm": 0.7555785531530169,
+      "learning_rate": 0.003,
+      "loss": 4.2069,
+      "step": 1878
+    },
+    {
+      "epoch": 0.01879,
+      "grad_norm": 0.7934996955230512,
+      "learning_rate": 0.003,
+      "loss": 4.196,
+      "step": 1879
+    },
+    {
+      "epoch": 0.0188,
+      "grad_norm": 0.7476047962254878,
+      "learning_rate": 0.003,
+      "loss": 4.199,
+      "step": 1880
+    },
+    {
+      "epoch": 0.01881,
+      "grad_norm": 0.732632895668458,
+      "learning_rate": 0.003,
+      "loss": 4.195,
+      "step": 1881
+    },
+    {
+      "epoch": 0.01882,
+      "grad_norm": 0.7676196664715533,
+      "learning_rate": 0.003,
+      "loss": 4.2236,
+      "step": 1882
+    },
+    {
+      "epoch": 0.01883,
+      "grad_norm": 0.774042907112019,
+      "learning_rate": 0.003,
+      "loss": 4.197,
+      "step": 1883
+    },
+    {
+      "epoch": 0.01884,
+      "grad_norm": 0.6501154660141607,
+      "learning_rate": 0.003,
+      "loss": 4.1919,
+      "step": 1884
+    },
+    {
+      "epoch": 0.01885,
+      "grad_norm": 0.5559497730489893,
+      "learning_rate": 0.003,
+      "loss": 4.1659,
+      "step": 1885
+    },
+    {
+      "epoch": 0.01886,
+      "grad_norm": 0.5108169963928473,
+      "learning_rate": 0.003,
+      "loss": 4.2034,
+      "step": 1886
+    },
+    {
+      "epoch": 0.01887,
+      "grad_norm": 0.4802312262485427,
+      "learning_rate": 0.003,
+      "loss": 4.1941,
+      "step": 1887
+    },
+    {
+      "epoch": 0.01888,
+      "grad_norm": 0.44491425211803826,
+      "learning_rate": 0.003,
+      "loss": 4.1789,
+      "step": 1888
+    },
+    {
+      "epoch": 0.01889,
+      "grad_norm": 0.4616155311993016,
+      "learning_rate": 0.003,
+      "loss": 4.1923,
+      "step": 1889
+    },
+    {
+      "epoch": 0.0189,
+      "grad_norm": 0.4932376764256785,
+      "learning_rate": 0.003,
+      "loss": 4.1724,
+      "step": 1890
+    },
+    {
+      "epoch": 0.01891,
+      "grad_norm": 0.5809213606136114,
+      "learning_rate": 0.003,
+      "loss": 4.2219,
+      "step": 1891
+    },
+    {
+      "epoch": 0.01892,
+      "grad_norm": 0.7827791825078997,
+      "learning_rate": 0.003,
+      "loss": 4.2063,
+      "step": 1892
+    },
+    {
+      "epoch": 0.01893,
+      "grad_norm": 0.9634537072256092,
+      "learning_rate": 0.003,
+      "loss": 4.1968,
+      "step": 1893
+    },
+    {
+      "epoch": 0.01894,
+      "grad_norm": 0.8975581900781968,
+      "learning_rate": 0.003,
+      "loss": 4.2047,
+      "step": 1894
+    },
+    {
+      "epoch": 0.01895,
+      "grad_norm": 0.670439659623389,
+      "learning_rate": 0.003,
+      "loss": 4.1794,
+      "step": 1895
+    },
+    {
+      "epoch": 0.01896,
+      "grad_norm": 0.774532069243431,
+      "learning_rate": 0.003,
+      "loss": 4.208,
+      "step": 1896
+    },
+    {
+      "epoch": 0.01897,
+      "grad_norm": 0.7575182930280945,
+      "learning_rate": 0.003,
+      "loss": 4.2077,
+      "step": 1897
+    },
+    {
+      "epoch": 0.01898,
+      "grad_norm": 0.6342553387149622,
+      "learning_rate": 0.003,
+      "loss": 4.1723,
+      "step": 1898
+    },
+    {
+      "epoch": 0.01899,
+      "grad_norm": 0.6743268097098117,
+      "learning_rate": 0.003,
+      "loss": 4.1867,
+      "step": 1899
+    },
+    {
+      "epoch": 0.019,
+      "grad_norm": 0.6675442258402475,
+      "learning_rate": 0.003,
+      "loss": 4.1804,
+      "step": 1900
+    },
+    {
+      "epoch": 0.01901,
+      "grad_norm": 0.6448176087551704,
+      "learning_rate": 0.003,
+      "loss": 4.2267,
+      "step": 1901
+    },
+    {
+      "epoch": 0.01902,
+      "grad_norm": 0.6309260453072276,
+      "learning_rate": 0.003,
+      "loss": 4.2038,
+      "step": 1902
+    },
+    {
+      "epoch": 0.01903,
+      "grad_norm": 0.5175825025006746,
+      "learning_rate": 0.003,
+      "loss": 4.1533,
+      "step": 1903
+    },
+    {
+      "epoch": 0.01904,
+      "grad_norm": 0.44558661360578977,
+      "learning_rate": 0.003,
+      "loss": 4.2075,
+      "step": 1904
+    },
+    {
+      "epoch": 0.01905,
+      "grad_norm": 0.4519539268740215,
+      "learning_rate": 0.003,
+      "loss": 4.1828,
+      "step": 1905
+    },
+    {
+      "epoch": 0.01906,
+      "grad_norm": 0.4534257888011715,
+      "learning_rate": 0.003,
+      "loss": 4.1993,
+      "step": 1906
+    },
+    {
+      "epoch": 0.01907,
+      "grad_norm": 0.4205401145833596,
+      "learning_rate": 0.003,
+      "loss": 4.1917,
+      "step": 1907
+    },
+    {
+      "epoch": 0.01908,
+      "grad_norm": 0.4717335154892585,
+      "learning_rate": 0.003,
+      "loss": 4.2064,
+      "step": 1908
+    },
+    {
+      "epoch": 0.01909,
+      "grad_norm": 0.5307249224766913,
+      "learning_rate": 0.003,
+      "loss": 4.1715,
+      "step": 1909
+    },
+    {
+      "epoch": 0.0191,
+      "grad_norm": 0.6322393539173273,
+      "learning_rate": 0.003,
+      "loss": 4.1858,
+      "step": 1910
+    },
+    {
+      "epoch": 0.01911,
+      "grad_norm": 0.6545216933386392,
+      "learning_rate": 0.003,
+      "loss": 4.1623,
+      "step": 1911
+    },
+    {
+      "epoch": 0.01912,
+      "grad_norm": 0.5663347338798967,
+      "learning_rate": 0.003,
+      "loss": 4.1711,
+      "step": 1912
+    },
+    {
+      "epoch": 0.01913,
+      "grad_norm": 0.5706339861620788,
+      "learning_rate": 0.003,
+      "loss": 4.188,
+      "step": 1913
+    },
+    {
+      "epoch": 0.01914,
+      "grad_norm": 0.6036176037925194,
+      "learning_rate": 0.003,
+      "loss": 4.2071,
+      "step": 1914
+    },
+    {
+      "epoch": 0.01915,
+      "grad_norm": 0.5703771194775653,
+      "learning_rate": 0.003,
+      "loss": 4.1909,
+      "step": 1915
+    },
+    {
+      "epoch": 0.01916,
+      "grad_norm": 0.6012322903506473,
+      "learning_rate": 0.003,
+      "loss": 4.173,
+      "step": 1916
+    },
+    {
+      "epoch": 0.01917,
+      "grad_norm": 0.7091647735109047,
+      "learning_rate": 0.003,
+      "loss": 4.1961,
+      "step": 1917
+    },
+    {
+      "epoch": 0.01918,
+      "grad_norm": 0.8129980711183025,
+      "learning_rate": 0.003,
+      "loss": 4.1895,
+      "step": 1918
+    },
+    {
+      "epoch": 0.01919,
+      "grad_norm": 1.0293282506765802,
+      "learning_rate": 0.003,
+      "loss": 4.2041,
+      "step": 1919
+    },
+    {
+      "epoch": 0.0192,
+      "grad_norm": 0.9313730998487869,
+      "learning_rate": 0.003,
+      "loss": 4.2061,
+      "step": 1920
+    },
+    {
+      "epoch": 0.01921,
+      "grad_norm": 0.8720092800409024,
+      "learning_rate": 0.003,
+      "loss": 4.1913,
+      "step": 1921
+    },
+    {
+      "epoch": 0.01922,
+      "grad_norm": 0.7996945150379353,
+      "learning_rate": 0.003,
+      "loss": 4.2091,
+      "step": 1922
+    },
+    {
+      "epoch": 0.01923,
+      "grad_norm": 0.8800137161869123,
+      "learning_rate": 0.003,
+      "loss": 4.2165,
+      "step": 1923
+    },
+    {
+      "epoch": 0.01924,
+      "grad_norm": 1.0182126450186646,
+      "learning_rate": 0.003,
+      "loss": 4.2271,
+      "step": 1924
+    },
+    {
+      "epoch": 0.01925,
+      "grad_norm": 0.9807294623812044,
+      "learning_rate": 0.003,
+      "loss": 4.2164,
+      "step": 1925
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 100000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 25,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.63140124966912e+16,
+  "train_batch_size": 512,
+  "trial_name": null,
+  "trial_params": null
+}